S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S o S o S o h h h e m e m e m nc c n c n r r r e e e o o o t t t M M M l l l o o o i i i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce Sandia Na e tional Laborcatories e c oc c o c o S h S h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n 20n 15n o o o i l i l li l o l o ol i i i t g Latboratory Dgirected g t a y a y ya . . . c c c r S o SResrearch & Dev o elopSment o r y y y e m e m e m T c c T c T r r r e e 2e015 Anenual Report e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y Sanodia National Laboratories 2015 LDRD Aynnual Report o o y l l l T o T o o T Twengty-five years of innovation for our nagtion g e y e y y e . . . c c c c c c S Throughout hisotory, imagSination and the capacity to dream have alwayos driven grSeat scientific o h h h e and technologicaml leaps forw e ard. This results from stepping outside the bmoundaries i e mposed by m nc mainstream scientific thinking c , and partly from an ability ton anticipate the future and w c hat it may n r r r o e demand of us. e o e o t t t M M M l l l o i The Atomic Energy Act of 1954 provideid the basis for U.S. Departmoent of Energy (DOE) nationial o l l l g i t laboratories to do just this—respond rapiidtly to research challenges angd developments at the i t g a a a cutting edge of science and technology, all while nurturing the best and brightest minds in science, yr r y r y y engineering, and technology. y y .T T . T . ce e c e c To help re-energize scientific, technological, and engineering excellence at our national laboratories, oc c o c o S hthe NatioSnal Defense Authorization Act of 1991 authohrized the laboratories to devote a relativeSly h msnmall portion of their research budget to “work of a creatinve and innovative nature…form the purpose of n m e e e ma o intaining the vitality of the laboratories in defense-related s o cientific disciplines.” Since then, this effort o c has b l eoen formallcy called Laboratory Directed Research and D l oevelopment (LDRD). c l o g g g r r r e Currently y directed bye DOE Order 413.2C, LDRD continues to be a y n important mechanism for our e y . . . c c c t S national labora o tories to atnSticipate, innovate, and deliver solutions for t o he most diSfficult and significant t o M e scientific and tec m hnical chaMlleenges facing our nation. m e M m c c c i l r e For the National Nuclear Securi irtly e Administration (NNSA) laboratories, such as Sandia, LrD e RD is the i l i t t M single most important source of initettM rnal investment in the future, and allows Sandia to: t M i t a a a i i i r l i • Maintain the scientific and technicalri l vitality of the Laboratories l i r t t t y a • Enhance the Laboratories’ ability to ayaddress current and future DOE/NNSA missions a y T r y • Foster creativity and stimulate exploratriTo y n of forefront science and technology r y T T • Serve as a proving ground for new conceptTs in research and development T e e e e • Support high-risk, potentially high-value reseearch and development e c c cc c c h h h h h h n n 25 n n o on o n l l l o o o o o o g g g l y ly y l o . o.YEARS . o c c c S g o S go S o g e m e m e m y For 25y years, our national laboratories, y c c c through the Congressionally authorized r e . r e LDRD pro.gram, have inve r seted in high- . c c c t t risk, potentially high-payoff retsearch and M o M development to oaddress nationMal security o S S challenges. Some of the nation’s mosSt i i i l i m l i impactful technologimes have come frloim m e t a e t a LDRD and allowing researchers to “thitnak e outside the box.” c r c r rc y y y r T r T Tr e e e e ee t M c h n o t M c h U. n S. o D W e a p s a h r in t g m to e n n , t D . o C f . Energy c h n t M o i l i l li l i t a r y T e c h S n e c o r e l o t M g i l i y t a . r y c 2 T o e c m h n o l o o g g y y . . c c o o m m S e c r l e i t t w M A w a V w A . I d i L ir l A r ec i B t c t i L v . y e a E 3 s b . . O d . o r N 2 T e L . . E y g I O a o x N t . v A s h c f D E T e d a f l P i e t u i e t a e O n A A s c s d d P s l n C c c a t e e T E d a i f i T L a e y o o s  o r r o t A t m : r b s e h e I r n e  D n C e r ( 1 c u C p l c N e d t d C i s s i t o . S h  O c r o t o n c s h p . A u i e A C n S 4 U n h r i v e o N c  r E s a t / B O t t E e d e B t r n  1 t A r h t h r o a q r . m E e I a t L e J s ) 9 L s a a b n o L c d u / N h d c i f u E d l / s t L t D D i b t r e O — e e 0 m i t I o e n p u i C t n o o r o N b i . T A r m r f e 6 e R r r e i l p B a e a t i e r T o y v ’ s i c m p Y . N h n m T l s o n d D t e e c s : s t J h e u C y . i r i r o r n n o t c m t . E I i S n L n s e a n e a f e t e t s T O a e b h t a p f t y l n A u C f g t o d A s r s a r h s e o s n h o l a e r N i t t . a l i i x T A c o i r l o s i h n c r B e r r o a n g t e t g T n r e e o . i g e i n o c I o o l g t a t m o d e h b O C o p n a s V h a l h D r e o s r t t y a f m t l i - e n a b r h c i i i e ; a p o R h a l r ; E t n o t o O t a m t e i - o i t i i t y l n i h e g g h p t s A s r v n v s . i o n r y t E a t e i c o s p e k r i a r i h u T y r i d n T r t o h n t e n s e a , f t a l e e n o e y N i l O o o y y t v o a q g c O c a r p p e a d r o s e b t r i i r u c t a a a o e p s n o o f e g R s c i o 4 c f c o o t n i t e p i s t i r d r o . . r i i r r o 1 t c c a e Y o r m b t d o s o e r o i e n o a s a r 3 n n R v o n c c d e n v n p m u w , e b t t p s m . t e D g s e t r e o a e o 2 n O c t i N a l l i a e i v i r a d l g i q r y p f t l B w o d p I t m o o a s c n r i i i N l i e h R u N l f o c e p l d h c m b t , w t i l y f y u i i s s l S n E o l e t a o L y u r r e l a i D h ’ l i w e r e m m p p A h r e i i C c a t a s n 3 b i t n m m h a m m p d e l , i d t s h y n b i d 2 . e T e g e b p t r u o e t e o p t a e h O 1 t e a o h i E h t h a c o e n e w i l r n l e n r 2 a v - r o n e i e e r x t D s a l t c v t t t ( p h c d S e e m n s p y m t c d h c O. n i a o ( p e d s e o e s n R l o a D ) L n l o s e i t n r r o e t c r r u n r n o L r o i n E t o d D e y n t u c r o d a c u e , c c r r p a t D f f e t S r a p c i a a a e D e R e b e f t i r s r r d u P l p o c e E i t d R s t l p i e y . a D o i i y e a d t m c p l f v u s t o r A p C o D y d t n t s i t r o e i o e ) b E o A n e p S t r e e t a R a u s c n i t l s O a o w . r n n s p o n n i d C i e r t n s o l C f s c T n e t e o c r r l i c m h i t v o r o a f t d r d t b h f p ( H i h c e L i y e e , e g e o l l g C g i l e i u l e e m l s r R n e a m a d l y e r n t r r o r r f A i R e h D e w n i a s d o t a e r e i p f s b a o ( f i c . z e m d e D N t s b r O e D t r r r O d t y i a n e ( e r o j o c h n s c d ) s D P a i n e E O a t r h t v , n o f e w g . r e t u c t h . d r y a i i A t n L E v t D e n c a l o d i e S e l n ie t a a t a e o . l ) h n d v t r n i r h d e ) E e b S r t b l n s r a d e e r a c e o m a o 1 ( V o e g c t f e r s o t c r a N p h t u n d r 0 c h t r q e e h t s E t a t o m a e t r d e 6 h a h t i u N n M h m u t e a t f s L o v t e - i o i N t e o e D t r r S s r 6 s f e t r O e m a a r e n i r e h w l N l A 5 i a n e r i r s t a t m o e P e e D o e h y l c . v t b i S ) s t p s i M d . i s a e o e L 1 O ; o e A m m e t l l i n , D n r o f E E h i t a a i e u c p s e R N a t / f n t s e m o N d f v f t t D T e r o a ; a e e N t y c A r n o a n p t S n d p d l t r a , r p A d D i o b r r g o e y O o v r c r a e E t a O o d m T t : r o O R 1 r a 0 y r D 4 e - e 2 1 2 E 3 -2 c . R 2 0 C 1 5 h n o l o o g g l y y i t . . c c a o o r m m y T e c h n o l o g y . c o m O IN ff I ic T e I A of T S E c D ie n B c Y e : S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o ySandia Naotional Laboratories 2015 LDRD Annual Report o y l l l T o T o o T g g Table of Contents g e y e y y e . . . c Program O c verview........................................................................c.................................... c ...............................................................................4 c c S o S o S o h h h e m e m e m Bioscience ........................................................................................................................................................................................................12 nc c n c n r r r e Computing & Information Sciencees ........................................................................................................................e...............................30 o o o t t t M M M l Engineering Sciences ................................................................................................l...................................................................................74 l o o o i i i l l l g i t Geoscience ..............................................................i.t...............................................................g............................................................i..t........108 g a a a yr r y r y y Materials Science ............................................................y...........................................................................................................................1y24 .T T . T . ce Nanodevices & Microsystems .............................................e..............................................................c....................................................164 e c oc c o c o S hRadiatiSon Effects & High Energy Density Sciences .............h..............................................................................................S...........204 h mn n m n m e e e oNew Ideas .......................................................................................................o...............................................................................................234 o c l o c l o c l o Defense Systems & Assessments.........................................................................................................................................................248 g g g r r r e y e y e y Energy & .Climate .........................................................................................................................................................................................326 . c c c t S o tS o S t o M International, Homeland, &M Nuclear Security .................................................................................................................................370 M e m e m e m c c c Nuclear Weapons .......................................................................................................................................................................................412 i r ir r i l e le e l i i i t t M Grand Challenges ....................................tt.. M ..........................................................................................................................t.. M .....................440 t a a a i Exploratory Express ........................................i............................................................................................................................i..............448 r l i lri l i r t t t y a ya a y Unpublished Summaries ..........................................................................................................................................................................478 r r r T y Ty y T T Awards & Recognition .......................................................T.......................................................................................................................48T0 e e e e e e c c cc c c h h h h h h n n n n o on o n On the Cover l l l o o o o o o g g g l Juan Eylizondo Decanini holds two compact, high-voltage nonlilnyear transmission lines. Juan leads a y l o project t.oc exploit nonlinear behavior in materials—behavior tha o t.’cs usually shunned because it’s so . c o S g unpredictabole. (17318 S 2) go S o g e m e m e m y y y c c c Kristina Czuchlewski is principal investigator for the 26-member PANTHER team, which has r . r . r . e c accomplished a number of breeakthroughs in rethinking how to compare moction and treajectories; c t t t M odeveloping software to representM remote sensor images, couple them with adoditional infoMrmation, o S S S i and present them in a searchable forim; and conducting fundamental research on visual cogniition. l m l m l m e i t (165535)e i t i te a a a c r c r rc y This playground structure represents a largyer-than-life nanoporous metal organic framework (MOyF) r T r T Tr to this Sandia National Laboratories research team of (clockwise from upper left) Michael Foster, e e e e ee Vitalie Stavila, Catalin Spataru, François Léonard, Mark Allendorf, Alec Talin and Reese Jones. The c c c t hteam made the first mteasurements of thermoelectrhic behavior in a MOF. (180898 ) h t M M M n n n oJudit Zádor’s KinBot code looks for 3D structures in cheomical reactions to automatically make o i l i l li l preodictions about behavior ofl potential reactions in combuostion for a given molecule. With these ol i i i g g g t predictions, scientists can identitfy the rates at which relevant reactions take place, information that t a y a y ya is critical to understanding combustion. (153342) . . . c c c r r r S o S o S o y y y e Ryan Davis anmd Sandia Neational Laboratories colleagues have developmed a methoed to recycle critical m T c and costly algae cultivationc nutrients phosTphate and nitrogen. (165714) c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 3 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report WEIVREVO MARGORP A message from Sandia’s Chief Technology Officer MARGORP SLAOG S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S On Nov. 5, 199o0, PresidenSt George H.W. Bush signed the National Defeonse AuthoSrization Act for o h h h e Fiscal Year (FY) 1m991, establ e ishing the Laboratory Directed Research andm Developm e ent (LDRD) m nc program. The act authorized D c epartment of Energy laboratnories to allocate a portion o c f their budgets n r r r o e toward innovative research and deevelopment that serves to moaintain their scientific and teechnical o t t t M vitality. M M l l l o o o i i i l l l g i t At Sandia, LDRD-funded work has been ai tmajor contributor to scientifigc understanding and i t g a a a technological advances. Our researchers work together across a broad spectrum of disciplines, yr r y r y y collaborating to advance the frontiers of sciencye and engineering in ways that are critical to Sandia’s y .T T . T . seven national security mission areas, which range from ensuring the safety, security and reliability of ce e c e c the U.S. nuclear stockpile to securing a sustainable energy future. oc c o c o S h S h S h mLnDRD is an essential component of Sandia’s strategy for nsustaining and developing thme world-class n m e e e scieonce, technology, and engineering capabilities needed too respond rapidly to evolving national o c secur l ioty needs asc they arise. LDRD research has contributed l too scientific and technical advancement c l o g g g r at a level that far excreeds what would be expected from a program of its size, less than 6% of Sandia’s r e FY 2015 b y udget. e y e y . . . c c c t S o tS o S t o M e Over the past 25 m years, SanMdeia’s LDRD program has produced significant m advances ine areas such as M m c microelectronics, materials scicence, defense, and advanced radar. These achievementsc, along with i l r e ir le significant intellectual impacts—e.g., highrly e cited i l i t t M i ttM peer-reviewed publications, patents, professti M onal i t a asociety fellowships and awards, and R&D 100 Awards a i i i r l i lri —demonstrate how Sandia’s LDRD investments slh i ape r t t t y a yathe scientific landscape and impact the state-of-the-aart y T r y r Ty technologies for national security. r y T T T T e e e e Thee LDRD program has played and continues to play e c c a criticccal part in Sandia’s ability to attract outstanding c c h h h h engineerhs and scientists, by creating an invigorating h n n n and productive research environment. The program n o on o n l has been partilcularly valuable in recent years as l o o o o g Sandia recruits oangd trains a new generation of talented g o l y scientists and enginlyeers. LDRD is a critical tool in y l o . c attracting highly sou o g.hct-after top talent to the Labs. . c o S g o S go S o g e m e m e m y Each year, researchers acroyss Sandia submit proposals y c c c for creative, forward-looking R&D projects that have r . r . r . e c e the potential to greatly benefit ocur nationael security c t t t M o M mission. In this report, you will findo descriptioMns of o S S S i i each of the projects funded in FY 2015, as well asi more l m l m l m e i t e i tinformation on Sandia’s LDRD program. i te a a a c r c r rc y y y r T r T Tr e e e e Rob Leland ee c c c t t t h h Chief Technology Officer h M M M n n Vice President, Science & Technology n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a4 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report PROGRAM OVERVIEW A Snapshot of Sandia’s LDRD Program 380 280 149 [dollars, K] [dollars, million] [projects] MEAN PROJECT SIZE TOTAL LDRD TOTAL LDRD PROJECTS PROGRAM COST 1127 37% LDRD HOURS CHARGED RESEARCHERS THAT BY NEW STAFF (< 5 CHARGED >10% LDRD YEARS AT SANDIA) 278 51% 86 62% of all of all Sandia Sandia LDRD-SUPPORTED POSTDOCS LDRD-SUPPORTED [FY 2011 - 2015] POSTDOC CONVERSIONS [FY 2011 - 2015] 1244 803 247 83 14 REFEREED TECHNICAL PATENTS SOFTWARE R&D 100 PUBLICATIONS ADVANCES ISSUED COPYRIGHTS AWARDS [CY 2011 - 2014] [FY 2011 - 2015] [FY 2011 - 2015] [FY 2011 - 2015] [CY 2009 - 2014] 30% of all Sandia 48% of all Sandia 50% of all Sandia 24% of all Sandia 74% of all Sandia MARGORP SLAOG S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S Sandia Natioonal LaboraStories, like all DOE/NNSA laboratories, is choarged withS working on tough o h h h e technical probmlems on be e half of the nation. Sandia’s LDRD program is man essentia e l element of m nc the Laboratories’ intent to p c rovide “exceptional service inn the national interest.” As c Sandia’s sole n r r r o e discretionary R&D program, LDeRD is foundational, leading-oedge R&D that nurtures aned enhances o t t t M core science and engineering capaMbilities, supports national security missions, and leads tMo the l l l o i creation of new capabilities. i o i o l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c oc c o c o S h S h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r FY 2015 LDRD Proir gram Statistics r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y LDRD and Early Career Staff Developmenty y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M Intellectual PrMoperty Resulting from LDRD M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 5 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report WEIVREVO MARGORP Program Structure Sandia’s research strategy arises from its laboratory strategy and is organized through program elements known as Investment Areas, each of which is focused on discipline- or mission-based research priorities set by upper management. The LDRD program elements mirror this structure. The Research Foundation Investment Areas provide cutting-edge foundational support for all of Sandia’s strategic national security missions. Mission Foundation Investment Areas create and nurture the ability to provide innovative solutions for NNSA, DOE, and other Federal agencies. Investment Area Mission Impact and/or Laboratory Capability F R snoitadnuo hcRaese Analyze, understand, and control the functions of biological systems in order to Bioscience reduce global chemical and biological dangers and secure a sustainable energy future. Computing & Advance the state of the art in computer and computational science and engineering, Information Sciences and information and data science relevant to national security. Integrate theory, computational simulation, and experimental discovery and Engineering Sciences validation to understand and predict the behavior of complex physical phenomena and systems. Perform world-class R&D focused on the properties, structure, phenomena and Geoscience processes associated with the earth’s geosphere, hydrosphere, and atmosphere. Nurture foundational materials capabilities by developing methodologies to enable Materials Science new understanding—or create enhanced understanding—of materials that are critical to our national security missions. Perform creative, leading edge, and high-impact R&D to discover new phenomena Nanodevices & at the nanoscale and microscale; and create or prove new concepts, devices, Microsystems components, subsystems, and systems. Radiation Effects & Advance the state of the art in radiation effects sciences, dynamic material High Energy Density properties, high energy density science, inertial confinement fusion, and pulsed power Sciences technology to enable stockpile stewardship and national security missions. Support pioneering research that may lead to game-changing breakthroughs in New Ideas science and technology that could eventually impact national security. F M snoitadnuo noissi S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S o S o S o h h h e m e m e m nc c n c n r r r e e e o o o t t t M M M l l l o o o i i i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c oc c o c o S h S h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e Defense Systems & Deveelop innovative systems, sensors, and advanced science and technoelogy solutions c c c t Assessments to detectt, deter, track, defeat, and defend against threats to our national stecurity. M o M o M o S S S i i i e l i t m Energy &e Climate Develop and clrie t ate capabilities to contribute to the nation’s energmy security andl i te m a resilience, economaic viability, and environmental sustainability. a c r c r rc y y y r T International, r Support innovative scieTnce and technology that enhances our abilities to provide Tr e e Homeland, & Nucelear effective advice, analyses,e technologies, and enterprise-level solutions to manage ee c Security risks from the world’s most dcangerous events. c t t t h h h M M M n n n Nurture a creative and vibrant science, technology, and engineering base to support a oNuclear Weapons o o i l deeip scientific understanding of currelnt and future NW products. li l o l o ol i i i g g g t t t a Grandy Challenges Address b a old science, technology and enginyeering challenges and provide ya . breakthrough solutions to critical national sec.urity challenges. . c c c r r r S o S o S o y y y e m eAnswer a key research question, within a relatively smhort timeframee, in