UK OFFICIAL [dstl] Quantum Information Processing Landscape 2020: Prospects for UK Defence and Security Authors: Andrew Middleton, Stephen Till, Lt Cdr Matt Steele RN Report Number: DSTL/TR121783 Date: June 2020 Cyber & Information Systems Division, Dstl Porton Down, Salisbury SP4 0JQ Strategic Command, Jt User Intelligence & Cyber © Crown Copyright 2020 Draft for comment Release Conditions: This document has been prepared for MOD and, unless indicated, may be used and circulated in accordance with the conditions of the Order under which it was supplied. It may not be used or copied for any non-Governmental or commercial purpose without the written agreement of Dstl. Authorisation: Project Manager: Fiona Beeching Technical Reviewer: Dr Andrew Bennett Authors: Andrew Middleton / Stephen Till / Lt Cdr Matt Steele RN Powered by the transistor and, later, the integrated circuit, technologies belonging to the 'First Quantum Revolution' came from understanding the nature of quantum mechanics. These devices are ubiquitous and critical to daily life: we wear them, we access and process information through them and they enable us to communicate. The 'Second Quantum Revolution' will see the introduction of technologies that exploit the subtler quantum effects and it is expected to have even greater impact. Many regard the rush to develop quantum computing as a new 'space race'. Through counterintuitive phenomena such as quantum entanglement and quantum superposition, a large-scale quantum computer is expected to surpass the best digital super computer by orders of magnitude and allow us to simulate the complexities of the natural world as never before. Globally, progress over the past 5 years has been remarkable, and in October 2019 Google claimed to have achieved 'quantum supremacy' – that is, a programmable quantum computer able to solve a problem that classical computers practically cannot. Within the UK during 2014, a coherent National Programme was planned and a first phase of five years Government support for translational R&D to build a sovereign quantum technology industry sector was announced in the Autumn Statement. A further five years Government support for the National Quantum Technology Programme (NQTP), which spans quantum timing, sensing, communications and computing and simulation, was announced in the 2019 Autumn Statement including the creation of a National Quantum Computing Centre. Concurrently, Dstl launched a DARPA-like project in 2014 to develop quantum technologies at pace principally focussed on quantum sensors for position, navigation and timing (PNT). In 2014, Quantum Information Processing (QIP) was judged too immature for near term Defence and Security benefit and only technology watch activities were carried on. However, progress has exceeded early expectations and a technology watch stance is no longer appropriate for MOD. This document critically surveys UK R&D in QIP hardware and software to inform what MOD investment should be considered during the near term (Era 1, 2020 – 2025), medium term (Era 2, 2025 – 2030) and further into the future (Era 3, 2030 and beyond). KP1 (1.1, p. 1): This document critically surveys emerging Quantum Information Processing technologies in the UK and the rest of the world. Recommends MOD re-consider its current technology watch-only stance on QIP. KP2 (1.2, p. 2): The UK Budget of 2018 announced government funding to develop a large-scale quantum computer, including algorithms, software, error detection, and workforce. KP3 (2, p. 6): Quantum computers are analogue machines operating fundamentally differently to conventional computers; initially they will be hybrid quantum-digital machines. KP4 (2.1, p. 7): Qubits can represent 0, 1, and all intermediate/complex numbers simultaneously. KP5 (2.2, p. 8): Different types of quantum computers compete with each other and with digital silicon solutions. KP6 (2.2.1, p. 9): Circuit model quantum computers have demonstrated superior performance for specific tasks (e.g. Google's Sycamore). KP7 (2.2.2, p. 10): Adiabatic quantum computers (quantum annealers) are available commercially from D-Wave Systems. KP8 (2.3, p. 10): Benchmarking is essential to determine whether digital technologies may be superior in the near-term. KP9 (3, p. 12): Quantum computers manipulate qubit states using lasers/electromagnetic fields; configured qubits form quantum gates implementing quantum algorithms. KP10 (3.1, p. 12): Future decision support systems (DSSs) will exploit Big Data; QIP will help manage data deluge. KP11 (3.2, p. 13): Existing classical algorithms must be recast to run on quantum computers. KP12 (3.2.1, p. 14): Few algorithms from the Quantum Algorithm Zoo are expected to have value for Defence and Security. KP13 (3.2.2, p. 15): Five NISQ algorithms of primary importance in Era 1 (Shor, Grover, QFT, QML, Quantum Annealing). KP14 (3.2.3, p. 15): Five NISQ algorithms of secondary importance in Era 1 (VQE, QAOA, DDQCL, QAE, PT). KP15 (3.2.4, p. 16): Two quantum algorithms requiring fault-tolerant computers may benefit Defence and Security in Era 3 (Quantum Simulation, Linear Equations). KP16 (3.3, p. 16): AI is increasingly used to deliver business functions. KP17 (3.3.1, p. 17): Automated data analysis using digital machines is becoming mature (e.g. The Automated Statistician). KP18 (3.3.2, p. 18): Neural nets have been developed for control, sensor processing, AI/ML, and situational understanding. KP19 (3.3.3, p. 20): Quantum neural nets offer potential for overwhelming quantum speedup. KP20 (3.3.4, p. 21): Training neural nets requires careful dataset selection; classical training data can be modified for quantum execution. KP21 (4, p. 23): QIP can be applied across all Enterprise activities where IT is in use. KP22 (4.1, p. 23): QIP will impact information processing and presentation to human decision makers through faster AI execution. KP23 (4.1.2, p. 24): In Enterprise Management (C4ISR), QIP's value lies in accurate, rapid pattern matching in Big Data. KP24 (4.2, p. 25): QIP impacts secure network transmission, authentication, and network immunity. KP25 (4.2.1, p. 26): Quantum networks enable unimpeded, secure data flow. KP26 (4.2.2, p. 27): Quantum signatures verify sender identity and are immune to quantum computer attack. KP27 (4.2.3, p. 28): Cyber mimic defence ('auto immune') systems can identify threats with quantum processing support. KP28 (4.3, p. 28): Image processing is a prime candidate for quantum advantage. KP29 (4.3.1, p. 29): Automated image searching enables recognition and tracking of events. KP30 (4.3.2, p. 29): Quantum image processing (QuImP) research exceeds quantum signal processing. KP31 (4.3.2.1, p. 30): Circuit model QuImP algorithms are at low TRLs due to hardware immaturity. KP32 (4.3.2.2, p. 30): Artificial neural nets running on NISQ machines (e.g. D-Wave) have broad defense applications. KP33 (4.3.2.3, p. 31): US industry has used D-Wave annealers for machine learning/image analysis for over a decade. KP34 (4.4, p. 32): Machine intelligence aids management of complex data. KP35 (4.4.1, p. 32): QIP will reduce data deluge by identifying correlations. KP36 (4.4.2, p. 33): QIP will extract fullest information from future quantum sensors. KP37 (4.4.3, p. 33): QIP neural nets offer situational understanding via image analysis and pattern detection in Era 1. KP38 (4.4.4, p. 34): Verification is critical for accurate situational understanding. KP39 (4.5, p. 35): Compact quantum neural nets can enable machine vision, orientation, navigation, and obstacle avoidance for robotics. KP40 (4.5.1, p. 36): QNNs offer transformational impact for autonomous vehicles. KP41 (4.5.2, p. 36): QIP will accelerate intelligent logistical systems. KP42 (4.5.3, p. 37): QNNs will revolutionize medical diagnostics and care delivery. KP43 (4.5.4, p. 38): Price is the primary constraint in domestic adoption. KP44 (4.6, p. 38): QIP contributes to Network Quantum Enabled Capability (NQEC) in future combat systems. KP45 (4.7, p. 40): QNNs will empower improved training and simulation. KP46 (5, p. 43): Modest MOD investment (~£5M over 5 years) would address gaps in UK NQTP portfolio. KP47 (5.1, p. 43): Information is fundamentally quantum; Era 1 (2020-2025) is critical for strategic investment. KP48 (5.1.1, p. 44): Organizations must prepare now to be 'quantum-ready'. KP49 (5.1.2, p. 45): Hardware platforms remain unsettled; algorithms present an early window of opportunity. KP50 (5.2, p. 46): UK NQTP works towards a sovereign, full-stack capability. KP51 (5.3, p. 48): Mid-TRL translational research is supported by Oxford-led QCS Hub. KP52 (5.4, p. 48): IUK and BEIS support NQCC commercialization via ISCF Wave 3. KP53 (5.4.1, p. 49): In 2020, IUK awarded £25.7M across 10 QIP projects under ISCF Wave 3. KP54 (5.4.2, p. 50): ISCF Wave 3 hardware spans leading platforms; software covers control and operating systems. KP55 (5.5, p. 51): ISCF Wave 3 lacks QNN coverage; second tranche in 2021 should support QNNs. KP56 (5.5.1, p. 53): Exemplar QNN challenge: NISQ algorithm to identify, localise, and track features in imagery data. KP57 (6, p. 54): NISQ computers are available now; QNNs provide near-term early wins before 2025. KP58 (7, p. 55): MOD should partner with NQTP to propose a QNN Challenge for ISCF Wave 3 in 2021 and build internal SQEP. In 2014, Dstl began a DARPA-like programme to develop quantum sensor demonstrators (quantum clocks and inertial force sensors) complementary to the UK National Quantum Technology Programme (NQTP). While QIP was initially watched from afar due to immaturity, recent rapid progress warrants reviewing MOD investment. The IT sector exceeds $5 trillion annually; QIP represents a second Information Age ('Quantum 2.0') exploiting superposition and entanglement. QIP promises quantum advantage in data analytics, AI, drug development, and materials science. Hardware development (trapped ion, superconducting, photonic) aims for 50-100 qubit NISQ machines in Era 1 (2020-2025) and large-scale fault-tolerant systems in Era 3 (2030+). Quantum computers are conceptually modern analogue computers functioning as co-processors in hybrid classical-quantum architectures. 2.1 Quantum Information: Unlike classical bits (0 or 1), qubits represent superpositions |ψ⟩ = α|0⟩ + β|1⟩. Measurement collapses the state. Preserving coherence requires extreme isolation. Entanglement provides true parallel connectivity. 2.2 Types of Quantum Computer: - Circuit Model: Logic gates act on qubits. Scalability is evaluated via Quantum Volume (IBM metric accounting for qubit count, connectivity, gate fidelity, coherence, and compiler efficiency). - Adiabatic Quantum Computers / Quantum Annealers: Ground state evolution on an energy landscape governed by the Ising Hamiltonian. Commercialized by D-Wave (Chimera and Pegasus architectures). 2.3 Competing Digital Silicon Technologies: High-performance classical alternatives include cloud computing, GPUs (massively parallel RISC), bespoke neuromorphic/neural net ASICs (Intel Loihi/NCS), and digital annealers (Fujitsu). 3.1 Information Processing in Defence & Security: Decision Support Systems (DSSs) require processing massive data streams across Situational Awareness, C3, Robotics, and Training. 3.2 Quantum Algorithms: - NISQ Era 1 Primary: Shor's algorithm (factoring), Grover's algorithm (database search), QFT (Fourier transform), QML (Machine Learning), Quantum Annealing. - NISQ Era 1 Secondary: Variational Quantum Eigensolver (VQE), Quantum Approximate Optimization Algorithm (QAOA), Data-Driven Quantum Circuit Learning (DDQCL), Quantum Auto-Encoder (QAE), Population Transfer (PT). - Era 3 Fault-Tolerant: Universal quantum simulation, HHL algorithm for linear systems. 3.3 QIP for Automation & Neural Networks: Neural networks compute in a single machine cycle on quantum hardware versus N^2 complexity on sequential digital processors. Training requires large datasets, making QNNs a major target for near-term quantum advantage. 