04.+Quantum+NAMS-2+Industry+Day+Slides.pdf

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov

The cornerstone of NASA’s current and future missions

A NASA Perspective on Quantum Computing

Eleanor G. Rieffel NASA Senior Researcher for Advanced Computing and Data Ana NASA Ames Quantum Artificial Intelligence Laboratory (QuAIL)

Shon Grabbe QuAIL Deputy Lead

NAMS-2 Industry Day NASA Ames Research Center Organization Code Briefings

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov 2

TI Division Portfolio

Autonomous Systems and Robotics (ASR) Enabling complex systems and operations through the development of technologies that can respond or adapt to changing conditions, knowledge, and constraints

Collaborative and Assistant Systems (CAS) Building advanced information systems to address data acquisition; data conditioning and integration; data storage, organization, and access; data visualization; and data analysis, mining, and extraction

Discovery and Systems Health (DaSH) Developing tools and methods for system health management, machine learning, physics modeling, computational materials modeling, and quantum computing

Robust Software Engineering (RSE) Increasing the reliability, robustness, and productivity of NASA's mission software through research, development, and application of automated software engineering technologies

FRET

Requirement analysis and formalization

CoCoSim Simulink model verification

IKOS

C/C++ code static analysis

AdvoCATE Assurance case automation

RACE/MESA

Run-time analysis of live data streams from airspace

AdaStress Adaptive stress testing and interpretable categorization

SYSAI

Model-based analysis of realizable goals in systems

R2U2 runtiSystem health management & me monitor

Concept of Operations

Requirements

Models

Code

Unit V&V

System V&V

Operations

Prophecy/SafeDNN Property inference, probabilistic analysis and formal explanations for neural networks

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n s 3

Why Quantum Computing at NASA?

Quantum computing has the potential to provide vastly more efficient computation for some applications

• Ability to compute what could not be computed even if every atom in the universe were a classical processor running for the entire age of the universe o e.g., factoring, certain material science applications

• Significant speedups in other areas such as optimization, machine learning

• Harnesses uniquely quantum effects

Green computations

• Low energy consumption

NASA constantly confronts massively challenging computational problems

• Computational capacity limits mission scope and aims

Pleiades supercomputer at NASA Ames

We do not build quantum computing hardware, but partner closely with groups that do

ESM FY19 TAPR

NASA QuAIL: Mandate and Partnerships:

Quantum hardware partners

Interagency partnerships

Google+NASA+ORNL collaboration

NASA QuAIL mandate:

Determine the potential for quantum computation to enable more ambitious and safer NASA missions in the future

Google

Rigetti

Partner on DOE NQI centers

Serve as SQMS Chief Scientist

T&E for multiple DARPA programs

International partnerships

Collaborations with other NASA Centers

• SCaN: Quantum network

(HQ, ARC, GRC, GSFC, JPL)

• Quantum comm., computing, sensors, networking (GRC)

• Follow-on from ARMD CAS QTech effort

QuAIL: ~15-person team + interns/visitors

ESM FY19 TAPR

NASA Quantum Artificial Intelligence Laboratory (QuAIL)

Application Focus Areas Planning and scheduling

Material science Fault diagnosis

Machine Learning Software Tools & Algorithms

Quantum algorithm design Mapping, parameter setting, error mitigation Hybrid quantum-classical approaches Compiling quantum algorithms to hardware Distributed quantum computing

Solvers & Simulators Physics-inspired classical solvers HPC quantum circuit simulators

Physics Insights Co-design quantum hardware

E. Rieffel et al. (2019), From Ansätze to Z-gates: A NASA view of quantum computing, Advances in Parallel Computing 34, 133-160 R. Biswas et al. (2017), A NASA perspective on quantum computing: Opportunities and challenges, Parallel Computing 64, 81-98

