ProprietaryandConfidential,D-WaveSystemsInc.

Software,CloudServicesandApplications

Copyright©D-WaveSystemsInc. 2

SoftwareDirections

• AlanBaratz newSVPSoftware&Apps

• Newsoftwaretoolsarchitecture– OCEAN

• Moreaccesstosystematmultiplelevels• Microkernel

• Intermediaterepresentations– QUBO

• SubjectMatterExperttools

• ImprovedCloudAccess• 3phaseimplementationstartinginJune

• “Quadrant”MachineLearningSoftware

• Conventionalimplementationtostart

ProprietaryandConfidential,D-WaveSystemsInc. 3

MajorInitiatives

• OceanToolsSuite– PhaseddeliverybeginningDecember,2017– Centraltocloudservice

• CloudServiceBusiness– Goal:Fasteraccessandadeveloperecosystem– Currentlyin4thof8sprints;Alphatestingunderway

• QuadrantMLApplicationsBusiness– Goal:LeverageSOTAgenerativeMLalgorithmstoengagethecommunity– FocusonCancerindicators/treatments,5GSDNs,imagedefectrecognition

• CreativeDestructionLabs(CDL)Participation– FocusedonattractingquantumorientedstartupstoD-Waveplatforms– FourtoptiercompaniesleveragingD-Wave

ProprietaryandConfidential,D-WaveSystemsInc. 4

Ocean1.0ToolsSuite

• GraphMappingandConstraintToolsChains• BeingIntegratedIntoCloudService• InvestigatingAdditionalProblemDecompositionApproaches

ProprietaryandConfidential,D-WaveSystemsInc. 5

ExampleUsage(Python)

importnetworkx asnximportdwave_networkx asdnximportdwave_micro_client_dimod asmicroimportdwave_qbsolv

urlc ='https://cloud.dwavesys.com/sapi'tokenc ='SE-bb7f104b4a99cf9a10eeb9637f0806761c9fcedc'solver_namec ='DW_2000Q_1'

structured_samplerc =micro.DWaveSampler(solver_namec,urlc,tokenc)samplerc =micro.EmbeddingComposite(structured_samplerc)samplerq =dwave_qbsolv.QBSolv()

cloudsi =dnx.structural_imbalance(G,samplerc,num_reads=10000)qbsolvsi =dnx.structural_imbalance(G,samplerq,solver=samplerc)

h={v:node_values[v]forvinG.nodes}J={(u,v):eval foru,vinG.edges}response=samplerc.sample_ising(h,J,num_reads=10000)

ProprietaryandConfidential,D-WaveSystemsInc. 6

CloudServiceBusiness

ServiceComponents

DemosOnlineTraining

CommunitySupport

PaidSupport

KnowledgeBase ProblemStatus

AccountManagement

Open-SourceTools

• NewUI/UXDesign&DevelopmentUnderway• Jupyter HubSelectedforO/LTrainingandProcessing

• Initialnotebookscomplete• SalesforceSelectedforCRM;ZenDesk forSupportandCommunity

• Configurationandintegrationunderway• NewQPUJobSchedulingModelBeingDeveloped

ProprietaryandConfidential,D-WaveSystemsInc. 7

QuadrantMLApplicationsBusiness

• Goal:LeverageSOTAGenerativeMLAlgorithmsforCommunityEngagementandMLApplicationOfferings

• EarlySuccessWorkingwithCustomerstoDevelopApplications– Siemens:UsedCRF-NNtoaccuratelyidentifymedicalinstrumentsinvideos– Huawei:UsedDiscreteDensityEstimationtodetectsleepingcelltowers– ProposaltoNIHforVAEtoidentifygenemarkersforbraindiseases

• LaunchofQuadrant.ai– Pressreleases,website,marketingcampaign

ProprietaryandConfidential,D-WaveSystemsInc. 8

KeyCDLCompanies

• OTILumionics– CreationofnewOLEDmaterials– LeveragingD-Waveforelectronicconfigurationcalculations

• SolidStateAI– ImprovingsemiconductorFAByieldandequipmentfailureprediction– LeveragingD-Waveforstrongclassifierdetermination

• ProteinQure– Newdrugdiscovery– LeveragingD-Wavefor3Dproteinfolding

• AdaptiveFinance– Highyieldequitity trading– LeveragingD-Waveforstrongclassifierdetermination