4.1 Civil & Military Enterprise Management (C4ISR): Military command networks depend on rapid intelligence sifting. QIP accelerates pattern matching and correlation across big data. 4.2 Information Security: - Quantum Key Distribution (QKD) and quantum teleportation for secure communications. - Quantum signatures for sender authentication. - Bio-mimetic 'cyber mimic defence' (CMD) / autoimmune networks assisted by QNNs for real-time intruder detection. 4.3 Data & Information Processing / Quantum Image Processing (QuImP): - Automated image search, feature decomposition, and tracking. - QuImP representations: QBIP, FRQI, NEQR, QST, QRMW, and QIC compression. 4.4 Management Aids (AI) & Situational Awareness: Fusing diverse sensor modalities to move from awareness to understanding. 4.5 Robotics & Logistics: Compact chip-scale quantum processors enabling real-time machine vision (MV), vector planning, autonomous transport, battlefield engineering, swarm control, and telemedicine. 4.6 Future Combat Systems (NQEC): Network Quantum Enabled Capability integrating resilient PNT, secure comms, and distributed planning. 4.7 Training & Simulation: AI-driven synthetic training environments leveraging Google TensorFlow Quantum (TFQ) and Cirq. 5.1 Opportunities & Threats: Economic and national security imperatives require developing sovereign QIP capability. Early adopters will gain strategic advantage in intelligence and autonomous systems. 5.2 Commercial Sovereign Capability: UK ecosystem includes academic excellence (QCS Hub), spin-outs (Oxford Quantum Circuits, Cambridge Quantum Computing, Riverlane), and the National Quantum Computing Centre (NQCC) at Harwell. 5.4 ISCF Wave 3: £153M government matching £205M industry funding across feasibility studies, collaborative R&D, and technology projects. 5.5 Recommended Strategy: MOD should fund a targeted £5M 5-year programme focused on Quantum Neural Networks (QNNs), collaborating with Innovate UK on a 2021 ISCF Wave 3 Challenge for automated imagery feature detection and tracking. Summary & Conclusions: Quantum computing is not decades away from military relevance; NISQ systems can deliver near-term value in neural net pattern recognition and logistics optimization within 5-10 years. Recommendations: MOD must immediately invest in building internal Suitably Qualified and Experienced Personnel (SQEP) in QIP, conduct benchmarking against digital silicon, and partner with NQTP/IUK on a dedicated QNN Challenge in 2021. Acknowledgements: Advice acknowledged from Roberto Desimone (BAE Systems), Viv Kendon (Durham), Sir Peter Knight (UKNQTP SAC), Roger McKinlay (ISCF), and Prof. D Paul (Glasgow). Appendix A: Introduction to quantum computing paradigms, Bloch sphere representations, qubit fidelity, quantum volume, circuit model, adiabatic/annealing, MBQC, and topological quantum computing. Appendix B: Hardware survey (IBM transmons, Google Bristlecone/Sycamore, Intel spin/superconducting qubits, Microsoft topological anyons, Honeywell trapped ions, D-Wave SQUIDs/Pegasus, photonic quantum computing, and digital silicon competitors). Appendix C: Primary Era 1 algorithms (Shor, Grover, QFT, Annealing, QML). Appendix D: Secondary Era 1 algorithms (VQE, QAOA, DDQCL, QAE, PT). Appendix E: Era 3 algorithms (Deutsch-Jozsa, Quantum Simulation, HHL Linear Solvers). Appendix F: Comparison of Boltzmann Machines, Restricted Boltzmann Machines (RBMs), Hopfield, Hamming, and Feedforward Neural Networks. Appendix G: Quantum image processing representations (QBIP, FRQI, NEQR, QST, QRMW, QIC). Appendix H: TensorFlow Quantum (TFQ) architecture and Cirq integration. Appendix I: International Quantum Programmes (Australia CQC2T, Canada NSERC/CFI/CIFAR/IQC, China $11B Hefei NQL/Micius/Alibaba, EU Quantum Flagship, Japan Q-LEAP/CREST, Russia RQTR/Skoltech, UK NQTP/NQCC, US National Quantum Initiative). Appendix J: Industry and UK supply chain capability (D-Wave, Google, IBM, Intel, Microsoft, Oxford Quantum Circuits, ATOS, Cambridge Quantum Computing, Riverlane).