Quantum-Enhanced NASA Applications

Quantum Tools & Programming

Novel Classical Solvers & Simulators

Quantum Comm & Networks

Quantum Algorithms

Fundamental Physics Insights

QuAIL team published

150+ peer-reviewed papers since

Focus: quantum computing, with connections to quantum networking, quantum communication, and quantum sensing

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n s

Exploration of Quantum Computing for NASA

Quantum Simulations for Aerospace Materials

Resource Allocation and Scheduling for Space Exploration

Air Traffic Management

Quantum ML for Earth Science Data Analysis

Secure Airspace

Communication

Quantum Optimization for Mission Planning & Coordination

Entanglement

Tunneling

Superposition

Objective: Find “BETTER” solution faster time-to-solution OR more precise solution OR not found by classical methods

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov 7

NASA’s Potential Quantum Computing Requirements (1/4) Algorithm development research

• Develop novel quantum and quantum-classical hybrid algorithms, including distributed algorithms, for combinatorial optimization, machine learning, data analysis, and simulation challenges in aeronautics, space, and Earth sciences

• Develop novel quantum and quantum-classical hybrid algorithms for partner applications in areas such as high-energy physics (HEP) and condensed matter physics (CMP)

• Develop novel quantum algorithms or innovative variants of existing quantum algorithms that leverage physical device properties, take advantage of strengths such as advanced gates, and fit within device limitations

• Advance hardware-algorithm co-design approaches, providing insight in the design of physical devices through co-design with algorithm advances

• Develop new methods for physical or algorithm approximations that expand the reach of quantum approaches for computational challenges of interest to NASA and partners

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov 8

Quantum Computing Requirements (2/4) Hardware and algorithms benchmarking research

• Design advanced performance metrics, and metrics of quantumness, suitable for benchmarking quantum algorithms and quantum hardware

• Generate application-related instances, including parametrized ensembles of instances, to benchmark algorithms and hardware, and combinations thereof

• Evaluate algorithms, hardware, and error mitigation approaches through run results on partner digital QPUs, on analog quantum hardware including but not limited to quantum annealers, and in simulation, evaluating against performance metrics

• Provide comparison with state-of-the-art classical approaches, and advance the state-of-the-art classical approaches, including classical approaches making use of novel classical hardware

• Develop theoretical models and innovative procedures for benchmarking physical gates, logical gates, and common subroutines on quantum processors

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov 9

Quantum Computing Requirements (3/4) Modeling and simulation research

• Develop theoretical models and innovative approaches to numerical simulation of realistic noise on hardware and approaches to mitigating the resulting errors within quantum algorithms

• Estimate resources (number of physical and logical qubits and gates, circuit depth, precision, runtime, classical processing, etc.) needed to run quantum algorithm for application-scale problems on future architectures

• Extend theoretical models of noise statistics for quantum device characterization and benchmarking experiments

Compilation and error mitigation research

• Compile quantum algorithms to efficient circuits that match the architecture of existing and proposed quantum processing units (QPUs) and to larger architectures composed of multiple network quantum and classical processors

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov 10

Quantum Computing Requirements (4/4) Workforce development

• Support NASA, partners, and the U.S. in developing the quantum computing workforce through seminars, journal clubs, and research talks Technical evaluation

• Support NASA in evaluations of proposed technical approaches and results, including evaluating technical reports and published papers

• Support NASA in technical evaluation of potential partner capabilities in the Quantum

Information Science (QIS) field Technical proposal development

• Support NASA in technical proposal development for internal and external opportunities Research and outreach support

• Develop automated processes and encourage tool adoption to make reporting, external visibility (e.g., webpages), and team coordination more effective and efficient

• Provide technical support for NASA sponsored quantum conferences

• Support NASA to conduct outreach to the academic community, promoting the work of

NASA Quantum Computing

Quantum Computing Going Forward

LAUNCH

COMPLE

X

GLOBAL

CIRCULATI

ON

Understanding and harnessing the power of quantum computing is a formidable challenge