Copyright©D-WaveSystemsInc. 9

Mission

Tohelpsolvethemostchallengingproblemsinthemultiverse:

• Optimization

• MachineLearning

• MonteCarlo/Sampling

• MaterialScience

Copyright©D-WaveSystemsInc. 10

CustomerApplicationAreas

• Lockheed/USCISI

– SoftwareVerificationandValidation

– Optimization– Aeronautics

– PerformanceCharacterization&Physics

• Google/NASAAmes/USRA

– MachineLearning

– Optimization

– PerformanceCharacterization&Physics

– Research

• LosAlamosNationalLaboratory

– Optimization

– MachineLearning,Sampling

– SoftwareStack

– SimulatingQuantumSystems

– Other(good)Ideas

• CS- 1

– Cybersecurity

• OakRidgeNationalLaboratory

– SimilartoLosAlamos

– MaterialScience&Chemistry

Los Alamos National Laboratory

D-Wave “Rapid Response” Projects (Stephan Eidenbenz, ISTI)

Round 1 (June 2016)1. Accelerating Deep Learning with

Quantum Annealing

2. Constrained Shortest Path Estimation3. D-Wave Quantum Computer as an

Efficient Classical Sampler

4. Efficient Combinatorial Optimization using Quantum Computing

5. Functional Topological Particle Padding6. gms2q—Translation of B-QCQP to

D-Wave7. Graph Partitioning using the D-Wave for

Electronic Structure Problems8. Ising Simulations on the D-Wave QPU9. Inferring Sparse Representations for

Object Classification using the Quantum D-Wave 2X machine

10. Quantum Uncertainty Quantification for Physical Models using ToQ.jl

11. Phylogenetics calculations

Round 2 (December 2016)1. Preprocessing Methods for Scalable Quantum Annealing2. QA Approaches to Graph Partitioning for Electronic

Structure Problems3. Combinatorial Blind Source Separation Using “Ising”4. Rigorous Comparison of “Ising” to Established B-QP

Solution Methods

Round 3 (January 2017)1. The Cost of Embedding2. Beyond Pairwise Ising Models in D-Wave: Searching for

Hidden Multi-Body Interactions3. Leveraging “Ising” for Random Number Generation4. Quantum Interaction of Few Particle Systems Mediated

by Photons5. Simulations of Non-local-Spin Interaction in Atomic

Magnetometers on “Ising”6. Connecting “Ising” to Bayesian Inference Image Analysis7. Characterizing Structural Uncertainty in Models of

Complex Systems8. Using “Ising” to Explore the Formation of Global Terrorist

Networks

Los Alamos National Laboratory 6/27/2017

2016 2017 %CombinatorialOptimization 5 5 10 45%MachineLearning,Sampling 2 2 4 18%UnderstandingDevicePhysics 2 1 3 14%SoftwareStack/Embeddings 1 1 2 9%SimulatingQuantumSystems 2 2 9%Other(good)Ideas 1 1 5%Total 11 11 22 100%

UseCaseTotal

The LANL Rapid Response Project results for 2016 and 2017 are available as PDF’s at:http://www.lanl.gov/projects/national-security-education-center/information-science-technology/dwave/index.php

UNCLASSIFIED Nov. 13, 2017 | 13

ISTI Rapid Response DWave Project (Dan O’Malley):Nonnegative/Binary Matrix Factorization

Imag

e cr

edit:

Lee

&

Seun

g, N

atur

e (1

999)§ Low-rank matrix factorization

• 𝐴 ≈ 𝐵𝐶 where 𝐵%,' ≥ 0 and 𝐶%,' ∈ {0,1}

•

𝐴

≈

𝐵

𝐶

§ Unsupervised machine-learning application• Learn to represent a face as a linear combination of basis images

• Goal is for basis images to correspond to intuitive notions of parts of faces

§ “Alternating least squares”1. Randomly generate a binary 𝐶

2. Solve 𝐵 = argmin6 ∥ 𝐴 − 𝑋𝐶 ∥: classically

3. Solve 𝐶 = argmin6 ∥ 𝐴 − 𝐵𝑋 ∥: on the D-Wave

4. Repeat from step 2

§ Results• The D-Wave NMF approach results in a sparser 𝐶 (85% vs. 13%) and denser but more lossy

compression than the classical NMF approach

• The D-Wave outperforms two state-of-the-art classical codes in a cumulative time-to-target benchmark when a low-to-moderate number of samples are used