an area of m Exploratory Express T c cucrrent or future str T ategic importance to Sandia. c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a6 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report PROGRAM OVERVIEW S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g Projgect SeleFcrtoimon I adnead tOov Perrsoijgehctt g e y e y y e . . . c c c c c c S Each year, thoe LDRD prSogram issues a Labs-wide Call for Ideas organoized througSh the Investment Area o h h h e leadership teamms. In resp e onse, staff members generate ideas and propmosals that e are directed to the m nc appropriate Investment Are c a selection committee for evnaluation. c n r r r e e e o o o t t t M The Sandia LDRD program is highMly competitive. In FY 2015, 866 short idea proposals weMre l l l o i submitted; the Investment Area seleiction committees invited 16o6 of those to submit full proiposals. o l l l g i t Ultimately, 71 new projects were fundietd, with 51 additional projectgs funded throughout the i t g a a a fiscal year. When added to ongoing projects, 380 projects were active in FY 2015. Each proposal yr r y r y y undergoes a rigorous review process, includying peer review by subject-matter experts. y .T T . T . ce e c e c NNSA guidance—as well as SNL’s internal LDRD processes—provides the framework for review oc c o c o S hand seSlection of Sandia’s LDRD portfolio. Idea evaluhation criteria include: alignment with SandSia h mnstrategy and potential impact to Sandia; technical mernit and feasibility; and leadinmg edge, high risk n m e e e oR&D character. For each idea selected, the respective Inovesment Area appoints a Portfolio/Project o c M l aonager whoc has the appropriate knowledge, experience l , oand position to leverage the potential c l o g g g r impact of the R&rD. Multi-year projects are reviewed each year, and must show sufficient technical r e progres y s and progreammatic alignment in order to receive conti y nued funding. e y . . . c c c t S o tS o S t o M e The Grand Ch m allenges pMoertfolio is an important investment for Sandia m , and there eare special M m c requirements for these largcer projects, beyond a longer, detailed proposal and a mocre rigorous review i l r e process. For example, they m irule st form an External Advisory Board (EAB) to guide clarrit e y of project i l i t t M vision and appropriate technicailt t f M ocus. The EAB critically reviews the project’s technicalt p M rogress i t a and planned R&D activities, and praovides insight about potential applications throughout the project a i i i r l i lifetime. lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e FY 2015 Idea and Project selection process and statistics. t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 7 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report WEIVREVO MARGORP S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Missiogn-Enabling Research g g e y e y y e . . . c c c c c c S LDRD projectso are chosenS for their technical o S o h h h e quality, their diffemrentiating e and programmatic m e m nc value to Sandia, and their relev c ance to DOE/ n c n r r r o e NNSA’s missions, as well as the naetional o e o t t t M security missions of the DepartmentM of M l l l o i Homeland Security, the Department of i o i o l l l g i t Defense, and Other Federal Agencies. i t g i t g a a a yr r y r y y The dollar amounts are greater than the FY y y .T T . T . 2015 program cost, since many projects are ce e c e c expected to benefit more than one mission and oc c o c o S hare thereSfore counted more than once. h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S Research Hioghlights tS o S t o M M M e m e m e m c c c i r Predictive Assessment of Stateir of Health and Lifetime of Components r i l e le e l i i i t t M Paiboon Tangyunyong, PI ttM t M t a a a NNSA’s stockpile stewardship mission is aimed at ensuring i i i r l i the safety, security, and reliability of wealrpions in the absence l i r t t t y a of underground nuclear tests. Understandiynag how new and a y T r y existing weapons components will behave thrr To y ughout the r y T T life of the system is critical to maintaining the stoTckpile. By T e eperforming accelerated aging tests on componenetse, researchers e e c ccan observe device behavior. In the absence of obvioccus c c hdegradation, other techniques must be used to understhand and h h h h pnredict aging effects. Researchers are using Sandia-devenloped n n pow o er spectrum analysis (PSA) to detect electrical differen onces o n l l l in devoices and determine whether PSA can detect aging effectos o o o o when degvices—such as commercial-off-the-shelf discrete g g l devices, di y odes, and capacitors—are subjected to accelerated ly y l o . o. . o life tests at elcevated temperatures and voltages. Initial results HIghc-magnification scanning electron c S g suggest the meothod can p S otentially be used to study aging micros go copy image o S f a focused ion beam o g e m e m e m y effects as a standalone technique or as a complementary cross-sectiyon from a bond-wire area in an y c c c r . technique to existing electrical t r esting methods, providing a unaged diode.. r . e cuseful tool for stockpile assurancee. c e c t t t M o M o M o S S S i l mPANTHER: Pattern Analytics to Suppiolrt High-Performance Exploitation and Rmeasoningi l m i i i e t e t te a Kristina Czuchlewski, PI a a c r c r rc y PyANTHER seeks to support high-consequence decision-y r T r T Tr making by unifying and advancing science across three e e e e ee key technical domains: sensor extraction, big data c c c t h t analytics,h and human analytics. The project results will h t M M M n enable analnysts to examine mountains of historical and n o current remotoe sensing data that would otherwise go o i l i l li l o l untouched, whileo also gaining meaningful, measurable, ol i i i g g g t t and defensible insights into overlooked geospatial- t a y a y ya temporal relationships and patterns. Capabilities . . . c c c r S o S rdeveloped by the PANTH o ER LDRD tSeam are being used o r y y y e m e for specific applications in nmew projectse sponsored by m T c c vTarious organizations. c T r r r e e e e e e t t t c M This image shows PANTHER’s geometric anMd temporal trajectocry analyses of air traffic patterns from 43,000M flights over c the continental United States on April 4, 2014. (Credit: Sandia researcher Andy Wilson) h i l i l h i l h i i i t t t n a8 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report PROGRAM OVERVIEW S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g gMission-Enabling Research g e y e y y e . . . c c c c c c S o S o R S esearch Highlights o h h h e m e m e m nc c n c n r r r e Radiography Signature Sceience of Homemade Explosives e o o o t M John Parmeter, PI t M t M l l l o o o i l The ability to accurately screen baggialge i l i i i gt for explosive materials is critical to aviattion g t g a a a security. While considerable research in this yr r y r y y area has focused on the detection of traditioynal y .T T . T . explosives, research on the X-ray radiography of ce e c e c homemade explosives (HME) has received less oc c o c o S h attentSion. In this project, Sandia researchers h S h mnused multi-energy computed tomography (CT) n m n m e omeasuremeents and theoretical calculations to o e o c invl o estigate thce X-ray attenuation properties of l o c l o variogus liquid and powder HME, demonstrating g g r r r e excelleynt agreemen e t between experiment and y e y theory in. many cases. Work was also carried . . c c c t S out in the deovelopmenttS of novel algorithms o S t o M M M e for the analysims of raw radeiography data. The m e m c project concluded in Septemcber 2015, and the Samples awaiting X-ray. c i l r e project team will continue re i sr le e arch on the X-ray r e i l i t t M radiography of various explosiviett s M as part of the Open Threat Assessment Platform (OTAtP M ) project. i t a a a i i i r l i lri l i r t t t y a ya a y T r y Sandia’s Twistact Technology: The Kr Te y y to Proliferation of Wind Power r y T T T T e Jeff Koplow, PI e e e e e c c Wccind power represents a significant renewable c c h h h h energhy source—however, traditional wind turbine h n n n generator architecture (gearbox-based) is difficult to n o on o n l scale up, olften resulting in failure to key components l o o o o g at multi-meoggawatt (MW) operation. Direct-drive g o l y generators, whlyich are less complex and lower y l o . c maintenance at o m. culti-MW scales, have traditionally . c o S g o S relied on rare earthg omagnets (v S ery high cost, reliable) o g e m e m e m y or high-current slip rinygs (short operational lifetime, y c c c high maintenance) to transmit power. Twistact r . r . r . e c e technology is a new architeccture for hiegh-current slip c t t t M o M rings, connecting an electrical ocircuit betwMeen moving o S S S i i and non-moving parts. Twistact eliminates thie need l mTwistact-based genreator technology will bel designed m l m e i t a to be a moduelar drop-in electrical generator fo i tr anext- for rare earth wind turbine magnets, addressinigt a a e critical technological vulnerability to US economic c r generation windc turbines. r rc y ysecurity identified by the DOE (2011 Critical y r T r T Tr Materials Strategy). Twistact consists of an electrically conductive belt and a transmission device e e e e ee that provides a continuous ultralow resistance path for current flow. The technology provides pure c c c t hrolling contact; direct tmetallic contact, with negligibhle voltage drop; a large electrical contact area; h t M M M ntwo parallel current paths; and extremely effective thnermal management. Twistact was selected for n oparticipation in the DOE’s LapCorp program. LabCorp aoims to accelerate the transfer of innovative o i l i l li l cleoan energy technologies froml the DOE’s National Laboraotories into the commercial marketplace. ol i i i g g g t Additionally, the technology recetntly won an Oustanding Technology Development Award from the t a y a y ya Federal Laboratory Consortium for Technology Transfer (Far West Region). . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 9 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report WEIVREVO MARGORP S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e A Worlyd-Class Research Communeity y y e . . . c c c c c c S o S o S o h h h e Sandia’s specializmed mission e s require highly motivated, qualified staff witmh deep exp e ertise, committed m nc to advancing the frontiers of s c cience and engineering throungh continual growth and de c velopment. n r r r o e The LDRD program supports somee of Sandia’s most accomplisohed scientists and engineeres, as well as o t t t M many promising early career researcMhers. M l l l o o o i i i l l l g i t Jon Madison i t Abraham Ellis g i t g a a a Black Engineer of the Year Great Minds in STEM, Outstanding Technical yr r y r y y Jon’s current NNSA-fundedy Achievement y .T research in 3D materials TSandia’s research on int.egration of solar and T . ce science is being used to eother renewables into the c grid has grown e c oc design better high-reliability c considerably, due in part to Aobraham’s c o S h S h S h components and systems for contributions as a researcher, team lead, and mn n m n m e enuclear weapons and other deparment manager. e o o o c l o c complex engineering systems. l o c l o g g g r r r e Somuri Pryasad e Susan Rempe y e y Asian Amer.ican Engineer of the Year . New Mexico Women of . c c c t S Somuri has rep o eatedly brtoSken new ground in the o InfluencSe Award t o M M M e understanding ofm friction aned wear in materials, m“LDRD hase had a major m c and made substantial contribuctions to national impact on myc career by i r security programs. ir funding researcrh that helps i l e le e l i i me and my colleagues find i t a t M Pavel Bochev ttM a innovative solution t sM to global t a i DOE Ernest Lawrence Award i problems in cancer druig r l i t “I am deeply honored to receive this award, l ri t therapy, water purificat l ioi tn, r y a ya a y which is a testament to the exceptional research and carbon dioxide capture.” r r r T y opportunities provided by Sandia and DOE. Since Ty y T e T joining Sandia, I’ve been very fortunate to interact T e Patrick Feng T e e with an outstanding group of researchers who eAsian American Engineer of the Year e c c stimulated and supported my work.” cc Patrick’s work focuses primarily on the c c h h h h lumhinescence properties of materials, or how h n n n n Maorgot Hutchins light is on emitted in response to various stimuli. In o n l Outstanding Young Engineer several olf his projects, Feng and his team develop l o o o o g Award, SME organic-boa g sed materials for the detection of g o l y Margot currently conducts fast neutrons ly , the signature for a variety of y l o . systems analysis for fissionable maot.erials. . o c c c S g onationaSl security, including go S o g e mcyber reseilience of critical Hongyou Fan m e m c y infrastructucre and Fred Kavli Distinguisyhed c y r . internationalr engagement on Lecture in Nanoscience., r . e e e t c the implementattion of nuclear Materials Research c t c M o detection architec M tures. Society o M o S S S i i Hongyou’s pioneering i l m l m l m e i t Steve Slutz e i t research in the field of i te a a a Fellow of the American Physical Society nanoparticle assembly c r c r rc y Steve’s MagLIF concept (magnetized liner inerytial and integration has y r T fusion), proposed in r 2010, is currently providing T supported a paradigm Tr e e realistic data about ne e utron production, a key eshift from nanoscience ee t c component of nuclear fustion. cdiscovery to practical c t h h h M M nanotechnologies. M n n n Daon Sinars o o i l Fellolwo of the American Physica i l S l ociety Tamara lK o olda l o i l i Dan’s cigtation reads, “For scientifici contributions Fellow, Soc g iety for Industrial and Applied g i t t t a and leadeyrship in the development ofa innovative Mathematics y ya X-ray radiog.raphy and spectroscopy diagnostics “Initial funding f.rom Sandia’s LDRD program . c c c r S for the study oof z-pinch phSysics, inertial r allowed us to devel o op techniquSes that are profoundly o r y y y e confinement fusimon, and highe energy density valuable in scientific amnd nationael security data m T c physics.” c Tanalysis programs.” c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a10 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report PROGRAM OVERVIEW S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y eA Programy Rich With Collaboration y e . . . c c c c c c S o S o S o h h h e Collaborationms between S e andia’s LDRD researchers and universities, mnational lab e oratories, m nc government agencies, and i c ndustry enhance Sandia’s futnure capabilities through an c influx of n r r r o e knowledge and skills. Sandia’se partners also benefit from coollaborative research resultes and o t t t M interaction with peers from outsidMe their organization. In addition to furthering science, pMartnerships l l l o i with academia often provide an impoirtant opportunity for Sandoia to recruit the world-class sicientists o l l l g i t and engineers needed for developmenitt of mission-critical lab capabgilities. Partnerships with ini tdustry g a a a enable technological breakthroughs developed through the LDRD program to be commercialized yr r y r y y under licensing agreements and brought to ymarket for the US public good. y .T T . T . ce e c e c In FY 2015, LDRD researchers at Sandia collaborated with nearly 100 experts across the US, as oc c o c o S hshownS below. Each type of partership is color-codedh. The size of each symbol corresponds to tShe h mnnumber of unique parterships at an individual institutnion. m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S Map by ESRI o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l i mResearch Excellence in Selrivice of the Nation m l i m e t e t te a a a c r c r rc y LDRD supports Sandia’s mission by investinyg in leading-edge research that advances the frontiers oyf r T r T Tr science and engineering critical to national security. The program is also instrumental in attracting e e e e ee and developing a world-class workforce of scientists and engineers, the people who make it possible c c c t hfor Sandia to achieve itts mission and goals. h h t M M M n n n oTo learn more, visit www.sandia.gov/ldrd. o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 11 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S o S o S o h h h e m e m e m nc c n c n r r r e e e o o o t t t M M M l l l o o o i i i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c Bioscience oc c o c o S h S h S h mn n m n m e e e Theo overarching goal of the Bioscience Investment Area is oto develop new competencies in biological o c scien l coe to addrescs two application areas in Sandia’s broad na l toional security mission — biodefense and c l o g g g r emerging infectiousr disease, and biofuels. r e y e y e y . . . c c c t S The research in o biodefentsSe includes developing better ways to detect, o characterizSe, and contain t o M e harmful pathoge m ns. The strMaetegy integrates advanced technologies with m an understaending of human M m c health and immune response. Tche goal is to improve the response to disease outbreakcs and to limit i l r e their spread. ir le r e i l i i i t t M ttM t M t a The research regarding the nation’s realiance on fossil fuels focuses on developing efficient, a i i i r l i economical biofuels that can replace or rlr e i duce current gasoline, diesel, and aviation fuel l i r t t t y a consumption. The research includes two soyaurces of energy: lignocellulose, or dry plant matter, and a y T r y algae. The aim is to find efficient and economir Tc y al methods to convert lignocellulose into fuels and to r y T T understand the factors that govern algal pond stTability and identify molecular mechanisms that can T e e e ebe used for lipid/fuel production. e e c c cc c c h h h h Pnrojects h n n h n o on o n l l l o Bio-Emoulative MOF-Based Lignin Degradation Catalysts ......................o.o...................................................................................14 o o g g g l Coupling Cyhemical Energy with Protein Conformational Changes to Tranlyslocate Small Molecules across y l o . o. . o Membracnes .....................................................................................................................c.....................................................................16 c S g o S go S o g e CRISPR Technologmy for Biodefeense and Emerging Infectious Disease Countermemasure Deveelopment .................17 m y y y c c c r e . Discovery of Anti-Viral Inhibitors ag r aeinst the Chikungunya Virus nsP2 Protease Dom.ain................. r ...e.......................18 . c c c t t t M oEKSG: A Universal Sample Prep TechnoMlogy for Multidimensional Bioscience ........................o.........................M................19 o S S S i i i l m l m l m i In Vivo High-Throughput Transcriptomics to Eilucidate the Spatial and Temporal Dynamics of Host- i e t e t te a a a Pathogen Interactions ....................................................................................................................................................................20 c r c r rc y y y r T Metal Organic Framerworks for Targeted, Triggered, STustained, and Systemic Delivery of Antibiotics ...................21 Tr e e e e ee cPredictive Pathogen Biology: Genome-Based Prediction ocf Pathogenic Potential and Countermeasures c t t t M hTargets ................................M..............................................................h....................................................................................................23 h M n n n Recoombinant Vesicular Stomatitis Virus for Therapeutic Antibodoy Epitope Mapping and Vaccine o i l i l li l Doevelopment ..................................l................................................................o....................................................................................24 ol i i i g g g t t t a Systems-Leyvel Synthetic Biology for Advaanced Biofuel Production ............y...........................................................................25 ya . . . c c c r r r S The Engineeringo and UndersStanding of Nanoparticle/Cellular Interactions ......o.........................S........................................26 o y y y e m e m e m T c Understanding and Engineering Lcignolysis for RenTewable Chemical Production...............................c..............................28 T r r r e e Unknown Pathogen Detection in Clin e ical Samples: A Neovel Hyperspectral Imaging and Single Cell e e t t t c M Sequencing Approach ........................M....................................c.........................................................................................M................29 c h i l i l h i l h i i i t t t n a12 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S o S o S o h h h e m e m e m nc c n c n r r r e e e o o o t t t M M M l l l o o o i i i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c oc c o c o S h S h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e This page inteentionally left blank. e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 13 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Bio-Emgulative MOF-Based Lignin Degradation Catalysgts g e y e y y e 180812 | Yea.r 1 of 3 | Principal Investigator: M. D. Allendorf . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc Lignin is the most abundant so c urce of renewable aromaticsn, with 200-300 Mtons/year c projected n r r r o e production by a US biofuels indusetry, that would process ~1B toons of biomass to meet DOeE goals. o t t t M However, there are currently no efficMient processes for extracting these aromatics and conveMrting l l l o i them to value-added chemicals and droip-in fuels. The technical ando economic challenges are i o l l l i i i gt staggering, due to the quantities of materital involved and lignin’s recalgcitrance to depolymerizatiotn. g a a a yr Conventional lignin degradation processes usre aggressive reagents and arye energy intensive (400- r y y y y 800°C) and yield complex product mixtures. Milder reaction conditions and narrower product .T T . T . cedistributions could be achieved using lignin-degradeing enzymes, but these are ctoo fragile to be e c ocpractical for large-scale biorefining. The objective of cthis project is to develop lignoin valorization c o S h S h S h mmnethods in which oxidative solubilization provides a feednstock for industrially robusmt catalysts n m e e e basoed on metal-organic frameworks (MOFs). MOFs are nanooporous materials with exceptional o c synth l eotic versaticlity arising from a structure comprised of me l otal ions linked by rigid organic groups. c l o g g g r Our strategy is to emrulate natural lignin degradation, wherein extra-cellular fungal enzymes r e y e y e y generate rea.ctive species to partially degrade lignin. The resulting, l.ow-molecular-weight products . c c c t S are metabolizeod by lignint-Sutilizing microorganisms, such as Sphingobiumo sp. SYK-6S. We will employ t o M M M e computational bimology and ine situ diagnostic probes to obtain structural amnd energeteic knowledge of m c c c enzyme-catalyzed reactions that we will translate into MOF structures. The lignin pretreatment will i r ir r i l i e employ an inexpensive homogene leoius catalyst (e.g., Fenton reaction) or chemical treatmeent to control l i t t M the molecular weight distribution o t f t Msolubilized products, thereby accelerating reaction wit t hM the MOF t a a a i and improving selectivity. i i r l i lri l i r t t t y a Conventional catalysts for lignin degradati y oan are either costly (e.g., Pt-group metals) or difficult to a y r r r T y separate from reaction products (e.g., vanadiuTmy complexes). New valorization strategies, combining y T T T T e the best features of enzymes with the robustnesse of heterogeneous catalysts, require fundamental e e e e research to understand reaction mechanisms and develop novel catalyst chemistries. c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a14 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Consgolidated Bioprocessing and Biofuels Productiogn Platform g e y e y y e 165822 | .Year 3 of 3 | Principal Investigator: R. W. Davis . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc Depleting fossil fuel reserve c s and environmental concernns are the major catalysts f c or research into n r r r o e alternatives for transportatione energy that are renewable aond carbon neutral. To achieeve current US o t t t M renewable fuels goals, approximaMtely 1 billion tons of residual biomass would need to be cMonverted l l l o i to biofuels. This research aims to rediuce biofuel production costos by building a consolidated i o l l l i i i gt bioprocessing (CBP) platform, reducingt processing steps. We propogse to utilize the proteolytict g a a a yr bacterium Bacillus subtilis as a chassis to enrgineer a microbial bioreactoyr that both degrades biomarss y y y y feedstocks and produces advanced infrastructure compatible biofuels. We will consolidate biomass .T T . T . ce pre-treatment and advanced biofuel productione to a single bioreactor platfcorm. CBP will reduce unit e c ocoperations and increase process efficiency by reduccing mass transport phenomoena and processing c o S h S h S h mnsteps and costs. Our proposed CBP chassis, B. subtilis,n produces various enzymes imncluding amylase n m e e e oand various proteases which make it ideal for CBP of higoh protein biomass, especially microalgae. o c Al l tohough B. sucbtilis accepts a variety of substrates for bioc l oonversion, optimization of amino acid c l o g g g r conversion to fuerls requires modification of carbon flux pathways and minimization of stress r e y e y e y response. genes. Therefore, the goal of the research with Univers.ity of California-Los Angeles is . c c c t S to: 1) investoigate comtbSinatorial genetic knock-outs of genes involveod in sporulSation and 2) use this t o M M M e platform for CmBP of proteein rich cellulosic biomass to produce advancmed biofuelse using the amino acid m c c c transamination pathway to produce fusel alcohols. i r ir r i l e le e l i i i t t M Summary of Accomplishments: ttM t M t a a a i We identified the amino acid consum i ption profile of B. subtilis, including fermentation inhibit i ors r l i t and limiting reagents. It was found tha lritt glutamine, cysteine, and proline were completely con l siutmed r y a ya a y within 48 hours; asparagine and serine were completely consumed within 96 hours. Within 144 r r r T y Ty y T hours, aspartic acid and arginine were completely consumed, while alanine and isoleucine were T T T e e ~75% consumed. Cysteine was found to be thee oenly essential amino acid required for robust growth e e c cof B. subtilis. In an effort to emulate the nisin-induccced phenotype (i.e., prevent starvation stress), we c c h h h h generated the strain B. subtilis DcodYDbkdBDrelA. Delhetion of stringent-response regulator/ppGpp(p) h n n n n osynthase relA in B. subtilis DcodYDbkdB led to increased bon iofuel production. This strain exhibited o n a dl oelayed response to amino acid starvation as evidenced lwoith the approximately 80% increase l o o o o in biogfuel titers. Furthermore, the strain achieved 60% of theg maximum NH yield and 28% of the g l y ly 3 y l o maximum . biofuels yield from amino acids and exhibited a partiaol- . starvation resistance that moves . o c c c S g towards com o plete consSumption of protein; maximum conversion tit go ers were acShieved within 48 o g e hours. The resmults indicatee that stress gene knockdown should be conmsidered as ae general strategy for m y y y c increasing biofuel yields frocm bioconversion of biomass c r . r . r . e e e c c c t Significance: t t M o M o M o S The R&SD effort and external collaboration bolsters biofuels and renewable energy for the eneSrgy i i i l m l m l m i security mission and assists with capabiility development for biomanufacturing with synthetic ibiology e t e t te a a a platforms. The results have been key to obtaining follow-on funding from DOE’s Office of Energy c r c r rc y Efficiency and Renewable Energy-Bioenergyy Technologies Office as a part of the Annual Operating y r T r T Tr Plan portfolio for Algae Biomass and Feedstock Logistics. e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 15 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Coupligng Chemical Energy with Protein Conformatiognal Changes to Translocate g e y e y y e Small Molecules across Membranes . . . c c c c c c S o S o S o h 180817 | Year 1 of 3 | Principal Investigator: S. Rempe h h e m e m e m nc c n c n r Project Purpose: r r e e e o All living cells rely on continuous exchange of diverse moleculaor species (e.g., nutrients, precursors, o t t t M M M l and reaction products) across cellular membranes for their normall function and survival. Membrane l o o o i l transporters are specialized molecular di levices that provide the machinery for selective and i l i i i gt t g t g a efficient transport of materials across the maembrane. The biological significance of membrane a yr transporters cannot be overstated, given theirr central role in a myriad of kyey cellular processes and r y y y y .T their subsequent targeting by a large number of Tpharmaceuticals. The signifi.cance of membrane T . cetransporters is also evident from a recent major foecus shift of experimental str c uctural biological e c oc c o c o S hstudies toSwards characterizing structural states arisingh during the function of these proteins. S h mCnurrently, our understanding of active membrane transpnort is limited. Experimental mmethods provide n m e e e o o o only snapshots of the energy-coupled transporter activity. Our goal is to obtain an understanding of c l o c l o c l o the full mechanism. Specifically, we endeavor to understand how chemical energy in a cell couples g g g r r r e with the myechanics of e protein conformational changes to producey directional translocation of a e y specific subs.trate across cellular membranes. To achieve this unders.tanding requires a dynamical . c c c t M S description of sotructural t t S rMansitions during the transport process and tohe free ene S rgy barriers t M o e m e m e m along the transport pathway. We propose to use molecular simulation to achieve those goals. Upon c c c i r successful completion, the projeirct will provide, for the first time, a detailed mechanistic rdescription i l e le e l i of the entire transport cycle for meimbrane transporter. The project uses Sandia’s high performance i t t M ttM t M t a computing capabilities, and extensivea expertise in molecular simulation of proteins. The work is in a i i i r l i collaboration with University of Illinois alrt i Urbana-Champaign. l i r t t t y a ya a y The research is high-risk due to challenges with determining complex pathways in biological r r r T y Ty y T reactions. The result will be new insights for directing transport of specific molecules important to T T T e eenergy and biodefense across membranes. e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a16 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T CRISgPR Technology for Biodefense and Emerging Ignfectious Disease g e y e y y e Countermeasure Development . . . c c c c c c S o S o S o h 180811 | Year 1 of 1 | Principal Investigator: O. Negreteh h e m e m e m nc c n c n r Project Purpose: r r e e e o Two urgent problems threaten national and global biosecuroity: 1) accelerating emergence of highly o t t t M M M l virulent, transmissible, and drug-resistant pathogens and 2) glolbally available, low-cost tools for l o o o i l creating and re-engineering organismi ls, increasing greatly the odds of the accidental or intenitlional i i i gt t g t g a manufacture and release of deadly pathoagens. The hallmark of these biothreats is genetic noveltay yr that evolved naturally or was introduced dreliberately to enhance viruleynce and multi-drug resistanrce, y y y y .T rendering existing countermeasures ineffectiTve. Traditional drug discover.y strategies simply cannot T . ce keep up with naturally occurring or potentially eengineered antiviral/antimic c robial resistance. New e c oc c o c o S happroaSches are desperately needed to develop effehctive countermeasures against a wide variSety of h mninfectious diseases. An ideal countermeasure would bne easily modified to target dmifferent pathogen n m e e e o o o classes and strains, and require minimal information about the pathogen. Clustered regularly c l o c l o c l o interspaced short palindromic repeats (CRISPR) is a method that can be used for genomic editing, g g g r r r e gene inyactivation/a e ctivation, or suppression of viral RNA replicyation with a two-component system e y that cons.ists of a RNA-guided endonuclease Cas9 and a guide RN.A (gRNA) complementary to the . c c c t M S nucleic acido target. Ou tS r Mmain goals for this project were to develop roegents and S protocols for applying t M o e m e m e m CRISPR-based countermeasures against virus infection. c c c i r ir r i l e le e l i Summary of Accomplishmentsi: i t t M We first used a host-directed CR tt IMSPR countermeasure strategy targeting the Ebola virus t M(EBOV) t a a a i receptor with our BSL-2 EBOV virus imodel (VSV-EBOVGP-GFP). We designed gRNA sequenices r l i lri l i r y t a targeting 600bp, 1200bp and 2100bp itnyato the open reading frame of human NPC1 gene and thtean y T r y used them along with Cas9 to create genormTy e-edited NPC1 knockout cells using VSV-EBOVGP- r y T T GFP as a selection pressure. This result demoTnstrated that targeting NPC1 using CRISPR systems T e e e e would prevent infection by Ebola GP –VSV. Addeitionally, we generated CRISPR reporter cells for e c c cc c c high-throughput efficiencies measurements. The design of the reporter places a target CRISPR cut h h h h nsite between a red fluorescent protein (RFP) and grehenn fluorescent protein (GFP) genes with the n h n oGFP gene out of frame. Endonucleases targeted to the conut site will induce frameshift to move GFP o n l l l o in forame. We have made single cell clones of these construco ots in HEK293T and HepG2 cells. Lastly, o o g g g we have made lentivirus-pooled libraries for CRISPR knockout (KO) screening that can identify host l y ly y l o factors in.volved in virus infection similar to genome-wide RNAi ot.echnology. However, CRISPR KO . o c c c S g screening iso a more robSust approach for identifying viral-host interagoctions dueS to the fact that RNAi o g e gene silencingm is typically e incomplete and CRISPR can completely supmpress gene e expression. We m y y y c c c r . expanded the GeCKO v2 librarry, verified the coverage of the plasmid expa . nsion usingr MiSeq Illumina . e e e c sequencing and have begun screening in HEK293T cells. In all, we developecd regents and protocols c t t t M ofor CRISPR-based countermeasur M es using this technology. o M o S S S i i i l m l m l m i i i e t Significancee: t te a a a c r The proposedc research directly addressesr the national security mission, specifically in the realm orf c y y y r T biosecurity and p r rotection against bioterrorisTm. The research supports DoD’s mission to develop Tr e e medical countermeeasures to ensure robust defeense against biological warfare threats. ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 17 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Discovgery of Anti-Viral Inhibitors against the Chikunggunya Virus nsP2 Protease g e y e y y e Domain . . . c c c c c c S o S o S o h 186364 | Year 1 of 3 | Principal Investigator: B. N. Harmon h h e m e m e m nc c n c n r Project Purpose: r r e e e o Chikungunya virus (CHIKV) is an emerging mosquito borne alpohavirus that causes devastating o t t t M M M l arthritic disease. CHIKV has been responsible for major outbreakls in Africa and Asia leading to l o o o i l serious morbidity and mortality in humai lns. The first cases of CHIKV were reported in the westei lrn i i i gt t g t g a hemisphere in late 2013 and nearly 1 millioan suspected CHIKV infections have occurred in the a yr Americas in the last year. The CHIKV mortalitry rate has been estimated toy be 1:1000. The US Nationarl y y y y .T Institute of Allergy and Infectious Disease (NIAITD) lists CHIKV as a category. C biodefense priority T . cepathogen due to the ability to engineer this pathogeen for mass dissemination. A c t present, there are no e c oc c o c o S hlicensed vSaccines or treatment options to combat CHIKhV infection, and research in the developmSent h monf therapeutics has been limited. To develop timely countnermeasures against this biomlogical threat, n m e e e o o o this project seeks to discover lead compounds for the treatment of CHIKV. Specifically, we will c l o c l o c l o discover drug-like small molecules that are inhibitors of the viral nsP2 protease (nsP2pro) using g g g r r r e advanced ybioinformat e ics, cheminformatics and experimental screyening methods. Viral nsP2pro e y is a cysteine. protease encoded as part of the viral non-structural po.lyprotein, and is essential for . c c c t M S processing of tohe polypro tS teMin into functional proteins and for viral reprooduction. A S t the end of the t M o e m e m e m proposed two year project, our goal is to have in hand small molecule inhibitors that meet three c c c i r criteria: 1) effective inhibition oirf nsP2pro proteolytic activity, demonstrated in vitro andr in vivo, 2) low i l e le e l i inhibition of human proteases andi low cytotoxicity, and 3) robustness to resistance due to sequence i t t M ttM t M t a mutations. a a i i i r l i t We seek to discover antiviral methods, in lrictluding the development of a novel FRET-based substra l itte r y a ya a y for high-throughput screening and computer-aided modeling and structural activity relationship r r r T y Ty y T for down selection of candidate inhibitors, to allow the identification of compounds that inhibit T T T e ensP2pro at nanomolar concentrations in vitro, anede inhibit disease in animal models. Unique combined e e c cexpertise in biodefense, modeling/simulation can accccomplish these objectives. c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a18 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T EKSGg: A Universal Sample Prep Technology for Mugltidimensional Bioscience g e y e y y e 173020 | .Year 2 of 2 | Principal Investigator: A. Hatch . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc Systems-level understandin c g of biological states and disnease mechanisms typically c requires n r r r o e multidimensional analysis of keey samples. Unfortunately, moultidimensional analysis is oeften difficult or o t t t M impossible using current sample pMreparation methods, due to inherent challenges purifyinMg multiple l l l o i analytes from small samples (e.g., humian clinical, animal model).o Each method generally recoivers only o l l l i i i gt certain sub-classes of analyte, yet splittting sample for multidirectiognal processing is problemattic or, g a a a yr for small samples, not feasible. Moreover, trreatments of sample and saymple-derived analytes can lread y y y y to analyte degradation and other undesirable effects (e.g., mixing of intra- and extra-cellular analytes), .T T . T . ce making interpretation difficult. We aim to addreess current deficiencies by dceveloping a powerful e c ocnew sample preparation competency. We will devcelop a novel electrokinetic stoep gradient (EKSG) c o S h S h S h mndevice for continuous processing of blood sample inpunt, enabling precise segregatmion of different n m e e e oanalyte classes (nucleic acids, proteins, nanoparticles, ceolls) into separate collection reservoirs o c fo l ro downstreacm analysis. EKSG will be applied to small vo l luomes (≤100 ml) of healthy whole blood, c l o g g g r tuned for simultarneous isolation of informative analytes: 1) mRNA, microRNA, and proteins from r e y e y e y white blo.od cells, 2) microRNA from extracellular vesicles, 3) mR.NA, microRNA, and proteins from . c c c t S extracellulaor RNA/prottSein nanoparticles, and 4) “free” extracellular oDNA, microSRNA, and proteins. t o M M M e We will similamrly apply EKeSG to infected blood for additional isolationm of such anaelytes from viral m c c c or bacterial pathogens. Nucleic acid fractions will be analyzed using quantitative polymerase chain i r ir r i l i e reaction (qPCR) and high-thro leuighput sequencing, and proteins using mass spectrometery. l i t t M ttM t M t a This R&D work explores a novel anad significant “disruptive” sample prep technology that could a i i i r l i dramatically improve a majority of biolrianalytical methodologies. Our novel microdevice configl iurations r t t t y a facilitate a clever enabling technology s y oalution. The possibility of rapidly isolating the multiple a y r r r T y proposed components from a single complTeyx sample is daunting and unprecedented. y T T T T e e e e e e Summary of Accomplishments: c c cc c c hWe validated a new approach to separating constituhents of biological samples. A clear distinction h h h h nwas made between the physical operational principlesn of the EKSG device and alternative methods n n o reported to date in the scientific and patent literature. W one achieved greater than 90% extraction o n l l l o o o o efficiency and greater than 90% purity for isolating multipleo key biological analytes including nucleic o g g g l acids, pyroteins and blood cells from a continuous sample input.ly The extraction process occurs within y l o seconds. .Tche ability to separate these multiple sample constitue o n.tcs with the speed and purity in a . c o S g continuous oformat is un S precedented. go S o g e m e m e m y y y c We designed, built, and testced more than one hundred EKSG microfluidic devices wcith several r . r . r . e iterative design cycles to achieeve the high purity and efficiencies demonstrated. We deesigned and c c c t t t M o built the imaging, fluidics and elecMtronics control system to operate the device o s and imageM the o S S S i separation processes. The operationial principles and scaling laws were explored by COMSOiL, l m l m l m e i t multiphysiecs modeling, to validate the intovelty of our approach, and the overriding physics prinictiplees a a a that enable optimization of devices for high efficiency and purity of separation. c r c r rc y y y r T The methods werre demonstrated with relevaTnt biological samples including blood. The system was Tr e e e e ee also proven to be capable of sorting small microliter volumes important for precious samples and also c c c t hproven to work for mutch larger volumes if desired. h h t M M M n n n oSignificance: o o i l i l li l Maony fundamental approachels in the biodefense mission sopace require sample preparation. The ol i i i g g g t most dangerous pathogens are dtifficult to study and infectious samples are a precious commodity. t a y a y ya Tradition.al sample processing methods waste sample and limit th.e extent of analysis that can be . c c c r r r S performed. oThis technoSlogy enables detailed analysis from these proecious samSples. The technology is o y y y e also faster, anmd leads to meore accurate multi-analyte analysis. m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 19 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T In Vivog High-Throughput Transcriptomics to Elucidatge the Spatial and Temporal g e y e y y e Dynamics of Host-Pathogen Interactions . . . c c c c c c S o S o S o h 173021 | Year 2 of 3 | Principal Investigator: R. Meagher h h e m e m e m nc c n c n r Project Purpose: r r e e e o Transcriptome sequencing (RNA-Seq) has the potential to revoolutionize studies of microbial o t t t M M M l pathogenesis by providing information on the dynamics of gene exlpression in both the host and l o o o i l pathogen. Although host response to baiclterial infection is well studied, the many mechanisms uisled i i i gt t g t g a in concert by pathogenic bacteria to survivea and proliferate within a host are poorly understood a yr for many bacteria. A novel Sandia-developed rapproach called Pathogen Cyapture can enrich r y y y y .T microbial transcripts 100-fold or more, allowingT complete recovery and sequ.encing of the microbial T . cetranscriptome without altering host transcripts. Hoewever, this technique is low c -throughput which is e c oc c o c o S han obstacSle to performing large transcriptomic studiesh. S h mn n m n m e Weo propose teo develop a high-throughput, automated Pathoogen Capture system to enable a host- e o c pathol o gen interacctions study using two model systems: primalr o y cell infection with multi-drug c l o r resistangt Klebsiella prneumoniae (part of an important class of emgerging, difficult-to-treat pathogens r g e with assoc y iated high meortality and morbidity), and an animal infec y tion model of Salmonella e y . . . t S typhimurium,c in o collaboratt S ion with the University of Colorado. By enrcic o hing for mi S crobial transcripts, t c o M M M