• New insights into physics, mathematics, and computer science

• Breakthroughs in engineering design to produce robust, reliable, scalable technologies

• Innovative algorithms and tools in computer and computational science

• Diverse workforce with appropriate expertise to leverage this new paradigm

Exciting future in quantum computing ahead of us

• Small-scale universal quantum chips becoming available

• New and better quantum algorithms, particularly quantum heuristics, emerging

• Compilation and performance capabilities improving rapidly

ENIAC (1946), the first “general-purpose” computer

The task of taking a problem and mapping it onto the machine was complex, and usually took weeks. After the program was figured out on paper, the process of getting the program "into" ENIAC by manipulating its switches and cables took additional days. This was followed by a period of verification and debugging […] (source: http://en.wikipedia.org/wiki/ENIAC) http://en.wikipedia.org/wiki/ENIAC

Backup

LAUNCH

COMPLE

X

GLOBAL

CIRCULATI

ON

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n swww.nasa.gov 13

NASA Quantum Artificial Intelligence Laboratory (QuAIL) Background

• Explores the potential of quantum computing for computational challenges arising in future agency missions

• Partners with domain experts across NASA, performs evaluations of emerging quantum hardware through partnerships with quantum hardware developers, and investigates quantum mechanisms that may be used to provide a quantum advantage

• Key goal is to advance the understanding of quantum computing’s potential to impact challenging computation problems for NASA missions in aeronautics, space and the earth sciences

• Work involves designing and evaluating novel quantum and hybrid quantum classical algorithms, innovative approaches to error mitigation, compilation, and noise characterization, algorithm-hardware co-design with partners, resource estimation, and advanced performance benchmarking and techniques for device characterization

New Era for Quantum Computing

… but so far only for a toy problem 😡😡

• Quantum hardware currently too small for solving practical problems intractable on classical supercomputers

• These devices need to scale up and become more reliable

So what to do in the interim?

• Unprecedented opportunity to invent, explore, and evaluate quantum algorithms empirically

At NASA

• Develop quantum and hybrid quantum-classical applications for computational challenges in aerospace and Earth science

• Algorithms and applications to enable safer, more ambitious, and greater time- and energy-efficient missions

Quantum supremacy achieved 😀😀

• Perform computations not possible on even largest supercomputers in reasonable time

• Google – NASA – ORNL collaboration

F. Arute et al. (2019), Quantum supremacy using a programmable superconducting processor, Nature

574, 505-510

Simulator for this work (qFlex) open sourced via https://github.com/ngnrsaa/qflex

More general simulator (HybridQ) open sourced via https://github.com/nasa/hybridq https://github.com/ngnrsaa/qflex

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n sBiswas, 25-27 May 2022

Diversity of Quantum Computing Hardware

General Purpose:

Universal quantum processors

Google Rigetti

Special Purpose:

D-Wave

Cold atomsQuantum annealers

ColdQuanta

Variety of physical implementations

• Superconducting: Alibaba, D-Wave, Google, IBM, Rigetti

• Ion-Trap: Honeywell, IonQ

• Photonic: PsiQuantum, Quandela, QuiX, Xanadu

• Cold Atom: ColdQuanta [on Int’l Space Station]

• Doped Silicon: Silicon Quantum Computing

• Anyons: Microsoft

NASA quantum computing R&D at QuAIL

• Not developing quantum hardware

• But engaged with variety of key players by providing important insights

• And thereby leverage hardware advances to explore applications of interest to NASA

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n sBiswas, 25-27 May 2022

Current status of quantum algorithms

A handful of proven limitations on quantum computing

Conjecture: Quantum Heuristics will significantly broaden applications of quantum computing

Quantum computing can do everything a classical computer can do

Provable quantum advantage known for a few dozen quantum algorithms and

Unknown quantum advantage for everything else

Status of classical algorithms

• Provable bounds hard to obtain

– Analysis is just too difficult

• Best classical algorithm not known for most problems

• Empirical evaluation required

• Ongoing development of classical heuristic approaches

– Analyzed empirically: ran and see what happens

– E.g. SAT, planning, machine learning, etc.

competitions

QuAIL Innovation in Quantum Algorithms Quantum Approximate Optimization Algorithm (QAOA) (Hadfield, et al.)