UNCLASSIFIED Nov. 13, 2017 | 14

ISTI Rapid Response DWave Project (Hristo Djidev):Efficient Combinatorial Optimization

§ Objectives• Develop D-Wave (DW) algorithms for NP-hard

problemsFocus: the max clique (MC) problem:

• Study scalability/accuracy issues and ways to mitigate them

• Characterize problem instances for which D-Wave may outperform classical alternatives

§ Results• The MC problem can be solved accurately

and fast on DWo but so can classical methodso no quantum advantage for typical

problem instances fitting DW (of size ~45)

• Running on larger (Chimera-like) graphso Chimera graph is

modified by merging a set of randomlyselected edges intoa vertex

o Resulting graphs ofsizes upto 1000 are used as inputs to MC problem

o DW beats simulated annealing, its classical analogue, by a factor of more than 106

o In order to see a quantum advantage for the MC problem, graph sizes should be > 300

o In order to see a quantum advantage for the MC problem, graph sizes should be > 300

orig.graph cliqueofmaxsize prob edges: 3068, Energy prob upper bound: 3930.5

-2 2 Vertices

-1 1 Couplers

Solution : -1 +1

200 400 600 800 1000

−50

−40

−30

−20

−10

0

Graph size

best

cliq

ue s

ize fo

und,

rela

tive

to D

wave

2X

Dwave 2XPPHaSA−clique fastSA−clique slowSA−isingfmc

0 200 400 600 800 1000

1e−0

21e

+00

1e+0

21e

+04

1e+0

6

Graph size

spee

dup

Quality comparison Speed comparison

UNCLASSIFIED Nov. 13, 2017 | 15

ISTI Rapid Response DWave Project (Sue Mniszewski): Quantum Annealing Approaches to Graph Partitioning for Electronic Structure Problems

§ Motivated by graph-based methods for quantum molecular dynamics (QMD) simulations

§ Explored graph partitioning/clustering methods formulated as QUBOs on D-Wave 2X

§ Used sapi and hybrid classical-quantum qbsolv software tools

§ Comparison with state-of-the-art tools§ High-quality results on benchmark (Walshaw),

random, and electronic structure graphsGraph N Best METIS KaHIP qbsolv

Add20 2395 596 723 613 602

3elt 4720 90 91 90 90

Bcsstk33 8738 10171 10244 10171 10171Minimize edge counts between 2 parts on Walshaw graphs.

k-Concurrent Partitioning for Phenyl Dendrimer.

k-parts METIS qbsolv2 705 706

4 20876 2648

8 22371 15922

16 28666 26003

k-Concurrent clustering for IGPS Protein Structure:resulting 4 communities share common sub-structure. Comparable to classical methods.

UNCLASSIFIED Nov. 13, 2017 | 16

ISTI Rapid Response Project (Carleton Coffrin):Challenges and Successes of Solving Binary Quadratic Programming Benchmarks on the DW2X QPU

§ Looking to the Future• I have drunk D-Wave Kool-Aid• RAN1 convinced me that the DW2X has huge potential• I believe, in the next 5 years, QPU’s will be very disruptive to

optimization research

Biswas, SMC-IT, 28 Sept 2017

Quantum Computing for NASA Applications

17

Data Analysisand Data Fusion

AirTrafficManagement

Mission Planning, Scheduling, and Coordination

V&V and Optimal Sensor

Placement

Topologically-aware Parallel

Computing

Anomaly Detection and

Decision Making

Common Feature: Intractable problems on classical supercomputers

Objective: Find “better” solution• Faster• More precise• Not found by classical algorithm

Current NASA Research in Annealing Applications

Machine Learning

Graph-based Fault Detection

Complex Planning and Scheduling

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

1617

18

19

20

21

22

23

24

25

26

27

28

29

30

31

3233

34

35

36

37

38

39 4041 42

43

44

45

46

• General Planning Problems (e.g., navigation, scheduling, asset allocation) can be solved on a quantum annealer (such as D-Wave)

• Developed a quantum solver for Job Shop Scheduling that pre-characterizes instance ensembles to design optimal embedding and run strategy – tested at small scale (6x6) but potentially could solve intractable problems (15x15) with 10x more qubits

• Analyzed simple graphs of Electrical Power Networks to find the most probable cause of multiple faults – easy and scalable QUBO mapping, but good parameter setting (e.g., gauge selection) key to finding optimal solution –now exploring digital circuit Fault Diagnostics

• Boltzmann sampling commonly used in Machine Learning, particularly Deep Learning. Quantum computing has provable advantage for some sampling problems. Demonstrated learning when using a QA as a Boltzmann sampler.