e we will discover tmhe transcrieptional sequence of events by which these pamthogens inefect the host, m c overwhelm the immune systemc, and resist treatment with antimicrobial drugs. c i r ir r i l e le e l i Our proposed pathogenesis investiigations will require large sample sets, which are uniquely enabled i t t M ttM t M t a by the proposed high-throughput plataform for Pathogen Capture. The device and methods are a r i l i broadly applicable to understanding thei lr d i ynamics of any microbial or viral pathogen in any hosti l i r t t t y a system, and the coupled host/pathogen trayanscriptomic data is a fundamentally new, innovative a y T r y technique to studying host/pathogen interactr Ti y ons. r y T T T T e The proposed research will generate new knowleedge about the pathogenesis of Salmonella and e e e e c cKlebsiella, which have biodefense relevance (Nationaccl Institute of Allergy and Infectious Disease c c h h h h (NIAID) Biodefense Categories B and C). The proof-of-choncept data and automated system h n n n n prooduced by this project will make us attractive partners foon r collaborations with leaders in microbial o n patholgenesis. l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a20 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Metagl Organic Frameworks for Targeted, Triggeredg, Sustained, and Systemic g e y e y y e Delivery of Antibiotics . . . c c c c c c S o S o S o h 173493 | Year 2 of 2 | Principal Investigator: C. E. Ashleyh h e m e m e m nc c n c n r Project Purpose: r r e e e o Engineered nanoparticles promise to revolutionize the prevoention, detection, and treatment of o t t t M M M l biological and chemical threats by enabling development of nexlt-generation adjuvants, affinity l o o o i l reagents, and therapeutics. Nanotecihlnology is especially critical for improving the bioavailabi lility i i i gt t g t g a of antibiotics and antivirals through targaeted accumulation of nanoparticles at sites of infection a yr and subsequent release of therapeutic payrloads in response to an appryopriate stimulus. However, r y y y y .T many existing nanoparticle delivery vehicles, Tincluding liposomes and poly.mersomes, suffer from T . ce limited capacities, uncontrollable release profilees, and complex, specialized c synthesis procedures e c oc c o c o S hthat mSust be re-adapted for each new cargo molecuhle, leading to drug- and disease-specific ‘onSe-off’ h mnapproaches. To address these limitations, we proposen to develop mesoporous silicma nanoparticles n m e e e o o o (MSNPs) and metal organic frameworks (MOFs) for targeted delivery of antibiotics to cells, c l o c l o c l o tissues, and organs infected with the Category A biothreat Francisella tularensis, which causes the g g g r r r e diseasey tularemia u e pon inhalation or absorption. MSNPs and MyOFs have unprecedented surface e y areas, syn.thetic flexibility, and chemical robustness and promise .to yield a new, differentiating . c c c t M S Sandia capaobility that t e S nMables pre-administration of therapeutics too laboratory S , first-responder, or t M o e m e m e m military personnel who are at risk of being exposed to a biological threat. Our proposed effort will c c c i r also facilitate event recoveryir by enabling effective decontamination of surfaces or burildings after i l e le e l i accidental or intentional releasie of a biothreat. Furthermore, MSNP and MOF technologies should i t t M ttM t M t a be easily extendable to delivery ofa nerve agent countermeasures, treatment of traumatic brain a i i i r l i injury, and creation of broad-spectrumlri vaccine adjuvants. Consequently, this work will help els i tablish r t t t y a Sandia as a leader in the fields of nanothyaerapeutics and nanoadjuvants. MSNPs and MOFs promaise to y r r r T y address many of the limitations of current Tnyanoparticle delivery vehicles. External sponsors typicalyly T T require in vivo evidence of a nanoparticle’s poTtential, however. We will demonstrate in vivo proof of T e e e e e e concept that MSNPs and MOFs are superior delivery vehicles for small molecule drugs, including c c cc c c hantibiotics and antivirals. h h h h h n n n n oSummary of Accomplishments: on o n l l l o Ovoer the course of this project, we synthesized MSNPs ando o optimized their hydrodynamic o o g g g diameter, surface area, pore size, and pore chemistry in order to achieve loading capacities for l y ly y l o physicoc.hemically disparate antibiotics that exceed those of othoe.r nanoparticle delivery vehicles . o c c c S g (e.g., liposomoes, polymeSrsomes) by 100 to 1000-fold. We then coategod antibiotiSc-loaded MSNPs with o g e biocompatiblem lipids and t e uned properties of the MSNP core and lipid mcoating to e e nable long-term m y y y c c c r . retention of encapsulated anrtibiotics in neutral-pH buffer and blood, as w . ell as tailorarble release . e e e c rates under intracellular conditions. Next, we demonstrated that modifyingc lipid-coated MSNPs c t t t M owith targeting and endosomolytic M ligands promotes highly selective uptake byo target cells M (i.e., cells o S S S i l mthat Francisella tularensis infects) folloi lwed by intracellular release of encapsulatedm antibioticsi,l both m i i i e t e t te a of which are critical to kill intracellular baacteria in vitro. Finally, we demonstrated that lipid-coataed c r MSNPs, whenc loaded with antibiotics and rtargeted to cells and organs that Francisella tularensis rc y y y r T infects, dramaticrally improve the efficacy of aTntibiotics. Impressively, lipid-coated MSNPs increased Tr e e the survival of mice e infected with Francisella tula e rensis from 0% for the ‘free’ antibiotic to 100%, ee c c c t hwhile also reducing thte dose from 40 mg/kg for the hfree antibiotic to 8 mg/kg, and reducing the h t M M M ndosing frequency from twice per day for the free antibniotic to once every other day. These data n o o o are truly revolutionary given the increasing prevalence of drug-resistant bacteria and are under i l i l li l i pre o paration for submission tol Niature. In addition to our wo o rk with MSNPs, we also synthesized ol i g g g t t t a MOFs a y nd demonstrated that they a have high in vitro biocompat y ibility and colloidal stability and are, ya therefore., promising as bioimaging agents and delivery vehicles f.or antibiotics and other medical . c c c r r r S countermeaosures. S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 21 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Significagnce: g g e Our resultys are the first-ever demonstration that e loading antibiotiycs in nanoparticles and targeting y e . . . c them to sitesc of infection can reduce the efficacious dcose, frequency,c and duration of treatment and c c S o S o S o h even recover the effectiveness of antibiotics against whhich bacteria have developed resistance. These h e m e m e m nc revolutionary data will accelercate Sandia’s ability to help renspond to outbreaks causedc by emerging n r r r e pathogens or biological weapons ein a rapid and cost-effective fashion. They will also estabelish Sandia o o o t as a major player in the field of nanotparticle-based medical countermeasures. t M M M l l l o o o i i i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c oc c o c o S h S h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a22 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Predgictive Pathogen Biology: Genome-Based Predigction of Pathogenic Potential and g e y e y y e Countermeasures Targets . . . c c c c c c S o S o S o h 180814 | Year 1 of 3 | Principal Investigator: J. S. Schoenihger h e m e m e m nc c n c n r Project Purpose: r r e e e o Pathogens with novel genomes arise from emergence of zooonotic agents, anthropogenic selection, o t t t M M M l or genetic engineering. Understanding molecular mechanisms lof pathogens requires years of work, l o o o i l but arresting an outbreak requires rai lpid assessment of virulence and countermeasure targeitls. i i i gt t g t g a We are grossly unprepared to respond tao an engineered biothreat attack: we cannot cope with a yr naturally emerging drug-resistant species.r Next-generation sequencinyg (NGS) technology enablesr y y y y .T rapid availability of complete genome sequenTces from outbreak-causing p.athogens. Large sets T . ce (N>100) of genome sequences from taxonomicaelly related pathogens of dif c fering virulence are e c oc c o c o S hnow puSblicly available and the cost for sequencing chontinues to fall: to apply this revolutionarSy h mncapability to outbreak response, we urgently need betnter tools to utilize these datma to reliably predict n m e e e o o o pathogenicity, identify pathogen-lethal targets, and predict countermeasure effectiveness. c l o c l o c l o Curregntly, best methods predict correct gene function ~60%g of the time, and ~30% of bacterial g r r r e protein y s have no asesigned function. While improved annotation y is necessary, we propose a more e y . . . t S fundamencta o l shift to at S data-driven comparative approach. We willc im o plement a S n analysis pipeline t c o M M M e wherein gene mand genomee features are harvested from large sets of remlated genoemes and statistical m c and machine learning methcods are used to predict drug resistance and virulence phcenotypes, similar i r ir r i l e to a Genome Wide Associationle Study (GWAS) but with richer feature sets, including hoerizontal gene l i i i t t M transfer descriptors. We will alsott Mdemonstrate cost-effective sequencing and assembly otfM hundreds t a a a of genomes of common pathogen species directly from clinical samples, and apply our pipeline to i i i r l i t predicting observed clinical phenotyp lr ei ts, showing that scaling to ~10,000 genomes will eventu l iatlly be r y a ya a y feasible. r r r T y Ty y T T This project will establish proof of principal anTd software tools for a new approach to predicting T e e e e e e pathogen behavior, and demonstration of scaling for sequencing and analysis of thousands of c c cc c c hpathogen genomes. h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 23 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Recomgbinant Vesicular Stomatitis Virus for Therapeugtic Antibody Epitope Mapping g e y e y y e and Vaccine Development . . . c c c c c c S o S o S o h 170801 | Year 3 of 3 | Principal Investigator: O. Negrete h h e m e m e m nc c n c n r Project Purpose: r r e e e o Identification and characterization of antibody binding sites (eopitopes) are important for the o t t t M M M l development of novel vaccines, therapeutics, diagnostics, and affilnity reagents. Several methods have l o o o i l been established to map epitopes on tariglet antigens, including x-ray co-crystallography of antibioldy- i i i gt t g t g a antigen complexes, array-based peptide scaanning, and ‘shotgun’ mutagenesis mapping. Combinatoraial yr techniques, such as phage display, are less labror- and cost-intensive but arye unable to reliably map r y y y y .T complex 3D conformational epitopes. Rapid andT cost-effective methods of m.apping of neutralizing T . ceantibodies binding sites at high resolution are curreently lacking. This project se c eks to develop a novel e c oc c o c o S htechnologSy based on recombinant vesicular stomatitish viruses (rVSV) that provides rapid informaStion h manbout the amino acid sequences of therapeutic neutralizinng antibody epitopes to aid min vaccine n m e e e o o o development for priority pathogens of national security concern. c l o c l o c l o Many prgiority pathogens classified as potential agents of bioterrgorism by the National Institute of g r r r e Allergy an y d Infectiouse Diseases (NIAID) require handling in high-le y vel biocontainment facilities and a e y . . . t S complex bioscaf o ety infrasttr S ucture (e.g., high containment labs, select acg o ent registra S tion/permits). This t c o M M M e project seeks to cmreate a rapeid capability for the high resolution mappingm of antibodye epitopes for BSL- m c 3 and BSL-4 biothreat pathogecns under BSL-2 containment. We will accomplish this gocal by creating i r ir r i l e libraries of pseudotypical rVSV vlaeriants displaying high-complexity randomly mutagenizeed biothreat l i i i t t M antigens. Once the rVSV libraries artteM developed, we aim to recover detailed information abtoMut the t a a a amino acid sequences of conformational epitopes recognized by multiple monoclonal antibodies or i i i r l i t polyclonal antisera in less than one week lr iutsing minimal amounts of each antibody. l i t r y a ya a y T r y Our high-risk/high-reward approach for rapidr T y and cost effective epitope mapping of antibodies r y T T that recognize biothreat pathogens would have aTn immediate impact in the development of novel T e e e e e e vaccines, therapeutics, diagnostics, and affinity reagents. Our established techniques for creating c c cc c c hrecombinant vesicular stomatitis virus (VSV) are poisehd to enable a revolutionary change in epitope h h h h mnapping of conformational antibodies. n n n o on o n l l l Summoary of Accomplishments: o o o o o g g g During the course of this project, three major milestones were achieved. First, under BSL-2 l y ly y l o containmen.t, we generated a recombinant virus expressing the Eboo.la virus (EBOV) glycoprotein in . o c c c S g preparation foro studies ofS virus entry, antibody response, diagnostic tegosting and thSerapeutic inhibitor o g e screening. This remcombinante construct is based on the current EBOV vacmcine vectore in phase III m y y y c c c clinical trials known as VSV. We also fused the VSV-P gene with a green fluorescent protein (GFP) in r . r . r . e e e corder to easily measure infection using fluorescent microscopy or microtiter pclate reader. We rescued c t t t M othe recombinant VSV-EBOVGP-GFPM virus in tissue culture cells and confirmed thoe presenceM of o S S S i EBOV-GP on the surface of the virions uising an antibody neutralization assay. Our second milesitone l m l m l m i i i e t a was met whene we performed proof-of-contcaept studies that tested antiviral strategies targeting thtea e c r EBOV lipid memcbrane against our BSL-2 Ebolra virus model. Our final milestone was achieved when rc y y y r T we characterized thre receptor usage of Cedar viTrus, a new henipavirus (RNA virus). The ability of Tr e eCedar virus to use mueltiple receptors indicates thee potential for a broader cellular tropism compared ee c c c t to Nipah virus and Hendrta virus. The techniques used to create recombinant VSV reagents and the t h h h M M M dnata obtained from this virus construct will be used in rensponse to future calls on Ebola virus or Nipah n relaoted viruses for studies on countermeasures, antibody roesponse, and/or diagnostic testing. o i l i l li l o l o ol i i i t Significagnce: t g g t a y a y ya NIAID prior.ity pathogens such as Nipah and Hendra viruses are con.sidered potential bioterrorism . c c c r S and agroterror o ism agentsS of national secrurity concern. There are curre o ntly no speScific therapies o r y y y e or licensed vaccinmes for maney of these highly pathogenic biodefense agenmts. This proeject benefits m T c c T c T the biosecurity mission through early R&D of a technology that can define the immunodominant r r r e e structures on viruses, which will feacilitate the deseign and development of improved vaccinees against e t t t c M biodefense pathogens. UnderstandinMg the full range cof viral epitopes targeted by the immunMe system c h i l is vital to designing better vaccines. i l h i l h i i i t t t n a24 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Systegms-Level Synthetic Biology for Advanced Biofguel Production g e y e y y e 170804 | .Year 3 of 3 | Principal Investigator: A. Ruffing . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc The US is currently reliant u c pon foreign fossil fuel resounrces, representing a signific c ant threat to n r r r o e national security. The limited seupply of fossil fuels presentso a significant challenge for feuture energy o t t t M provision, and the accumulation oMf carbon dioxide from fossil fuel combustion may lead toM significant l l l o i changes in the global climate. The deivelopment of a renewable eonergy source, such as biofueils o l l l i i i gt derived from microalgae, will help lead tto energy independence forg the US while also addressintg g a a a yr concerns of fossil fuels supply and climate rchange. However, microalgayl fuel production is currentlry y y y y limited by the low natural productivities of microalgae. Synthetic biology techniques may be applied .T T . T . ce to improve the fuel-producing capacity of thesee microorganisms, yet the imcpact of this strategy is e c oclimited by the one part (or circuit) at a time approacch of traditional synthetic boiology. To reach the fuel c o S h S h S h mnproduction rates necessary to make this process econnomically viable, microalgae mmust be genetically n m e e e omodified at a systems level to optimize the entire genomoe for fuel production. Hence, new methods o c an l do syntheticc biology tools must be developed for these m l iocroalgal hosts. This project will develop c l o g g g r a systems-level srynthetic biology approach for targeted genetic manipulation of microalgae. With r e y e y e y this techn.ique, multiple genetic targets will be modified in a singl.e transformation step, leading . c c c t S to a significaont reductitoSn in the time required for strain developmenot. As an exaSmple, consider a t o M M M e strain requirinmg ten geneteic modifications for improved fuel productiomn: the tradietional approach m c c c would typically require ten months for strain development, assuming a conservative four weeks per i r ir r i l i e genetic modification. The proj l ee icted systems-level approach would construct the samee strain in only l i t t M one month. Additionally, the syst tt eMms-level synthetic biology technique would generate v t aMriants t a a a i with different combinations of the tein targeted modifications. Given the inherent non-lineariity of r l i lri l i r y t a biological systems, these combination vt ya a riants may have unknown synergistic effects, resultingt a in y r increased fuel production and informing orur basic understanding of fundamental microalgal biologry. T y Ty y T T T T e e Summary of Accomplishments: e e e e c cThis project advances the development of a multipccle promoter replacement technique for systems- c c h h h h level optimization of gene expression in a model cyanhobacterial host: Synechococcus sp. PCC 7002. To h n n n n orealize this multiple-target approach, key capabilities weon re developed, including a high-throughput o n deltoection method for advanced biofuels, enhanced transfol ormation efficiency, and genetic tools l o o o o for Sygnechococcus sp. PCC 7002. Nile Red and Bodipy based ngeutral lipid staining methods were g l y ly y l o develope . d for characterization and quantification of free fatty aoc . ids and alkanes, potential biofuel . o c c c S g products. Th o ese staininSg methods may be integrated into high-thro g u o ghput screSening methods for o g e characterizingm modified peroduction strains. For modification of our cymanobacteriael host Synechococcus m y y y c sp. PCC 7002, we identifiedc and characterized neutral integration sites within the hcost genome. r . r . r . e c Parameters used for transformeation of Synechococcus sp. PCC 7002 were alcso optimizeed, including c t t t M o cell concentration and length of hMomology arms for genome integration. Addit o ionally, exoMnuclease o S S S i activity was detected as a potential fiactor for reduced transformation efficiency in Synechocoiccus sp. l m l m l m i i i e t PCC 7002e, and high genome copy numbter was identified as an obstacle in generating mutants itn theis a a a c r strain. To enabcle high-throughput characterrization of genome integration and gene expression, wer c y y y r T characterized flu r orescent protein reporters a T nd native promoters in Synechococcus sp. PCC 7002. Tr e e Three fluorescent perotein reporters were expreessed in Synechococcus sp. PCC 7002 and shown to ee t cbe useful for imaging, tflow cytometry, and quantificcation of gene expression. Twenty-four native c t h h h M n promoters were charactMerized in Synechococcus sp. PC n C 7002 using a fluorescent protein reporter; n M othese promoters will serve as a toolset for engineering goene expression level and regulatory patterns. o i l i l li l o l o ol i i i Signigficance: g g t t t a The pro y ject impacts the DOE’s misasion to address energy secur y ity through the development of ya . . . r transformcative science and technologry. The development of a systcems-level synthetic biology c r S o S o S o y y y e technique for mmicroalgaee will reduce the time required to optimize strmains for thee commercial m T c production of renewable, dcrop-in replacemTent fuels. The technique will likely have ca broader impact T r r r e e on the field of synthetic biologey, advancing otheer areas of national interest such as reneewable e c t M replacements to petroleum-basetdM chemical feeds c tocks. These renewable energy and chet Mmical c applications are key focus areas of the DOE’s Bioenergy Technology Office and APRA-E. h i l i l h i l h i i i t t t n a a n a 25 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T The Enggineering and Understanding of Nanoparticleg/Cellular Interactions g e y e y y e 165609 | Yea.r 3 of 3 | Principal Investigator: C. J. Brinker . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc Stable nanoparticles capable o c f specific binding to target cenlls and delivering high dose c s of n r r r o e therapeutic compounds would bee transformational for treatmoent of disease by making dreug delivery o t t t M into diseased cells more efficient, whMile reducing toxic side effects in healthy cells and tissueMs. We l l l o i recently invented a new composite nanoicarrier termed a ‘protocello.’ Targeted protocells, reportied o l l l i i i gt in a 2011 Nature Materials cover article cotmbine the advantages of FDgA-approved liposomes (lowt g a a a yr inherent toxicity, immunogenicity, long circulration times) and porous partyicle nanocarriers (stability r y y y y .T and an enormous capacity for multiple cargos). A T s recognized in a Nature Mat.erials News & Views T . cecommentary accompanying our report, “The propereties engineered into this [protcocell] system elegantly e c oc synergize to approach the goal of an ideal targeted-deliv c ery agent.” However, it was coautioned that c o S h S h S h mtnherapeutic efficacy established only in vitro (i.e., in cell cnulture) “will need to be testedm in animal models, n m e e e whoere the complexity of events leading up to encounter of a nanooparticle with its target cell may have a o c signifi l coant impactc on the ultimate outcome.” We completely ag l roee with this assessment; accordingly, c l o g g g r the overall purpose rof this project is to rigorously engineer nanoparticle characteristics such as size, r e y e y e y shape, dens.ity, surface charge, and extent of modification with polym.ers and targeting ligands, and . c c c t S determine the oeffects ont sSpecific and non-specific binding, circulation toime, biodisStribution, and t o M M M e toxicity in vivo. Omur use of steate-of-the-art imaging employing an accessimble egg emberyo model has m c c c allowed us uniquely to reveal as yet unrecognized structure-activity relationships needed to achieve i r ir r i l i e efficient targeted delivery in vivo l .e i e l i t t M ttM t M t a a a Summary of Accomplishments: i i i r l i t This project employed a variety of in vitro lri atnd in vivo methods along with modeling and simulatio l ni t r y a ya a y and advanced imaging to explore and quantify nanoparticle/cellular interactions with the goal of r r r T y Ty y T establishing needed SARs. A major focus was to use the ‘protocell’ as a modular platform in order T T T e eto understand, engineer, and perfect NP/cellular ieneteractions. The in vivo behavior, viz. the flow e e c ccharacteristics, specific versus non-specific binding, accnd uptake by the innate immune system, of c c h h h h ENPs was evaluated using an ex ovo chick chorioallantoihc membrane model or CAM. CAM imaging h n n n n enaobled us to engineer the physicochemical characteristicson of ENPs and their in vivo behavior so o n as to lmoinimize non-specific binding to endothelial and innatel iommune cells and, therefore, enhance l o o o o circulatigon time and binding frequency to target cells. Safer-by-dgesign concepts were developed for g l y ly y l o silica and ra . re earth oxide nanoparticles. o. . o c c c S g o S go S o g Significance: e m e m e m y y y c Engineered nanoparticles (ENcPs) capable of specific binding to target cells and delivercing high r e . doses of therapeutic compounds r ecould be transformational for treatment of .disease by m r eaking . c c c t drug delivery into diseased cells motre efficient, while reducing toxic side effects in healthy ctells and M o M o M o S tissues. ESNPs are also of interest for a variety of sensing, diagnostic, and imaging functions that cSould i i i l m l m l m i provide early detection of disease or the oinset of infection and enable monitoring of the course oif e t e t te a a a therapy. Underlying these topics and crucial to the safe use of ENPs is development of fundamental c r c r rc y structure-activity relationships (SARS) in the cyontext of nanoparticle/cellular interactions in vivo. This y r T r T Tr is essential for advancing the burgeoning field of nanoparticle therapeutics, sensors, imaging agents, e e e e ee cand prophylactics for chem/bio defense and other DocD supported areas like traumatic brain injury. c t t t h h h M M M Bneyond fundamental science, this project contributed grneatly to the protocell intellectual property n o o o portfolio where, based on successful responses to office actions, we were issued a second i l i l li l comproehensive US patent on targleted protocells. Additionallyo, protocell technology was successfully ol i i i g g g t transferred to the biotechnology indtustry. Aspects of this work also informed a newly funded Grand t a y a y ya Challenge p.roject on developing nano-therapeutic platforms for the. delivery of clustered regularly . c c c r r r S interspaced shoort palindroSmic repeats (CRISPR) components. o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a26 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Refergeed Communications: g g e C.J. Brinyker, “Silica @ Cells: A Special Biotic/Abiotic e Interface,” Plenaryy Lecture at the Third International y e . . . c Conference con Advanced Complex Inorganic Nanomateriacls (ACIN), Namur, cBelgium, 2015. c c S o S o S o h h h e J.L. Townson, Y-mS Lin, P. Durefee, C.E. Ashley, and C.J. Brinker, “Protocells: Mesmoporous Silieca Nanoparticle m nc Supported Lipid Bilayers for Tacrgeted Delivery,” presented at nthe Fourth International Confercence on n r r r e Multifunctional Hybrid and Nanomeaterials, Sitges/Barcelona, Spain, 2015. e o o o t t t l M J. Sun, E. Jakobsson, Y. Wang, and C.JM. Brinker, “ Nanoporous Silica-Balsed Protocells at Multiple ScaMles for l o o o i Designs of Life and Nanomedicine,” Life, ivol. 5, pp. 214-229, January 2015. i l l l i i i gt t g t g a Y-R Lou, L. Kanninen, B.J. Kaehr, J.L. Townsona, J. Niklander, R. Harjumaki, C.J. Brinker, and M. Yliperttula, “aSilica yr y Bioreplication Preserves Three-Dimensional S r pyheroid Structures of Human Pyluripotent Stem Cells and Hep r Gy2 y .T Cells,” Scientific Reports, vol. 5, Article 13635, SeptTember 2015. . T . ce e c e c ocK. Butler, P. Durfee, C. Theron, C.E. Ashley, E.C. Carnes,c and C.J. Brinker, “Protocells: M o odular Mesoporous Silica c o S hNanopa S rticle-Supported Lipid Bilayers for Drug Deliveryh,” to be published in Small. S h mn n m n m e oP.E. Johnsone, P. Muttil, D. MacKenzie, E.C. Carnes, J. Pelowitz,o N.A. Mara, W.M. Mook, S.D. Jett, D.R. Dunphye, G.S. o c Timl omins, and Cc.J. Brinker, “ Spray-Dried Multiscale Nano-Biocolmoposites Containing Living Cells,” ACS Nano, volc. l o 9, pp.g 6961-6977, June 2015. g g r r r e y e y e y B. Sun, S. P.okhrel, D.R. Dunphy, H. Zhang, Z. Ji, X. Wang, M. Wang, Y-P L.iao, C.H. Chang, J. Dong, R. Li, L. Mädler, . c c c t S C.J. Brinker, Ao.E. Nel, andtS T. Xia, “ Reduction of Acute Inflammatory Effectso of Fumed SiSlica Nanoparticles in t o M M M e the Lung by Adjmusting Silanoel Display through Calcination and Metal Doping,m” ACS Nano, veol. 9, pp. 9357-9372, m c September 2015. c c i r ir r i l e le e l i A.S. Dobroff, R. Rangel, L. Guzman-iRojas, C.C. Salmeron, J.G. Gelovani, R.L. Sidman, C. Bologa, T. Oprea, C.J. i t a t M Brinker, R. Pasqualini, and W. Arap, “ tt LMiagand-Directed Profiling of Organelles with Internalizing Ph t aMge Libraries,” t a i Current Protocol in Protein Science, vol. 30i, 1-30.34, February 2015. i r l i lri l i r t t t y a J.C. Harper, B.D. Carson, G.D. Bachand, W.Dy.a Arndt, M.R. Finley, C.J. Brinker, and T.L. Edwards, “Laser Macahined y r r r T y Plastic Laminates: Towards Portable DiagnostiTcy Devices for Use in Low Resource Environments,” Electroanalyysis, T T vol. 7, pp. 2503-2512, November 2015. T T e e e e e e c cD.R. Dunphy, P.H. Sheth, F.L. Garcia, and C.J. Brinker, “E ccnlarged Pore Size in Mesoporous Silica Films Templated c c hby Pluronic F127: Use of Poloxamer Mixtures and Increahsed Template/SiO2 Ratios in Materials Synthesized by h h h h n n n Evaporation-Induced Self-Assembly,” Chemistry of Materials, vol. 27, pp. 75-84, December 2014. n o on o n S.Sl.o Chou, Y-K Huang, J. Kim, B. Kaehr, B.M. Foley, P. Lu, C. Dykstlroa, P.E. Hopkins, C.J. Brinker, J. Huang, and l o o o o V.P. Dgravid, “Controlling the Metal to Semiconductor Transition of MgoS2 and WS2 in Solution,” Journal of the g l Americayn Chemical Society, vol. 137, pp. 1742-1745. January 2015. ly y l o . o. . o c c c S g Y. Fu, B. Li, Y-oB Jiang, D.RS. Dunphy, A. Tsai, S-Y Tam, H. Fan, H. Zhang, D.M. Rgoogers, S.B. RSempe, P. Atanassov, o g e J.L. Cecchi, andm C.J. Brinker,e “Atomic Layer Deposition of L-Alanine Polypeptimde,” Journal oef American Chemical m y y y c Society, vol. 136, pp. 15821-15c824, October 2014. c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 27 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report ECNEICSOIB S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Undersgtanding and Engineering Lignolysis for Renewgable Chemical Production g e y e y y e 173019 | Yea.r 2 of 3 | Principal Investigator: S. Singh . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc The purpose of this project is t c o build upon the Sandia expenrtise in lignin characterizat c ion and n r r r o e conversion to develop a microbiael chassis capable of convertinog lignin generated by the beiofuel, pulp, o t t t M and paper industries into targeted reMnewable chemicals. M l l l o o o i i i l Lignin is the only source of renewable arlomatics, and utilization of lignin for high value co-produlcts i i i gt t g t g a will enable biofuel industries to become cosat competitive with petrochemicals. The prospects of a yr r y r y y macromolecular lignin being utilized as raw mayterials for fuels and chemicals have been highly touted, y .T but no conversion technologies have been realizTed. Towards this goal, this pr.oject explores two T . ce e c e c routes: 1) a bio-route based on an engineered organism by borrowing the heterologous enzymes for oc c o c o S hlignin depSolymerization, transportation of the lignin mhonomers and subsequent modification intoS h mn n m n m value-added chemicals and 2) a hybrid process—a chemical process followed by a bio-based process e e e o o o that utilizes the engineered organism from the first route. c l o c l o c l o r Bacillus gsubtilis was crhosen as a suitable bacterial host for this prgoject as it possesses the ability to r g e secrete th y e lignolytic eenzymes along with other advantages. By pe y rforming omics studies on an array e y . . . c c c t S of different mic o robes andtS substrates, we are aiming to gain knowledge o of differenSt transporters t o M M M e involved in ligninm uptake, thee central metabolism of the lignolytic organismm and identeify novel m c c c enzymes involved in the lignin pathway. We are also evaluating two aromatic transporters for their i r ir r i l e capability in uptake of a range of loeligomeric compounds derived from lignin. Our final proeduct will be l i i i t t M a robust synthetic biology toolbox t ttaMilored for lignin coupled with an engineered microbial c t hMassis and t a a a i conversion routes for lignin conversion.i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a28 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report BIOSCIENCE S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Unkgnown Pathogen Detection in Clinical Samples:g A Novel Hyperspectral Imaging g e y e y y e and Single Cell Sequencing Approach . . . c c c c c c S o S o S o h 165607 | Year 3 of 3 | Principal Investigator: B. Carson h h e m e m e m nc c n c n r Project Purpose: r r e e e o Detection of pathogens in pre-symptomatic stages of infectoion is challenging. Most diagnostic o t t t M M M l techniques are insufficiently sensitive to detect infection and idlentify viruses in particular at l o o o i l these early stages. Recently, second-igleneration sequencing has been applied to this problemi.l i i i gt t g t g a However, despite substantial progress, tahe signal-to-noise ratio inherent in this technique makeas yr pathogen identification in clinical samples rprohibitive, thus leaving treaytment decisions dubious. r y y y y .T We address this critical problem by identifyinTg spatial and spectral signatu.res of host response that T . ce differentiate uninfected human blood cells frome those infected with a patho c genic virus without e c oc c o c o S hrequirSing a priori knowledge of its identity. We will ihntegrate hyperspectral imaging with microSfluidic h mncell sorting to: 1) separate individual infected and uninnfected cells and 2) use highm-throughput n m e e e o o o sequencing to analyze individual cells and pools of cells. We envision this will greatly improve the c l o c l o c l o ability to identify pathogens in clinical samples since pathogen-negative cells from the same sample g g g r r r e will seryve as contro e l. Only single cell analysis can verify the accyuracy of this technique, and our e y success w.ill greatly improve our competitiveness for the Nationa.l Institutes of Health and other . c c c t M S potential spoonsors. Mo tS rMeover, this new capability would transform noot only the S ability to diagnose t M o e m e m e m infection with unknown pathogens early in disease, but also the identification, characterization, c c c i r and countermeasure developirment for emerging and engineered viruses. Finally, the hryperspectral i l e le e l i imaging cell sorter we are develioping will have countless applications in diverse biological problems i t t M ttM t M t a including infectious disease, biofueals, and analysis of microbial consortia. a i i i r l i lri l i r y t a Summary of Accomplishments: t ya t a y T r y We have designed and tested a laminate mr Ti y crofluidic device capable of axially focusing a stream ofr y T T cells. A rapid-actuation solenoid valve provideTs a short burst of pressure to divert a target cell into anT e e e e adjacent flow stream bound for a sample vial. Wee have developed image analysis software to select e c c cc c c cells that meet user-specified criteria (e.g., color, size, intensity) and have integrated this software h h h h nwith the solenoid valve control software to allow forh anutomated sorting. n h n o on o n Wle have designed and constructed a prototype of a novell hyperspectral imaging-based cell sorter. l o o o o o o Usingg this instrument, we have demonstrated collection of hygperspectral confocal fluorescence g l imagesy of individual flowing beads/cells. An invention disclosurlye has been filed for this technology. y l o . o. . o c c c S g We have ideontified speSctral and spatial biomarkers indicative of viragol infection Sin macrophage cells o g e stained with tmhe vital dyee Acridine Orange. The biomarkers are robustm and permite classification of m y y y c c c infection state as early as six hours post-infection with Adenovirus. Identification and classification r . r . r . e e e c are enabled by the development of image analysis algorithms to identify celcls and provide cell c t t t M ostatistics and by logistic regressioMn classification methods. o M o S S S i i i l i mWe have performed successful sequelnicing of as few as 10-100 cells with detectiomn of virus. Wl ie m e t e t te a a a accomplished single cell sequencing, but this proved to be less robust and sensitive to contamination. c r c r rc y y y r T Significance: r T Tr e e e e ee The project has provided initial demonstration of a new method for identification and isolation of c c c t hvirus-infected cells frotm their uninfected counterpahrts without the use of virus-specific reagents. This h t M M M nmethod benefits the DOE/NNSA and DHS biodefensen missions by facilitating groundbreaking studies n o o o i on l the identification of man-imade and emerging pathogen l s. Further, this capability can benefit DOE/ li l o l o ol i NNSA bioenergy programs sincie it enables the screening of genetically engineered algal strains and i g g g t t t a microbyial communities for enhancaed biofuels production. y ya . . . c c c r r r S Refereed CoommunicatSions: o S o y y y e S.M. Anthony, Am. Carroll-Poretillo, and J.L. Timlin, “Dynamics and Interactions mof Individuale Proteins in the m T c c T c T Membrane of Living Cells,” Single Cell Protein Analysis, A.K. Singh, editor, Springer, 2015. r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 29 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T g g g e y e y y e . . . c c c c c c S o S o S o h h h e m e m e m nc c n c n r r r e e e o o o t t t M M M l l l o o o i i i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c Computing & Information Sciences oc c o c o S h S h S h mn n m n m e e e Theo Computing and Information Sciences (CIS) LDRD Invesotment Area (IA) champions innovative o c resea l roch and devcelopment that advances the state of the art l ion mathematics, information sciences, c l o g g g r and computing relevrant to Sandia’s national security missions. As these applications continue to r e technically y deepen aned broaden (internally and externally), there i y s an ongoing need to refresh e y . . . c c c t S and advance un o derlying ktSnowledge and capabilities. The scope of the I o A includes Scomputer and t o M e computational sc m ience andM eengineering, information science (including m m athematicse and cognitive M m c sciences), and aspects of cyberc sciences and data sciences. The IA is pursuing several sctrategic i l r e objectives. First, we want to dev irele lop the technologies that will underpin the success of fru e ture i l i t t M generations of Sandia computers. Ti th tM ese technologies include scalable system architecturest, M system i t a software and computational methodsa, and tools for both scientific and data-centric computing. They a i i i r l i also include approaches to building futurlr e i machines out of non-CMOS technologies. Second, wel w i ant r t t t y a to develop and exploit synergy between coymaputing and cyber capabilities. Third, we want to advanace y T r y the science and engineering of trust and resilir Te y nce in computing and information systems. r y T T T T e e e e e e c cProjects cc c c h h h h h h n n n n (Acotive) Learning on Groups of Data with Information-Theoretic on Estimators ....................................................................32 o n l l l o o o o A Domain-Specific Language for Distributed Tensor Computations ...o...................................................................................33 o g g g l y ly y l o A Framework . for Wind Turbine Design under Uncertainty ...............................o.. . .......................................................................34 . o c c c S g Active Learning oin the Era of S Big Data ................................................................................go......................... S ........................................35 o g e m e m e m y y y c Adaptive Bayesian Inference for Pcrediction .......................................................................................................c..............................36 r . r . r . e e e c c c t Adaptive Multimodel Simulation Infrasttructure (AMSI) ............................................................................................t..................37 M o M o M o S S S i Advanced Data Structures for Improved Cyiber Resilience and Awareness in Untrusted Environments ......i.........38 l m l m l m i i i e t e t te a Advanced Uncertainty Quantification Methodsa for Circuit Simulation ...............................................................................a.39 c r c r rc y y y Analyst-to-Analyst Variability in Simulation-Based Prediction ................................................................................................40 r T r T Tr e e e e ee APEX: Application Characterization for Exascale Systems ........................................................................................................41 c c c t t t h h h M Cnognitive Computing for SecuMrity .........................................................n................................................................................................42 n M o o o i Couplling Computational Models: Firom Art to Science ..........................l......................................................................................44 li l o l o ol i i i g g g t Data-Driven Optimization for the Desigtn and Control of Large-Scale Systems ................................................................45 t a y a y ya . . . Decision Analyctics for Complex Supply Chain Networks .......................................c.....................................................................46 c r r r S o S o S o y y y e Detecting Lateral Mmovement oen Internal Networks..........................................................m.........................e...................................48 m T c c T c T r r r e e Efficient Probability of Failure Calcuelations for QMU u e sing Computational Geometry..........................e.......................49 e t t t c M Enabling Bidirectional Modality TransitiMons in Collaboraticve Virtual Environments .......................................M................51 c h i l i l h i l h i i i t t t n a30 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Gameg Theory for Proactive Dynamic Defense and Attack Mitigatiogn in Cyber-Physical Systems ............................52 g e y e y y e Graph Lea.rning in Knowledge Bases .....................................................................................................................................................53 . c c c c c c S o S o S o h HostWatch: Situational Awareness of Machine State for Chybersecurity .............................................................................54 h e m e m e m nc c n c n r Identification of Markers of Highr RANS Uncertainty for Model Improvement in Engineering Frlows ......................56 e e e o o o t M In Situ Compressed Sampling and RteMconstruction of Exascale Unstructured Mesh Datasets ...........t..M.......................57 l l l o o o i l Kernel and Meshless Methods for Partiaill Differential Equations ............................................................................i..l..............58 i i i gt t g t g a a a Modeling Information Multiplexing in the Hippocampus ............................................................................................................59 yr r y r y y y y .T Operationally Relevant Cyber Situational AwarenTess Tool Development ............................................................................6T0 . ce e c e c ocPIMS: Memristor-Based Processing-in-Memory-and-Sctorage .................................................o...............................................62 c o S h S h S h mnReducing Computation and Communication in Scientific Conmputing: Connecting Theory t m o Practice ...................63 n m e e e o o o c Sclaoling up Semciconductor Quantum Computers through Multislcoale Analysis ..................................................................6c4 l o g g g r r r e Simulatyion Capabilit e y and Computational Assessment of Memristorsy as Beyond-CMOS Logic and Memory e y Dev.icces ........................................................................................................................c...........................................................................65 . c t S o tS o S t o M M M e Strong Local-Nmonlocal Coupeling for Integrated Fracture Modeling ......................m.........................e.........................................66 m c c c i r Sublinear Algorithms for In Situi rand In-Transit Data Analysis at Exascale ...........................................r................................68 i l e le e l i i i t t M Topological Design Optimization of ttCMonvolutes in Next-Generation Pulsed Power Devices ............. t ..M.......................70 t a a a i i i r l i Towards Rigorous Multiphysics Shock-Hylrdiro Capabilities for Predictive Computational Analysis ...........l..i............71 r t t t y a ya a y r User-Accessible Unified Manycore Performanrce-Portable Programming Model .........................................................r...72 T y Ty y T T T T e Versatile Formal Methods Applied to Quantum Infoermation ....................................................................................................73 e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 31 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T (Activeg) Learning on Groups of Data with Informationg-Theoretic Estimators g e y e y y e 181061 | Yea.r 1 of 3 | Principal Investigator: W. P. Kegelmeyer . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc Machine learning techniques i c n areas such as classificationn, anomaly detection, and re c gression n r r r o e typically focus on single points ase the unit of interest; the undeorlying data is projected intoe a feature o t t t M space before the algorithms ever seeM it. This projection, however, is not always natural, and itM may l l l o i discard key information. In many settingis, we wish to perform machoine learning tasks on objectsi that o l l l i i i gt best can be viewed as a set of lower-level otbjects: for example, imagesg can be successfully viewedt as a g a a a yr collection of local patches. In order to use trarditional techniques on thesey set-based objects, we mustr y y y y create a single vector that represents the entire set. Though there are various ways to summarize .T T . T . cea set as a vector, we can discard less information aned potentially require less wcork in feature e c ocengineering by operating directly on sets of feature vcectors. o c o S h S h S h mn n m n m One method for machine learning on sets is to consider them as samples from some unknown e e e o o o underlying probability distribution over feature vectors. Each example has its own distribution: if we c l o c l o c l o are classifying images as sets of patches, each image is defined as a distribution over patch features; g g g r r r e each classy of images is e a set of patch-level feature distributions. Wye can then explore kernel methods, e y . . . based on statcistical estimates of distances between probability districbutions, to classify samples from c t S o tS o S t o M these distributions. BecausMe we obtain a kernel, we can exploit the substantial literature of kernel M e m e m e m c methods to solve many types ocf learning problems. This research, in collaboration Carcnegie Mellon i r University, is applicable wherevierr a deeper analysis of sets of things (rather than individrual samples) i l e le e l i i i t t M is required. ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a32 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T A Dogmain-Specific Language for Distributed Tensogr Computations g e y e y y e 181062 | .Year 1 of 1 | Principal Investigator: T. G. Kolda . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc The growing complexity of c mathematical modeling requnires increasingly complex d c ata structures to n r r r o e be partitioned onto massively eparallel-distributed computinog systems. We are particulearly interested o t t t M in the distribution of multi-way arMrays, also known as tensors. A matrix is an example of a Mtwo-way l l l o i array, and much previous work has foicused on optimal partitionoing schemes to reduce the coist of o l l l i i i gt collective communications (i.e., those otperations that involve data egxchange between groups oft g a a a yr compute nodes). We consider the problemr for n-way arrays, which greyatly increases the complexitry of y y y y the problem, but is necessary for facilitating advanced simulations in domains such as computational .T T . T . ce chemistry, advanced data analysis, and other apeplications. Each communicaction pattern has e c ocsome structure underlying it. Our goal is to providce analysis capabilities for opotimizing collective c o S h S h S h mncommunications. n m n m e e e o o o The goal of this project, in collaboration with University of Texas-Austin, is to create a theory for c l o c l o c l o multilinear computation in distributed-memory environments, which formalizes data distributions, g g g r r r e collectiyve commun e ications, and the effect of collective communyications, while still being powerful e y . . . enough toc derive high-performance algorithms for multilinear opecrations. By creating a library c t S o tS o S t o M that implements the theMory, tests can be performed to validate or refute the theory. This project M e m e m e m c investigates new ways to excpress distributed objects for n-way arrays. c i r ir r i l e le e l i Summary of Accomplishmentsi: i t t M ttM t M t a Large-scale datasets in computatioanal chemistry typically require distributed-memory parallel a r i l i methods to perform a special operatiiolri n known as tensor contraction. Tensors are multidimeinls i ional r t t t y a arrays, and a tensor contraction is akin tyao matrix multiplication with special types of permutatioans. y T r y Creating an efficient algorithm and optimirzTe y d implementation in this domain is complex, tedious, r y T T and error-prone. To address this, we have devTeloped a notation to express data distributions so T e e e e e e that we can apply automated methods to find optimized implementations for tensor contractions. c c cc c c hBy considering the spin-adapted coupled cluster sinhgles and doubles method from computational h h h h nchemistry, we have used our methodology to producen an efficient implementation. Experiments n n o performed on the IBM Blue Gene/Q and Cray XC30 dem ononstrate both improved performance and o n l l l o o o o reduced memory consumption. o o g g g l y ly y l o Significan.ce: o. . o c c c S g Research suopporting mSultiple DOE, DoD, and other government migsosions oftenS relies on o g e mathematicalm modeling aend advanced simulation, computationally commplex operaetions. Partitioning m y y y c c c and executing such processes onto massively parallel distributed computing systems is nontrivial. The r . r . r . e e e c success of our results could potentially numerous disciplines reliant on highc performance computing. c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 33 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T A Framgework for Wind Turbine Design under Uncertaginty g e y e y y e 173867 | Yea.r 2 of 3 | Principal Investigator: M. S. Eldred . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc Wind energy is an increasingly c vital component of the electnricity generation system in c the United n r r r o e States. In 2009, over 3% of the naetion’s total electricity generaotion came from wind, and ae report by o t t t M DOE outlines plans to satisfy 20% ofM the national energy budget with wind power by the yeaMr 2030. l l l o i DOE estimates that as much as a 35% inicrease in annual wind enerogy production can be obtaineid o l l l i i i gt from improvements in wind turbine aerodtynamics. Achieving these gagins in performance will onlyt be g a a a yr possible by using higher fidelity methods to mrodel the aerodynamics. Theyrefore, a new framework forr y y y y wind turbine design under uncertainty, that makes effective use of high-fidelity computational fluid .T T . T . cedynamics, tools will be developed in collaboration weith Stanford University. Thce ability to accurately e c ocassess the effect of all sources of uncertainty using pcredictive high-fidelity modelos will enable c o S h S h S h mbnusinesses to effectively estimate the investment risk in nwind energy solutions and pmromote growth n m e e e in tohis energy sector. o o c l o c l o c l o Currently, wind turbines are analyzed and designed by low-fidelity tools since they are g g g r r r e computatiyonally inexp e ensive, allowing direct evaluation of quantityies of interest (QoI) involving power e y . . . output and mcaximum loading for a broad range of input cases. For necwer, more aerodynamically c t S o tS o S t o M aggressive designs, there isM little hope of obtaining accurate QoI ranges using the simple low-fidelity M e m e m e m c models. But, by effectively augcmenting a large number of low-fidelity simulations withc a smaller i r number of high-fidelity evaluatiiorns within our proposed multifidelity framework, we wilrl accurately i l e le e l i i i t t M and efficiently quantify the uncertatt in M ties in our QoIs. t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a34 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Activge Learning in the Era of Big Data g g e y e y y e 173667 | .Year 2 of 2 | Principal Investigator: W. L. Davis, IV . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc As the volume of data grow c s, so does the amount of comnputational power and the h c uman ingenuity n r r r o e required to harness that poweer. Processing the data is not tohe problem. The critical isseue is that, o t t t M while the amount of data is growiMng, the time it takes a human to answer a question todayM is the same l l l o i amount of time it will take the humani to answer it tomorrow. Anod tomorrow, it will cost morei. While o l l l i i i gt many tasks have been automated, sucht as telephone operators, somge tasks will always prefer tot have g a a a yr a human in the loop. Such tasks may includre sensitive, high-risk applicaytions in medical or security r y y y y areas; they also include areas that require more qualitative assessments of quality: the “I’ll know it .T T . T . ce when I see it” quality that is difficult to automatee. c e c oc c o c o S hCurrenSt active learning research generally neglectsh the valuable information that can be gaineSd h mn n m n m from studying the relationship between the specific machine learning algorithms employed and e e e o o o the active learning heuristics themselves. This project aims to tackle the two major impediments c l o c l o c l o to implementing active learning for big-data in practice: 1) the logistics of query distribution and g g g r r r e collectiyon and 2) th e e lack of efficient algorithms with guaranteeys. This research, in collaboration with e y . . . the Univercsity of Wisconsin-Madison, will both extend the scalabiclity of active learning approaches c t S o tS o S t o M and expand the use of aMctive learning techniques by lowering the barrier of complexity. In addition M e m e m e m c to these practical advancemcents, this research will advance the science of machinec learning by i r enhancing the theoretical unidrerstanding of active learning constraints and guaranteers. i l e le e l i i i t t M ttM t M t a Summary of Accomplishments: a a r i l i To facilitate the development, testingi l, r i and deployment of active learning for real applicationsi l, i we have r t t t y a built an open-source software system foyar large-scale active learning research and experimentataion. y T r y The system, NEXT, provides a unique platfroTy rm for real-world, reproducible active learning researcrh y . T T NEXT enables machine learning researchers tTo easily deploy and test new active learning algorithms T e e e e e e and enables applied researchers to employ active learning methods for real-world applications. c c cc c c h h h h NEXT is accessible through a REST Web API (applicahtion programming interface) and can be easily h n n n n odeployed in the cloud with minimal knowledge and expeon rtise using automated scripts. NEXT provides o n relsoearchers a set of example templates and widgets that claon be used as graphical user interfaces to l o o o o collecgt data from participants. g g l y ly y l o In additio.n to the NEXT software itself, this research produced r o e.sults that show how . o c c c S g experimentaotion can hSelp expose strengths and weaknesses of activgoe learning Salgorithms, in o g e m e m e m y sometimes unexpected and enlightening ways. y y c c c r . r . r . e e e c Significance: c c t t t M oCollecting large-scale datasets foMr building analytical models for national secuority can be Mdifficult o S S S i and time-consuming. NEXT providesi a flexible and general-purpose active learning platformi that l m l m l m i i i e t is versatilee enough for a large breadth otf applications. This allows us to easily replicate past (antd e a a a c r future) algoritchm implementations, experirments, and applications. This research enables analyst-rc y y y r T driven data collerction in a much more efficienTt manner, reducing overall costs for data procurement Tr e e and increasing the efidelity of our analytical modeels. This ultimately enhances our ability to utilize ee c c c t analytical models in arteas such as cybersecurity and satellite data analysis. t h h h M M M n n n oRefereed Communications: o o i l K. lJoamieson, S. Katariya, A. Desh i plande, and R. Nowak, “Sparse Dl oueling Bandits,” presented at the Artificial Intelli- l o i l i i i t genceg and Statistics Conference, San Dtiego, CA, 2015. g g t a y a y ya K. Jamieso.n, L. Jain, C. Fernandez, N. Glattard, and R. Nowak, “NEXT: A. System for Real-World Development, . c c c r r r S Evaluation, aond ApplicatiSon of Active Learning,” Neural Information Processiong Systems, MSontreal, Canada, 2015. o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 35 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Adaptigve Bayesian Inference for Prediction g g e y e y y e 188256 | Yea.r 1 of 1 | Principal Investigator: T. M. Wildey . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc Performing rigorous predictive c simulation necessitates thant any and all available data c be incorporated n r r r o e into the modeling process in ordeer accurately quantify uncertaointies in the model predicteion. o t t t M Statistical Bayesian inference is the mMost frequently used approach for incorporating data inMto l l l o i uncertainty estimates. Statistical Bayesiian inference uses data ando an assumed error model to iinform o l l l i i i gt posterior distributions of model input varitables and model discrepanciges. The resulting posterior t g a a a yr densities are then propagated forward througrh the model to quantify they effect of these uncertaintiers y y y y on the model quantities of interest. Markov Chain Monte Carlo (MCMC) sampling (or the special case .T T . T . ceof Gibbs sampling) is the most popular approach foer solving the Bayesian invercse problem. However, e c ocMCMC often requires an infeasible number of modecl evaluations and can fail to coonverge to the true c o S h S h S h mpnosterior distribution. n m n m e e e o o o We propose to pursue a new approach for stochastic inversion that: c l o c l o c l o g g g r • Has a solid matrhematical foundation r e y e y e y • Does no.tc rely on MCMC sampling to generate samples of the po. csterior distribution . c t S o tS o S t o M • Does not require a staMtistical assumption on the model error M e m e m e m c • Leverages the adaptive recsponse surface approximations techniques developed act Sandia i r ir r i l e le e l i t t M Our approach will utilize a reformuiltt a M tion of a measure-theoretic inversion technique to solvt M e a non- i t a statistical Bayesian inference problema. This reformulation allows us to reduce the computational cost a i i i r l i of approximating the posterior distributilroin using goal-oriented adaptive sampling and approximlaition r t t t y a methods. Our algorithm will concentrate hyiagh-fidelity model evaluations in regions of the parametear y r r r T y space that significantly contribute to the unceTrytainty in specified quantities of interest and are most y T T T T e informed by the available data. e e e e e c c cc c c hSummary of Accomplishments: h h h h h Wne compared our new non-statistical Bayesian approachn with the standard statistical Bayesian n n app o roach and demonstrated that the new approach can ea o snily generate comparable sets of samples o n l l l o o o o from the posterior distribution without relying on MCMC. Weo also produced numerical results that o g g g l demonstrayte adaptive refinement of a sparse-grid surrogate modleyl based on both the prior and the y l o posterior di.stributions. Our method demonstrated improvements i o n. the computational cost required . o c c c S g to achieve a givoen error to S lerance. go S o g e m e m e m y y y c c c Significance: r . r . r . e e e cInverse problems are ubiquitous in computational science and engineering. Ofcten, parameters of c t t t M ointerest cannot be measured directlyM and must be inferred from observable data.o The approaMch o S S S i developed in this project will assist engiineers working to understand and quantify the uncertainities l m l m l m i i i e t a for a broad raenge of national security applticaations throughout the NNSA and other agencies that t a e c r require efficient cpredictive simulation with qurantified uncertainty. rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a36 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Adapgtive Multimodel Simulation Infrastructure (AMgSI) g e y e y y e 166140 | .Year 3 of 3 | Principal Investigator: M. W. Glass . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc At present, the implementa c tion of a massively parallel mnultiscale/multiphysics simu c lation requires n r r r o e expertise, not only in the physiecal domain of interest, but aloso in parallel programming eand software o t t t M engineering. While there have beeMn attempts to construct software frameworks to ease tMhe l l l o i construction of such simulations, theise typically require that comoponent codes adhere to ceritain o l l l i i i gt specific interfaces and data structures. tIncorporation of software cogmponents used in these t g a a a yr simulations can require a great deal of exprertise and effort. Through a ypartnership with Rensselaerr y y y y Polytechnic Institute, this fundamental research is focused on developing multiphysics code-coupling .T T . T . ce frameworks for legacy components. We will deveelop a simulation infrastruccture to facilitate the e c ocimplementation of multimodel adaptive simulatiocns, allowing the incorporatioon of proven legacy c o S h S h S h mncomponents. This infrastructure will leverage high-levnel programming techniquesm and a variation n m e e e oon the typical component-based architectures to facilitaote the easy integration of legacy software o c co l moponents fcor use in the infrastructure. By removing the l orequirement of low-level programming c l o g g g r expertise in orderr to construct these simulations, domain-specific experts and industry-level users r e y e y e y will be ab.le to reuse proven legacy software components in orde.r to simulate the physical phenomena . c c c t S of importanoce to themt.S o S t o M M M e m e m e m c The design and implementaction of a system to facilitate the combination of low-levcel code i r components with arbitrary fuirnctional interfaces and data structures poses a numberr of mathematical, i l e le e l i i i t t M logical, and computational challett n M ges, especially in conjunction with the usability goals stt M ated above. t a The primary foci in the constructioan of physical simulations are efficiency, performance, and accuracy. a i i i r l i Thus, the low-level systems underlyinlrgi the high-level computation stages pose an entirely diflfierent r t t t y a set of challenges to be overcome in thei y ra implementation. a y r r r T y Ty y T T Summary of Accomplishments: T T e e e e We initially focused on designing a set of simulaetion-state abstractions and implementing mechanisms e c c cc c c to model and track them over the life of a simulation. The development of these concepts allowed h h h h h h nthe initial implementation of the adaptive multiscale snimulation infrastructure (AMSI). The primary n n ocapabilities of AMSI focused on maintaining the simulatonion meta-state, as well as planning, managing, o n l l l o ando implementing the parallel communication required foro omultiscale data coupling to occur in a non- o o g g g intrusive manner. Non-intrusive use of AMSI was a key area of focus, to allow minimal modification to l y ly y l o existing s.ingle-scale codes for use in general multiscale simulatioo.ns facilitated by AMSI. . o c c c S g o S go S o g We demonstrated the utility of the initial implementation of AMSI, and applied the method to a e m e m e m y y y c test biological tissue problecm, coupling together a macroscale finite element code acnd a microscale r e . quasistatics code. These resu r ltes were discussed, along with core data stru.ctures and r alegorithms used . c c c t in the implementation of AMSI, itn a talk at the SIAM Conference on Parallel Processing ftor Scientific M o M o M o S CompuSting 2014. We also demonstrated the use of new capabilities developed for dynamicallSy i i i l m l m l m i controlling adaptive multiscale simulatiions: specifically, for scale-sensitive load balancing opeirations. e t e t te a a a These results were presented at the SIAM Conference on Computational Science and Engineering c r c r rc y 2015 and included additional discussion ony the biological tissue test application. y r T r T Tr e e Significance: e e ee c c c t hThis project saw the imtplementation of the scale-sehnsitive load balancing capabilities of AMSI, as well h t M M M nas their inclusion in the biological tissue multiscale simnulation. Further exploration of the capabilities n o o o of the scale-sensitive load balancing and the optimal methods of use in general multi-scale simulations i l i l li l o l o ol was then conducted. Initial work is being conducted on ‘load-balancing’ the execution of individual i i i g g g t t t a scales iyn a multiscale simulation so a as to further optimize the tiyme-to-solution and facilitate greater ya dynamic .control over the simulation. . . c c c r r r S o S o S o y y y e Refereed Commmunicatioens: m e m T c W.R Tobin, D. Fovargue, V.W.Lc. Chan, and M.S. S T hephard, “Load Balancing Multiscale Simulactions,” presented at T r r r e e the SIAM Conference on Computatieonal Science and Eengineering, Salt Lake City, UT, 2015. e e t t t c M W.R. Tobin, D. Fovargue, and M.S. SheMphard, “Parallel Incfrastructure for Multiscale Simulation,” preMsented at the c h i l SIAM Conference on Parallel Processing fori S l cientific Computihng, Portland, OR, 2014. i l h i i i t t t n a a n a 37 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Advancged Data Structures for Improved Cyber Resiliegnce and Awareness in g e y e y y e Untrusted Environments . . . c c c c c c S o S o S o h 180820 | Year 1 of 3 | Principal Investigator: C. A. Phillips h h e m e m e m nc c n c n r Project Purpose: r r e e e o This project focuses on fundamental research in write-optimizoed data structures (WODS) with o t t t M M M l applications to resilient high performance data management. WOlDS enable systems to ingest l o o o i l and index data faster, by orders of magniiltude, than with previous data structures. There are i l i i i gt t g t g a write-optimized data structure replacemenats for B-trees (for databases) and for Bloom filters (for a yr approximate membership queries). Developerd for commercial databases,y WODS have not been r y y y y .T applied in security settings to their full potentialT. . T . ce e c e c ocWe will create WODS to enable network-security-mconitoring components to stroeam high-speed c o S h S h S h network data to disk and perform complex queries on that data during subsequent analyses. The mn n m n m e addoitional speeed of WODS can create new opportunities foor cybersecurity monitoring tools to e o c supplo o rt complexc and real-time queries. We will add featuresl t o o WODS to make them more useful for c l o cyberstrgeam monitoring. These include aging out data and handgling significant data repetition. We g r r r e will quant y ify performaence improvements on realistic cyberstream y s. e y . . . c c c t S We will design odata struc t tSures that combine the speed of WODS with soecurity/reSsilience (history- t o M M M e m e m e m independent storage). A snapshot from history-independent data structures would not reveal c c c i r anything about the past historyi rof the data structure, such as the order that elements werre i l e le e l i inserted. We will explore fundameintal tradeoffs between history independence and data-structure i t t M ttM t M t a performance in main and external meamory. a i i i r l i Finally, we will consider improved efficie lrnicy for oblivious RAMs, which hide data-access pattern l si r t t t y a ya a y from intruders and untrusted platform operators. We will investigate improved efficiency and/or r r r T y security for big data platforms such as Hadoo T py and Accumulo. y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a38 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Advagnced Uncertainty Quantification Methods for gCircuit Simulation g e y e y y e 173331 | .Year 2 of 3 | Principal Investigator: E. R. Keiter . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc This research aims to devel c op reliability methods for ananlog circuit simulation, with c the goal of n r r r o e mitigating the expense of largee ensemble Xyce circuit calcuolations. The primary outpute will be o t t t M discovery of a set of advanced meMthods to determine time-dependent failure probability oMf nuclear l l l o i weapons (NW) circuits. Historically, NiW circuit designers and anoalysts rely almost exclusiveliy on o l l l i i i gt brute-force nested sampling methods, rtequiring expensive numbergs of Xyce simulations. To addtress g a a a yr this, our research is focused on aspects of runcertainty quantification (UyQ) that can be used to r y y y y mitigate this computational expense. We have recently included the addition of analytical parameter .T T . T . ce sensitivities in Xyce for steady state and transieent analysis, and surrogate mcodeling via principal e c occomponent analysis on field data. c o c o S h S h S h mn n m n m In the final year of the project, we plan to focus on two main areas: 1) hierarchical uncertainty e e e o o o quantification and 2) the use of analytic parameter sensitivities in calibration and UQ. Research Area c l o c l o c l o 1 involves decomposing a circuit into subcircuit or device models, where each submodel may be g g g r r r e charactyerized with e a lower-dimensional representation, such ays a polynomial chaos expansion or a e y . . . surrogatec model, as mentioned above. The hierarchical approach hcas the potential to make UQ more c t S o tS o S t o M feasible for complex circMuits that are computationally expensive. Research Area 2 focuses on the M e m e m e m c adjoints that Xyce now prodcuces, which can be used to reduce the number of functicon evaluations i r both for UQ and calibration mirethods. We plan to investigate this in the context of comrpact models i l e le e l i i i t t M and the construction of derivativtt e M -enhanced surrogates. t M t a a a i The successful development of this riesearch will reduce the number of simulations required ito assess r l i lri l i r t NW circuit failure due to hostile radiatiton. It is anticipated that this will impact weapons qualifitcation y a ya a y r activities (e.g., Qualification Alternatives tro the Sandia Pulsed Reactor), enhanced surveillance r T y Ty y T T programs, and various NW life extension progTrams. Advanced UQ methods have not been previouslyT e e e e attempted for production-level circuit simulatioen. This project addresses several issues that will e c crender uncertainty quantification and failure analccysis more feasible for large-scale NW radiation- c c h h h h nhardness circuit problems. h n n h n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 39 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Analysgt-to-Analyst Variability in Simulation-Based Prgediction g e y e y y e 173028 | Yea.r 2 of 3 | Principal Investigator: M. R. Glickman . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc Despite the tremendous numb c er of compute cycles devotend to computer-based simul c ation, the value n r r r o e of all this computation is only realeized when coupled with a difoferent kind of computationeal process o t t t M that is performed by human beings sMeeking to refine their understanding and make predictioMns. l l l o i Moreover, from informal observation anid discussions with analystso, we have become aware thait, even o l l l i i i gt when provided the same problem, source tdata, simulation tools, and spgecific questions, it is unliketly g a a a yr that two analysts will arrive at precisely the srame predictive judgments. y r y y y y .T Analyst-to-analyst variability is an indicator of pTotential uncertainty in simul.ation-based assessment. T . ce e c e c What impact might such uncertainty have on our national security mission? Are there effective ways oc c o c o S hto mitigatSe associated risk and/or improve decision-mahking? We don’t know because there have S h mn n m n m been few, if any, formal studies of analyst-to-analyst variability in simulation-based assessment. In e e e o o o lieu of such studies, analyst judgment is developed through professional training and mentoring and c l o c l o c l o subjected to peer review. Despite these wise practices, there remains a large disparity between our g g g r r r e relatively yintuitive app e roach to analyst judgment and our formal tyreatment of other stages of the e y . . . simulation-bacsed assessment pipeline such as hardware architecturecs and numerical algorithms. c t S o tS o S t o M More alarmingly, psycholoMgical research continues to reveal consistent biases in human judgment and M e m e m e m c decision-making, biases that wcere only divined via formal analysis. c i r ir r i l e We are conducting both: a) in sitlue observational studies of simulation-based judgment ase practiced l i i i t t M at Sandia and b) formal, controlled e ttxMperiments in which subjects are given carefully chosen t Mdata t a a a i and tools and asked to derive their besti assessment. The goal of this project is to identify and i r l i lri l i r t document significant factors that underliet variability in simulation-based assessment to prepare t y a ya a y r for investigating how this understanding mighrt be leveraged to boost predictive performance. This r T y Ty y T T novel project leverages psychological findings wTithin the already highly interdisciplinary enterprise of T e e e epredictive computer simulation of physical processees. e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a40 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T APEXg: Application Characterization for Exascale Sygstems g e y e y y e 173031 | .Year 2 of 2 | Principal Investigator: S. D. Hammond . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc Scalable parallel application c s and computer architecturens are a critical capability fo c r DOE and NNSA n r r r o e missions, NNSA’s Advanced Simeulation and Computing Proogram, and many other natioenal security o t t t M related programs. As the communMity continues to pursue significant increases in the perfoMrmance l l l o i of supercomputing hardware, we arei quickly finding that the maopping of existing algorithms iand o l l l i i i gt programming models to leading-edge htardware is becoming highly gstrained. This is, in part, a retsult g a a a yr of the changing approach to delivering harrdware performance–paralleylism is offered instead of r y y y y increased core clock frequency–as well as implicit assumptions regarding the behavior and structure .T T . T . ce of machines that no longer continue to hold. Wheen combined with the challcenges of lowering power e c occonsumption and the reduced reliability of futurec hardware systems used at scoale, the use of existing c o S h S h S h mnproduction codes on future supercomputers without mnodification now seems unlmikely. Application n m e e e odevelopers are, therefore, facing the prospect, for perhaops the first time in two decades, of needing to o c fu l nodamentallyc rearchitect, and in some cases rewrite, larg l eo sections of key production code. Whilst c l o g g g r a full applicationr rewrite may seem an appealing route, the cost and time associated with this activity r e y e y e y alone me.ans it cannot be undertaken for more than a handful of c.odes or libraries before it becomes . c c c t S prohibitive. oAn alternatStive approach is to evolve applications througoh the identSification of key kernels t o M M M e and libraries amddressing peerformance, power, and reliability concerns min a step-weise manner. m c c c i r We are proposing a suite of birinary analysis and instrumentation tools to support devrelopers in i l e le e l i i i t t M characterizing their existing apptl t i M cations, with a specific focus on identifying areas of contc M ern relating t a to performance, data motion, and raeliability. The output of these tools will relate program level a i i i r l i objects such as data structures, functliroins, loops, or even code segments, to specific program ml ietrics r t t t y a of interest in exascale systems. These m y eatrics are being identified by research using the Sandia-laed y r r r T y NNSA/ASC future architecture test bed prToygram and collaborations with leading industry vendors.y T T T T e e e e e e Summary of Accomplishments: c c cc c c hThe tools produced in this project, the first of any nahtional laboratory, have been used to provide h h h h nspecific and accurate calculation of achieved levels of nfloating point and integer vectorization. Our n n o research was able to demonstrate that vectorized floati onng point calculations, as expected by the o n l l l o o o o developer, show poor correlation with hardware-reported poerformance counters, indicating that o g g g l tools buyilt upon these counters are likely to be provide incorrelcyt information to NNSA/ASC code y l o teams. Co. cllaboration with Intel persuaded the Advisor-XE tools o g.rcoup to develop a vectorization . c o S g analysis extoension base S d largely on the approach illustrated in this pgoroject. Fur S ther, through the o g e m e m e m y development of methods for assessing application structure, we providyed insight into where large y c c c r . scientific codes spend the morst significant amount time executing instruct.ions. The dremonstration . e e e c c c t of this analysis on the SIERRA ADtAGIO and ALEGRA Albany codes show the ability of thtis research S M oto scalSe to production-class work M loads. Extensions of this work to multi-threadoed applica M tionsS has o i i i l msince been employed on a number of rlesearch problems including analysis of kernmels for Trilinlos, m i i i e t e t te a Albany, SIERRA, and Kokkos–many of wahich are actively targeting the ASC Trinity and other a c r c r rc y platforms. We also extended the analysis myethods to the Xeon Phi Knights Corner platform and y r T then, with collabroration from Intel, limited iniTtial analysis of program structures for the Xeon Phi Tr e e e e ee Knights Landing instruction-set architecture (ISA) to be used in Trinity. The project provided the first c c c t t t hassessment of program instruction behavior for thehse platforms and was reported to several ASC h M M M nprojects to help guide application porting and optimiznation activities. n o o o i l i l li l Sigonificance: l o ol i i i g g g t The project has provided significtant insight into program behavior for production-class algorithms, t a y a y ya and in so.me cases, full production codes. The assessment of this b.ehavior is enabling production code . c c c r r r S developers too understaSnd weaknesses in code performance and areoas where oSptimization may need o y y y e m e m e m to be considered. The instruction profile data is being provided to industry vendors to help improve T c c T c T r processor/ISA design for futurre products that may be selected for NNSA/ASC producrtion computing e e e e e e platforms. The project, therefore, is helping to improve algorithm performance on both contemporary t t t c M and future systems. M c M c h i l i l h i l h i i i t t t n a a n a 41 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Cognitigve Computing for Security g g e y e y y e 165613 | Yea.r 3 of 3 | Principal Investigator: E. Debenedictis . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc The original project purpose w c as to study applications of mnemristor-based neural netw c orks to n r r r o e security systems, but the scope cheanged over its lifetime. Thesoe “new” security systems weere to have o t t t M had two features: they would have aM novel physical security feature that stores critical inforMmation l l l o i in memristors, which could be quickly eriased or otherwise used foro information protection. Thei o l l l i i i gt cognitive features would also be able to letarn certain security relevantg tasks on the fly, such as t g a a a yr identification of malware. These objectives chranged and expanded over tiyme. r y y y y .T The project identified that the original direction Twas not practical based on c.urrent memristor T . ce e c e c technology, which led to development of a comprehensive theory on when certain new transistor- oc c o c o S hreplacemSent device could outperform conventional cohmplementary metal oxide semiconductorsS h mn n m n m (CMOS). The original idea of malware detection also expanded over time to an understanding of e e e o o o how to embed a general-purpose computer into a neural network–leading to a reconciliation of the c l o c l o c l o current “Turing machine” class computer that is a ubiquitous office machine and the other main class g g g r r r e of computyer in the wo e rld, which are the brains of animals. y e y . . . c c c t S Based on our n o ew researtcSh direction, we sought to identify the scaling o limits of arStificial neural t o M M M e networks built frmom analog eelectronics, applying the resulting know-howm to a securiety application as m c c c a demonstration. There has been science fiction-class vision for many decades that mankind would i r ir r i l e make artificial neurons that can dleuplicate the human brain. While realizing this vision is liekely far l i i i t t M off, there is a reasonable intermedia tttMe goal of having artificial neurons and related brain-ins t pMired t a a a i techniques address useful problems at simaller scale. In fact, there is a good possibility that artifiicial r l i lri l i r t neural networks could fill a role as a “Beyotnd CMOS” or “Beyond Moore” computing system. t y a ya a y r r r T y To make progress toward such an intermediatTey goal requires a clear understanding of the capabilities y T T T T e and limitations of artificial neurons as engineerede subsystems, with an ultimate goal of understanding e e e e how to make reliable, manufacturable artificial neural networks at appropriate scale. This includes c c cc c c hdeveloping CMOS analog electronics in Sandia’s MESAh facility that fill the equivalent role of neuron h h h h n n n cell bodies. The CMOS analog electronics will be integrated with evaporatively deposited memristors n o on o n to fill the role of synapses and accompanied by the neural network learning theory to learn/program l l l o o o o suitable applications. o o g g g l y ly y l o We identifie . d a security application that would create a useful neuroa . l network at a manufacturable . o c c c S g scale, which is e o ssentially San encryption device implemented with artifi go cial neuronSs. The device is o g e physically constrmucted so it ies encased in its neurons. Disassembly wouldm damage thee encryption m y y y c c c parameters, so the system would neither encrypt again nor reveal the keys. r . r . r . e e e c c c t M o Summary of Accomplishments: t M o t M o S The projeSct began with extensive characterization of memristors. The activity began in an S i i i l m l m l m e i t environment ewhen memristors were beliei tved to have “figure 8” curve traces–there was heavy deibtatee a a a about the precise shape of the figure 8. The noteworthy result is that we characterized memristor c r c r rc y y y more comprehensively than other organizations (at the time), discovering that memristors actually r T r T Tr e e had additional internael state and “noise,” both facto e rs invalidating the entire concept that the device ee ccould be characterized by a curve of any sort. This didc not mean memristors were less useful, but c t t t h h h M rather that they would onlyM work in different circuits than had been studied by that time. M n n n o o o A second development was a theoretical framework for energy efficiency of these new non-CMOS i l i l li l device o s. Essentially, CMOS is widlely believed to have a theore o tical energy minimum of kT ln(2) per ol i i i g g g t t t gate operation. Analysis of memristor circuits showed a minimum, also in units of kT, but instead of a y a y ya the coefficie.nt being the constant ln(2), it scaled with N and the amo.unt of precision. In other words, it . c c c r r r S is not a bad devoice in termSs of energy efficiency, but its behavior is quitoe different Sthan transistors. o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a42 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T This lged to an exploration of computer architectures that wougld be compatible with memristors– a g e host ofy other devices with unrealized R&D pot e ential. The overayrching result from this research is the y e . . . c discovery cof a way to make general-purpose compcuters, which is acn important development. With this c c S o S o S o h discovered, we noted, interestingly that the circuits shcaled up the number of small neural networks h e m e m e m nc whereas most of the researcch community expected a sinngle neural network to scalec up in size. n r r r e e e o o o t Significance: t t M M M l As this project concludes, it is contributing to the national dialolg on new computers. The project l o o o i i i l i developed a new computing model inlciorporating aspects of the von Neumann computer archliitecture gt t g t g a and neuromorphic computers (or brainsa), accompanied by an energy and power model and a yr r y r y y experimental and theoretical validation. Thyese ideas were adopted internally and influenced ongoinyg .T T . T . research in adaptive neural algorithms for threat detection. The results, in collaboration with insights ce e c e c from Hewlett-Packard Laboratories’ Fellow Stan Williams, were developed into whitepapers for the oc c o c o S hOffice S of Science and Technology Policy (OSTP), withh the combined set of ideas being announc S ed as h mn n m n m e an OSTP Neanotechnology-Inspired Grand Challenge for the Next Decade on the “Sensible Machinee.” o o o c OSl o TP is essen c tially using this set of ideas to set an R&D agle o nda for the National Strategic Computing c l o Initiagtive’s fundamental research arm at the national level. g g r r r e y e y e y Refereed. cCommunications: . c . c t M S F. Rothgangeor, B.R. Evan t s S , MJ.B. Aimone, and E.P. Debenedictis, “Training Neuoral Hardwa S re with Noisy Compo- t M o e m e m e m nents,” in Proceedings of the International Joint Conference on Neural Networks, pp. 1-8, 2015. c c c i l r e A.J. Lohn, P.R. Mickel, J.B. Aimon irele , E.P. Debenedictis, and M.J. Marinella, “Memristors as Synaprs e es in Artificial i l i i i t t M Neural Networks: Biomimicry Beyott n M d Weight Change,” in Cybersecurity Systems for Human Cognitti M on Augmenta- t a tion, pp. 135-150, September 2014. a a i i i r l i t M. Marinella, et al., “Development, Chara l c ritterization, and Modeling of a TaOx ReRAM for a Neuromorp l hiitc Accel- r y a ya a y erator,” presented (invited) Meeting Abstracts. No. 42, The Electrochemical Society, vol. 64, pp. 37-42, 2014. r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 43 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Coupligng Computational Models: From Art to Scienceg g e y e y y e 173025| Yea.r 2 of 3 | Principal Investigator: P. B. Bochev . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc The purpose of this project is t c o formulate, develop, and denmonstrate new and mathem c atically n r r r o e rigorous approaches for the coupeling of disparate numerical moodels. e o t t t M M M l For many mission-critical DOE applications, accurate and efficienlt coupling of trusted simulation l o o o i i i l codes could expand scope, and deliver nelw modeling capabilities for a fraction of the time and colst i i i gt t g t g a of new development. Of particular importaance to DOE/NNSA, and the main focus of this project, a yr r y r y y are coupled computational models (CCM) for hyigh-consequence analyses of impact fuzes and asset y .T protection. Current state-of-the-art CCM posesT the coupling as a “hard” con.straint on the models. T . ce e c e c Mathematically correct implementation of such couplings can be difficult, and is often substituted by oc c o c o S hsimpler adS hoc solutions. Such tactics, while expedient,h are frequently mathematically questionabSle h mn n m n m (e.g., regarding stability), physically inconsistent (e.g., regarding conservation), and sometimes e e e o o o computationally catastrophic (e.g., leading to clearly inadmissible results). We propose to significantly c l o c l o c l o improve the robustness, accuracy, and physical fidelity of CCM by developing a new mathematically g g g r r r e rigorous, yyet adaptabl e e, coupling strategy that casts the coupling iynto a constrained optimization e y . . . problem. c c c t S o tS o S t o M M M e Posing CCM as am constrainede optimization problem—to minimize the mismmatched “eenergy” between m c c c states subject to physical constraints—enables a reversal of roles, where the coupling adapts to i r ir r i l e the physical models. Furthermorele, since the basic approach relies on generic optimizatione ideas, l i i i t t M adaptable optimization-based coup ttliMngs have significant potential for other multi-physics p t rMoblems, t a a a i and for informing future research directiions in solving such problems. i r l i lri l i r t t t y a The proposed adaptable coupling strategy yias a first-of-its-kind application of optimization ideas to a y r r r T y CCM that aims to deliver an advanced simulatTiyon capability. y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a44 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Datag-Driven Optimization for the Design and Contgrol of Large-Scale Systems g e y e y y e 180822 | .Year 1 of 2 | Principal Investigator: D. P. Kouri . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc This project will fuse data, u c ncertainty, and simulation inn novel algorithms and discr c etizations for the n r r r o e rapid and robust solution of miession-critical optimization prooblems. e o t t t M M M l Decisions are often formulated as optimization problems. In mlany applications, these problems l o o o i i i l are simulation-based and involve partlial differential equations, ordinary differential equationls, or i i i gt t g t g a differential algebraic equations. Such praoblems are often large-scale and the governing simulatioan yr r y r y y typically involves uncertain input parameteyrs. An example application is the reliability testing of y .T components using direct field acoustic testingT (DFAT). In DFAT, one must d.etermine the speaker T . ce e c e c output that reproduces target vibration profiles through the component while accounting for oc c o c o S hmateriSal uncertainties. Another application is to deshign “off-normal tests” in which one interroSgates h mn n m n m a hardware or software system to discover vulnerabilities. To design a robust, off-normal testing e e e o o o harness, one must account for uncertainty in system components. For such problems, using data to c l o c l o c l o reduce uncertainty is crucial. g g g r r r e A comm y on solutione approach for these problems is to assume a y known probability distribution for the e y . . . c c c t S uncertain pa o rameterst, Sdiscretize the governing dynamics using stan o dard numeSrical approximation t o M M M e schemes, and msolve the reesulting finite-dimensional optimization probmlem. In maney applications, such m c c c as reliability engineering, assuming a known probability distribution can lead to unreliable system i r ir r i l e performance with potentially lceatastrophic outcomes. To circumvent this issue, we proepose novel l i i i t t M discretizations and optimization t tfMormulations that do not assume a known probability di t sMtribution, t a a a i but rather use data to determine an oiptimal solution that is robust to our ignorance of the truie r l i lri l i r t probabilistic characterization of the untcertain inputs. The expected outcome of this project is at y a ya a y r rigorous and broadly applicable frameworrk for optimal design, control, and decision making in ther T y Ty y T T presence of uncertainty that combines data, sTimulation, and optimization. T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 45 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Decisiogn Analytics for Complex Supply Chain Networgks g e y e y y e 178670 | Yea.r 2 of 2 | Principal Investigator: G. K. Kao . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc The purpose of this project is t c o develop decision-support tnechnologies to enable decis c ion makers n r r r o e to perform risk-based cost-benefiet prioritization of security inovestments to manage suppely chain o t t t M integrity and risk. We propose a newM systematic approach to examine the lifecycle phases (eM.g., design, l l l o i implementation, testing, deployment, miaintenance, retirement) of osupply chains by leveraging ai o l l l i i i gt new security risk metric based on the degrtee of difficulty adversaries wgill encounter to successfultly g a a a yr execute the most advantageous attack scenarrio. The key challenge is thaty the scale of the end-to- r y y y y end supply chain lifecycle problem is too large and complex. To manage this complexity, we propose .T T . T . cea hierarchical decomposition methodology in modeeling the supply chain lifecyccle. Our framework e c occonsists of: 1) hierarchical representation, 2) macro sctate-of-health assessment, 3o) adversary c o S h S h S h manction evaluation, 4) difficulty and consequence evaluationn, and 5) optimization. Them hierarchical n m e e e reporesentation modeling enabled us to decompose the supoply chain problem into manageable o c piece l so, such thatc risk and state-of-heath evaluation can be pe l roformed. High-level metrics will enable c l o g g g r decision makers andr analysts to identify critical areas for further deep dives. The macro state- r e y e y e y of-health as.sessment is analogous to measuring vital signs before ad.ministering any diagnoses. . c c c t S Data analytic toechniquest wSill address data integrity issues, where deepo dive analySsis will consist of t o M M M e generating advermsary scenareios based on functional grammars. Difficultym and conseqeuence metrics m c c c will enable decision makers to overcome complexity of quantifying security risk, and will be suited i r ir r i l i e for cost-benefit optimization. Ne l we i optimization techniques, such as multilevel optimizatieon methods, l i t t M will be explored. By design, the hier t a tMrchical and recursive approach will address the supply t cMhain t a a a i problem comprehensively at various deipths. Our methodology will help manage the complexityi of r l i lri l i r y t a the problem, such that analysts can undertsyt a and and gain insights into the end-to-end supply chaint t a o y r make risk-informed, effective decisions. r r T y Ty y T T T T e eSummary of Accomplishments: e e e e c cWe designed a new holistic supply chain lifecycle decccision analytic framework that enables decision c c h h h h makers to perform risk-based cost-benefit analyses to ohptimally apply mitigation options to address h n n n n vulonerabilities in supply chains. The framework is demonstor n ated by a tool chain consisting of four o n majolro components: l o l o o o o g g g l 1. Hiyerarchical Supply Chain Representation ly y l o . o. . o c c c 2. Sandia’s Difficulty and Consequences Risk Based Assessment S g o S go S o g e m e m e m y 3. Semi-Automatic Grammar Based Vulnerability and Mitigation Assyessment y c c c r . r . r . e 4. Optimization Based Deciseion Analytics e c c c t t t M oTwo data representation models, lev M eraging graph databases, and traditional relaotional data M bases, o S S S i l mwere developed to capture the hierarchiilcal lifecycle supply chain. These representatmion modelsi l m i i i e t e t te a enabled high-level state-of-health evaluatioan based on our newly developed supply chain risk metraics. c r We then developced a set of context free gramrmar, which were applied to our representation models rc y y y r T to generate an attacrk surface. Our method generTated attack graphs that were not bounded by subject Tr e ematter experts’ input, e and provided a more compreehensive attack surface for assessment. Bi-level ee c c c t hoptimization problems wtere formulated for attack graphh assessment. These optimization techniques h t M M M ennabled analysts to identify attractive attack paths basedn on adversary’s difficulty, defender’s n o o o consequences, and other constraints. i l i l li l o l o ol i t A patengt, Framework and Methodolo i gty for Supply Chain Lifecycle Agnalytics, was also submitted, and g i t a is currentlyy under review. Furthermoare, we demonstrated this fraymework on multiple-use cases. ya . . . r We applied ocur framework to real life prorcesses and electronic compconents, which also resulted in c r S o S o S o y e publication demo m nstrating aepplications. y m e m y T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a46 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Signigficance: g g e Our resyearch builds the capabilities to help str e engthen the founydation of the nation’s information y e . . . c and commcunication technologies critical infrastruccture supply chacin. This problem addresses issues c c S o S o S o h that will not only have a national impact, but will alsoh have a global impact. We have developed new h e m e m e m nc assessment methodology thcat advances the science in sunpply chain risk managemecnt. The models n r r r e developed will help federal ageencies (e.g., DOE, DoD, DHS) to evaluate security impacte in designing o o o t and developing a viable and secutre global supply chain. This research will, no doubt, provtide insights M M M l l l o to the national security risk on the reliance of the global supply cohain. o i i i l l l i i i gt t g t g a Refereed Communications: a a yr r y r y y G. Kao, J. Hamlet, R. Helinski, J. Michalski, and Hy. Lin, “Macro Supply Chain Security Decision Analytics,” preseynt- .T ed at the Annual Computer Security Applications ConTference (ACSAC), 2015. . T . ce e c e c ocG. Kao, H. Lin, B. Eames, J. Haas, A. Fisher, J. Michalski, cJ. Blount, J. Hamlet, E. Lee, J. Gaouthier, G. Wyss, R. Helins- c o S hki, D. Fr S anklin, and L. McLay, “Supply Chain Lifecycle Dechision Analytics,” in Proceedings of International C S arnahan h mn n m n m e Conference oen Security Technology (ICCST), pp. 1-7, 2014. e o o o c G. l Koao, J. Hamlcet, R. Helinski, J. Michalski, H. Lin, and M. Shakam l uori, “Supply Chain Security Decision Analytics: c l o r Macrgo Analysis,” inr Proceedings of IAEA Computer Security in a Nucleagr World, 2015. r g e y e y e y M. McCro.ry, G. Kao, and D. Blair, “Supply Chain Risk Management: The. Challenge in a Digital World”, in Proceed- . c c c t S ings of IAEA Coomputer Sec tSurity in a Nuclear World, 2015. o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 47 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Detectging Lateral Movement on Internal Networks g g e y e y y e 188321 | Yea.r 1 of 1 | Principal Investigator: S. Martin . . c c c c c c S o S o S o h h h e Project Purpose:m e m e m nc After gaining access to a netwo c rk (via phishing, drive-by-donwnloads, etc.), attackers at c tempt to n r r r o e move laterally within that networek, with the goal of stealing croedentials, gaining access toe additional o t t t M computers, and ultimately compromMising servers and stealing secrets. Currently, there are liMmited l l l o i methods for detecting lateral movemenit within a network, and it iso unknown to what degree i o l l l i i i gt intruders can be detected using authentictation data. Authentication isg difficult to obtain and parset, g a a a yr and it is not known to what extent existing algrorithms will uncover lateraly movement in networks. r y y y y We propose to develop automatic lateral movement detection methods by applying machine learning .T T . T . cealgorithms to authentication data. e c e c oc c o c o S hOne sourSce of user-based data relevant to detection ofh lateral movement is authentication data, Sas h mn n m n m might be found in Kerberos logs, or even packet capture data. Researchers at LANL have assembled e e e o o o a new unique dataset containing authentication data. The data includes red team activity as ground c l o c l o c l o truth. This dataset is ideal for developing and refining our algorithms. g g g r r r e We will in y vestigate thee use of machine learning algorithms (anoma y ly detection and/or supervised e y . . . c c c t S learning) to aut o omaticallytS identify lateral movement within a network. o We will apSply our algorithms t o M M M e to LANL’s augmemnted authenetication dataset; if successful, our approachm will greatlye benefit m c c c cybersecurity personnel, improving the chances of detecting targeted attacks in a timely manner. i r ir r i l e le e l i i i t t M Summary of Accomplishments: ttM t M t a a a In order to establish the feasibility of using authentication data to detect lateral movement within i i i r l i t internal networks, we utilized two datas l eritts collected at LANL. One dataset spanned a nine-mon l tiht r y a ya a y period with very few details per event, and the other spanned a two-month period with many details r r r T y per event, including red team activity. We ana T lyyzed the data using simple statistics and various y T T T T e machine learning algorithms, including graph layoeut, dimension reduction, supervised learning, and e e e e c canomaly detection algorithms. For both datasets, ourcc results were encouraging, and pointed the way c c h h h h towards further analysis and eventual application. h h n n n n For o the nine-month dataset, we constructed both a bipartit one user-computer graph and user activity o n l l l o o o o profiles for each user over time. The user-computer graph shoowed definite groups of users and their o g g g computers on the LANL network, and how these groups were connected. l y ly y l o . o. . o For the two-mconth dataset, we constructed personal authentication csubgraphs (PASs), which are c S g o S go S o g e user-computer g m raphs for a geiven user. Simple graph statistics were comp m uted for eaech user on that m y y y c user’s PAS, including: numberc of nodes, number of edges, diameter, max in degree, macx out degree, r . r . r . e and number of user changes. We ethen used these statistics as input features for support veector c c c t M o machines (SVMs) and anomaly detetcMtion algorithms in order to detect red team a o ctivity. ThteM SVMs o S were ableS to detect red team activity with up to 77.5% accuracy, with the area under the receiverS- i i i l m l m l m e i t operating-chaeracteristic (ROC) curve scoirtes up to 0.81. The best anomaly detection algorithm wai ts e a a a able to achieve an area under the ROC curve of 0.85 on the same task. c r c r rc y y y r T r T Tr Significance: e e e e ee Once an attacker gains access to a single computer on a network, he/she will attempt to move c c c t t t M hlaterally within the networMk. The goal is to steal credenhtials, gain access to additional computers, h M n n n and ultimately compromise servers and obtain data. Sometimes this sequence of events can happen o o o i very lquickly. A system that autoimatically alerts cybersecuritly personnel would significantly impact li l o l o ol i multipleg security and consumer doimains.. g g i t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a48 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report COMPUTING & INFORMATION SCIENCES S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Efficgient Probability of Failure Calculations for QMgU using Computational Geometry g e y e y y e 165617 | .Year 3 of 3 | Principal Investigator: S. A. Mitchell . . c c c c c c S o S o S o h h h e Project Purpomse: e m e m nc This project focuses on new c algorithms for uncertainty qnuantification (UQ) and opt c imization (Opt), n r r r o e as well as mesh generation. Theese algorithms are of generalo use for a broad variety of reeliability and o t t t M design questions related to NNSAM’s stockpile stewardship mission. M l l l o o o i i i l The key technical challenge is improvling the efficiency of UQ and Opt methods (i.e., the accurlacy i i i gt t g t g a that can be obtained in a given computataional time). In realistic applications, the computational taime yr r y r y y is dominated by function evaluation calculaytions, which are typically large scale. Thus, efficiency y .T is increased if we can obtain the same informaTtion using fewer function ev.aluations, where each T . ce e c e c function evaluation is a sample point in our approach. Hence, efficiency translates into innovative, oc c o c o S hadaptiSve sampling approaches that place samples ath locations where the most overall informaStion h mn n m n m can be gained. There are both challenges and benefits related to the use of surrogates, which are e e e o o o explicit functions constructed from prior samples. What is unique to our algorithms is the heavy c l o c l o c l o use of geometric concepts. Instead of boxes, our sample’s neighborhoods are spheres or Voronoi g g g r r r e cells. Thye size of th e ese neighborhoods is adapted, based on infoyrmation gained from new samples. e y . . . Further, incstead of sample points, we consider sample lines, planesc, hyperplanes, etc. Sometimes this c t S o tS o S t o M recursive-by-dimensionMs organization must be handled numerically; other times, we gain immense M e m e m e m c efficiency by exploiting the canalytic form of a sphere or surrogate. The general-purpcose algorithms i r developed in this project willi rbe included in Sandia’s publically available Dakota optimrization and UQ i l e le e l i i i t t M toolkit, as well as in stand-alone tm tM eshing software. t M t a a a r i l i Summary of Accomplishments: i lri i l i r t t t y a The intent of the project was to utilize cyoamputational geometry and computer graphics techniquaes to y T r y develop new algorithms for quantified marr Tg y ins of uncertainty, a form of uncertainty quantificationr. y T T Algorithm development focused on using samTpling techniques—notably hyper-line sampling and T e e e e e e sphere packing sampling—together with Voronoi diagrams and surrogate functions. These new c c cc c c halgorithms are ideal for applications involving the ushe of numerical integration and surrogate h h h h nfunctions of some form. Additionally, algorithm develonpment yielded a number of mesh-generation n n o applications that utilize the same core sampling techniq onues and Voronoi diagrams. Many of the o n l l l o o o o techniques developed in this project were deployed in stando-alone demonstration software. Some of o g g g l the moyre practical capabilities—notably some uncertainty quanlytification and optimization algorithms, y l o together. wcith a library for analytic surrogate functions for interp o.oclating data—were deployed in the . c o S g Dakota optiomization to S olkit. go S o g e m e m e m y y y c c c Significance: r . r . r . e e e c Our results strengthened our external visibility, which helps enhance the victality of the workforce c t t t M o(e.g., for recruiting talented new rMesearchers). Partnering with academia durinog algorithmM o S S S i l mdevelopment also allowed us to gain iilnsight into the techniques of other fields (e.gm., computeirl m i i i e t a graphics aned geometric modeling), and teanabled the application of our core algorithms to other tfiaeleds. c r Our results adcvanced mathematical and cormputer science capabilities for conducting science andr c y y y r T engineering activrities by providing more efficTient and new meshing and opt/UQ algorithms, with Tr e e enhanced understaending of the key mathematices underlying these techniques. The methods were ee c c c t vetted on problems wtith the same structure as mission-relevant applications. The project also had a t h h h M M M nstrong cross-disciplinary approach, integrating effortsn from computational and engineering sciences, n o o o so as to consider method applications. A number of natural follow-on projects emerged based on i l i l li l theo research outcomes, and selveral of these were funded. Foinally, the project spawned technical ol i i i g g g t achievements and numerous pubtlished papers, and spanned general sampling techniques, giving rise t a y a y ya to uncert.ainty quantification, global optimization, modeling, and .meshing techniques. . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a a n a 49 n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m Sandia National Laboratories 2015 LDRD Annual Report SECNEICS NOITAMOFNI & GNITUPMOC S S S S S S e e e e c e c c e r r r c e c e e c r t M r t M t M r e e e i i i l l l t i t i i t t t t M a M a a M r r r y y y i T i T T i l l l i e i e e i t t t c c c a a a h h h r n r n n r y o y o o y l l l T o T o o T Refereegd Communications: g g e A.A. Rushdyi, S.A. Mitchell, C.L. Bajaj, and M.S. Ebeida, “ e Robust All-Quad Myeshing of Domains with Connected y e . . . c Regions,” prescented at the International Meshing Roundtablce, Austin, TX, 201c5. c c S o S o S o h h h e M.S. Ebeida, S.A. Mmitchell, A. Paetney, A.A. Davidson, S. Tzeng, M.A. Awad, A.H. Mmahmoud, ande J.D. Owens, “Exer- m nc cises in High-Dimensional Samplicng: Maximal Poisson-Disk Sampnling and k-d Darts,” Series: Mcathematics and n r r r e Visualization, Topological and Statisticeal Methods for Complex Data, pp: 221-238, 2015. e o o o t t t l M S.A. Mitchell, M.A. Mohammed, A.H. MaMhmoud, and M.S. Ebeida,” Delaunlay Quadrangulation by Two-Mcoloring l o o o i Vertices,” presented at the International Mesihing Roundtable, London, England, 2014. i l l l i i i gt t g t g a a a yr r y r y y y y .T T . T . ce e c e c oc c o c o S h S h S h mn n m n m e e e o o o c l o c l o c l o g g g r r r e y e y e y . . . c c c t S o tS o S t o M M M e m e m e m c c c i r ir r i l e le e l i i i t t M ttM t M t a a a i i i r l i lri l i r t t t y a ya a y r r r T y Ty y T T T T e e e e e e c c cc c c h h h h h h n n n n o on o n l l l o o o o o o g g g l y ly y l o . o. . o c c c S g o S go S o g e m e m e m y y y c c c r . r . r . e e e c c c t t t M o M o M o S S S i i i l m l m l m i i i e t e t te a a a c r c r rc y y y r T r T Tr e e e e ee c c c t t t h h h M M M n n n o o o i l i l li l o l o ol i i i g g g t t t a y a y ya . . . c c c r r r S o S o S o y y y e m e m e m T c c T c T r r r e e e e e e t t t c M M c M c h i l i l h i l h i i i t t t n a50 a n a n r r r o y y o y o T T T l l l o e e o e o c c c g h h g h g n n n y y y o o o . l l . l . o o o c c c g g g o y y o y o . . . m c c m c m o o o m m m