• Generalization of Farhi et al. Quantum

Approximate Optimization Algorithm framework inspired by use cases and hardware compilation considerations

• Supports problems with hard constraints, restricting evolution to feasible subspace

• Introduced more general ”mixing operators”:

• Bitflip mixers

• XY mixers

• Controlled Bitflip mixers

• Controlled XY mixers

• Permutation mixers

Quantum-accelerated constraint programming (Booth, et al.)

• Constraint programming (CP) is a paradigm used to model and solve constraint satisfaction and combinatorial optimization problems

• Show how quantum algorithms, can accelerate CP, at both the levels of inference and search

• Introduce a quantum-accelerated filtering algorithm for the alldifferent global constraint

• Propose frameworks for the integration of quantum filtering algorithms within both classical and quantum backtracking search schemes

S. Hadfield et al. (2019), From the quantum approximate optimization algorithm to a quantum alternating operator Ansatz, Algorithms 12, 34 – recipient of Algorithms 2020 Best Paper Award

K.E.C. Booth et al. (2021), Quantum-accelerated constraint programming, Quantum, 5, pp. 550

T h e c o r n e r s t o n e o f N A S A ’ s c u r r e n t a n d f u t u r e m i s s i o n s 18

Example: Hardware-Algorithm Co-Design, Gate-Model Quantum Alternating Operator Ansatz introduced more general “mixing operators”

S. Hadfield et al. (2019), From the quantum approximate optimization algorithm to a quantum alternating operator Ansatz, Algorithms 12, 34 Z. Wang et al. (2020), XY-mixers: analytical and numerical results for QAOA pausing, Phys. Rev. A 101, 012320 M. Streif el al. (2021), Quantum algorithms with local particle number conservation: Noise effects and error correction, PRA

Approximation ratio for 3-coloring a triangle: QAOA with standard X-mixer (left) and with XY-mixer (right)

Problem instance:

Max-K-Colorable subgraph

Collaboration with Rigetti under DARPA ONISQ project to evaluate QAOA on their gate-model processors

Rigetti developed novel calibration techniques for this family of gates inspired native implementations of these general mixing operators on Rigetti hardware

Example: Hardware-Alg. Co-Design: Q. Annealing

J. Marshall et al. (2019), Power of pausing: Advancing understanding of thermalization in experimental quantum annealers, Phys. Rev. Applied 11, 044083 Z. Izquierdo et al. (2021), Ferromagnetically shifting the power of pausing, Phys. Rev. Applied 15, 044013

The power of pausing: additional control of anneal schedule, increases performance by orders of magnitude

Heat map of probability of solution, depending on pause location and pause lengthAnneal time regions. Purple region is where pause is expected to be effective.

An appropriately placed pause can improve TTS as well as the probability of success

- on embedded instances of application interest

- consistent region in which pausing helps

Quantum Computing for Aeronautics

Quantum annealing approaches to de-conflict optimal trajectories for air traffic management

T. Stollenwerk et al. (2020), Quantum annealing applied to de-conflicting optimal trajectories for air traffic management, IEEE Trans. Intell. Transp. Syst. 21, 285-

Current efforts

• Quantum algorithms for material simulation with applications to batteries & energy storage

Feasibility of QIS to address cybersecurity of communication availability for air traffic of the future

First-ever demo incorporating solutions from a quantum system into an UAV flight test

N. Cramer et al. (2020), Enhanced UAS availability via vehicle-to-vehicle routing scaled experiments, AIAA Aviation Forum, 2020-2850

Z. Izquierdo et al. (2022), The Advantage of pausing:

parameter setting for quantum annealers (to appear)