01 0 11 1 10 01 0 0 01 0 0

CircuitBreakers

Sensors

Observations

IN: configs.

OUT: params.

QA {J , h}

D-Wave run results: established baseline performance for QA on these applications

Scheduling Applications

Job-Shop scheduling: Complete quantum-classical solver framework with pre-

processing, compilation/run strategies, decomposition methods

D. Venturelli, D. J.J. Marchand, G. Rojo, Quantum Annealing Implementation of Job Shop Scheduling, arXiv:1506.08479

Eleanor G. Rieffel, Davide Venturelli, Minh Do, Itay Hen, Jeremy Frank, ParametrizedFamilies of Hard Planning Problems from Phase Transitions, AAAI-14.E. G. Rieffel, D. Venturelli, B. O'Gorman, M. B. Do, E. Prystay, V.N. Smelyanskiy, A case study in programming a quantum annealer for hard operational planning problems, Q. Information Processing, 14, (2014)

Comparison with state-of-the-art application-specific algorithms:current best planners

Scheduling problems as testbed for resource-bounded tailored embedding methods

10 20 30 40 50

Problem size n: number of tasks

Med

ian

Run

time

[sec

]

0.01

0.1

1

10

100

1000

10,000

Planner Comparison: All Scheduling Problems

FF:α=1.11 ± 0.061LPG: α=0.69 ± 0.139M: α=0.1 ± 0.007Mp: α=0.54 ± 0.035

Solved problems with 6 machines and 6 jobs: analyzed scaling of tractability

Graph coloring

Mars Lander activity scheduling

Airport runwayscheduling

• T. Tran, M. Do, E. Rieffel, J. Frank, Z. Wang, B. O'Gorman, D. Venturelli, J. Beck, A Hybrid Quantum-Classical Approach to Solving Scheduling Problems, SOCS’16

• T. Tran, Z. Wang, M. Do, E. Rieffel, J. Frank, B. O'Gorman, D. Venturelli, J. Beck, Explorations of Quantum-Classical Approaches to Scheduling a Mars Lander Activity Problem, Workshops AAAI’16

QA-guided tree search

A. Perdomo-Ortiz et al., On the readiness of quantum optimization machines for industrial applications arXiv:1708.09780

Fault DiagnosisFirst comprehensive study addressing the readiness of

quantum annealing for real-world applicationsSix different algorithms (SA, PT-ICM, QMC, SAFARI, SAT-based, and DWave2X)In all three problem Hamiltonian representations (PUBO, QUBO, Chimera)

Addressed future quantum annealer design for quantum advantage in applications with practical relevance

• What is the impact of higher-order terms? • Need for non-stoquastic Hamiltonians? • Impact of connectivity? …

Compresseddata

i~d⇢✓(t)dt

= [H✓, ⇢✓]i~d⇢✓(t)

dt= [H✓, ⇢✓]

✓

Rawinputdata

Quantumsampling

Measurement

Hiddenlayers

ClassicalgenerationorreconstructionofdataIn

ference

Qua

ntum

processing

Classicalpre-a

ndpost-p

rocessing

Visibleunits Hiddenunits Qubits

Trainingsamples

Generatedsamples

• Hybridproposalthatworksdirectlyonalow-dimensionalrepresentationofthedata.Newparadigm:UsedeeplearningtoassistQMLimplementationinnear-termQC

Anear-termapproachforquantum-enhancedmachinelearning

Benedetti,Realpe-Gomez,andPerdomo-Ortiz.Quantum-assistedHelmholtzmachines:Aquantum-classicaldeeplearningframeworkforindustrialdatasetsinnear-termdevices.arXiv:1708.09784 (2017).