Quantum algorithms to optimize information sharing in bandwidth-limited autonomous settings

Quantum Computing for Earth Science

Quantum-assisted associative adversarial network

N. Gao et al. (2020), High-dimensional similarity search with quantum-assisted variational autoencoder, KDD ‘20, 956-964

Quantum-assisted variational autoencoder

• “Similarity” search applied to NASA’s MODIS dataset

• Multi-variate time series of land cover change

• 17 years of satellite observations over 36 spectral bands

• Extend to segmentation, reconstruction, super-resolution

Optimizing quantum heuristics with meta-learning

M. Wilson et al. (2019), Quantum-assisted associative adversarial network: Applying quantum annealing in deep learning, arXiv:1904.10573

M. Wilson et al. (2019), Optimizing quantum heuristics with meta-learning, arXiv:1908.03185

Quantum Computing for Space Exploration

Quantum computing for mission planning and scheduling

E. Rieffel et al. (2015), A case study in programming a quantum annealer for hard operational planning problems, Quantum Inf. Process. 14, 1-36 T. Tran et al. (2016), Explorations of quantum-classical approaches to scheduling a Mars lander activity problem, Workshops of 30th AAAI Conf. on AI Planning for Hybrid Systems

Quantum networks to provide distributed quantum resources via entanglement swapping

• Vision of a space-ground quantum network

• Applications to long baseline telescopes, sensor arrays, and distributed quantum computing

Quantum Tools Development

Use subsystem codes to attain less resource-intensive error mitigation [Jia17]

HPC simulation of quantum circuits

Temporal planning approaches to compiling quantum algorithms [Ven18] Open quantum system simulation [Mar20]

State-of-the-art quantum circuit simulator used to support quantum supremacy demonstration

• Open source qFlex code:

https://github.com/ngnrsaa/qflex

• Open source HybridQ [MAN21] code:

https://github.com/nasa/hybridq

[Jia17] Z. Jiang and E.G. Rieffel, (2017), Non-commuting two-local Hamiltonians for quantum error suppression, Quantum Information Processing [Ven] D. Venturell et al., (2018) Compiling quantum circuits to realistic hardware architectures using temporal planners, Quantum Science and Technology, 3(2), 025004.

[Hon22] H.Y. Hu et al., (2022), Local shadow tomography: Efficient estimation of error-mitigated observables, arXiv:2203.07263 [Man21] S. Mandrà, et al., (2021) HybridQ: A Hybrid Simulator for Quantum Circuits, arXiv:2111.06868 [Mar20] J. Marshall, et al. (2020) Characterizing local noise in QAOA circuits, IOP SciNotes 1, 025208

Local shadow tomography used to estimate error-mitigated expectation values on noisy quantum computers [Hon22]

HybridQ: A Hybrid Quantum Simulator for Large Scale Simulations

Hardware agnostic quantum simulator, designed to simulate large scale quantum circuits.

Can run tensor contraction simulations, direct evolution simulation and Clifford+T simulations using the same syntax Features:

Fully compatible with Python (3.8+) Low-level optimization achieved by using C++ and Just-In-Time (JIT) compilation with JAX and Numba, It can run seamlessly on CPU/GPU and TPU, either on single or multiple nodes (MPI) for large scale simulations, using the exact same syntax User-friendly interface with an advanced language to describe circuits and gates, including tools to manipulate/simplify circuits.

Recent Improvements:

Commutations rules are used to simplify circuits (useful for QAOA) Expansion of density matrices as superpositions of Pauli strings accepts arbitrary non-Clifford gates, Open-source (soon!) project with continuous-integration, multiple tests and easy installation using either pip or conda

Open source code available at https://github.com/nasa/HybridQ S. Mandrà, J. Marshall, E. G. Rieffel, R. Biswas, HybridQ: A Hybrid Simulator for Quantum Circuits, arXiv:2111.06868