Feasibility study: Using quantum-classical hybrids to assure the availability of the UAS Traffic Management (UTM) network

against communication disruptions

Kopardekar, P., Rios, J., et. al., Unmanned Aircraft System Traffic Management (UTM) Concept of Operations, DASC 2016

Future • Higher vehicle density• Heterogeneous air vehicles• Mixed equipage• Greater autonomy• More vulnerability to

communications disruptions

Newly funded effort in aeronautics

Explore quantum approaches to• Robust network design• Track and locate of a moving jammer • Secure communication of codes

supporting anti-jamming protocols

30 month effort: harness the power of quantum computing and communication to address the cybersecurity challenge of availability

Joint with NASA Glenn, who are working on QKD for spread spectrum codes

Prior work (NASA-DLR collaboration): T. Stollenwerk et al., Quantum Annealing Applied to De-Conflicting Optimal Trajectories for Air Traffic Management

Copyright©D-WaveSystemsInc. 23

Mission

Tohelpsolvethemostchallengingproblemsinthemultiverse:

• Optimization

• MachineLearning

• MonteCarlo/Sampling

• MaterialScience

Quantum Computing at Volkswagen:Traffic Flow Optimization using the D-Wave Quantum Annealer

D-Wave Users Group Meeting - National Harbour, MD 27.09.2017 – Dr. Gabriele Compostella

The Question that drove us …

27.09.2017 K-SI/LD | Dr. Gabriele Compostella 25

Is there a real-world problem that could be addressed with a

Quantum Computer?

YES: Traffic flow optimisation

Everybody knows traffic (jam) and normally nobody likes it.Image courtesy of think4photop at FreeDigitalPhotos.net

27.09.2017 K-SI/LD | Dr. Gabriele Compostella 26

Public data set: T-Drive trajectory

https://www.microsoft.com/en-us/research/publication/t-drive-trajectory-data-sample/

Beijing• ~ 10.000 Taxis• 2.2. – 8.2.2008

data example:

27.09.2017 K-SI/LD | Dr. Gabriele Compostella 27

Result: unoptimised vs optimised traffic

27.09.2017 K-SI/LD | Dr. Gabriele Compostella 28

Volkswagen Quantum Computing in the news

27.09.2017 K-SI/LD | Dr. Gabriele Compostella 29

27.09.2017 K-SI/LD | Dr. Gabriele Compostella 30

HETEROGENEOUS QUANTUM COMPUTING FOR SATELLITE OPTIMIZATIONGID EON BAS S

BOOZ AL L EN HAM ILTON

September 2017

BO O Z AL L EN • DI G I T A L

Booz Allen Hamilton Restricted, Client Proprietary, and Business Confidential.

Traveling Salesman

Vehicle Routing

Logistics

Circuit Design

Network Design

Manufacturing

Machine Learning

Artificial Intelligence

Robotics

System DesignOptimization

Combinatorial Chemistry

Drug Discovery

QUANTUMANNEALING HAS MANY REAL-WORLDAPPLICATIONS

BO O Z AL L EN • DI G I T A L

Booz Allen Hamilton Restricted, Client Proprietary, and Business Confidential.

CONCLUSIONS

18

+ As problems and datasets grow, modern computing systems have had to scale with them. Quantum computing offers a totally new and potentially disruptive computing paradigm.

+ For problems like this satellite optimization problem, heterogeneous quantum techniques will be required to solve the problem at larger scales.

+ Preliminary results on this problem using heterogeneous classical/quantum solutions are very promising.

+ Exploratory studies in this area have the potential tobreak new ground as one of the first applications ofquantum computing to a real-world problem

Display Advertising Optimization by

Quantum Annealing Processor

Shinichi Takayanagi*, Kotaro Tanahashi*, Shu Tanaka†*Recruit Communications Co., Ltd.

† Waseda University, JST PRESTO

（C）Recruit Communications Co., Ltd.

Behind the Scenes

35

DSPSSP

RTB

AdvertiserPublisher

Impression

SSP: Supply-Side PlatformDSP: Demand-Side PlatformRTB: Real Time Bidding

AD

AD

AD

1.0$

0.9$

0.7$

Winner!

（C）Recruit Communications Co., Ltd.

CTR Prediction with Machine-Learning

• Machine-Learning (ML) tech is often used for CTR prediction

• ML has succeeded in this field

36

Click F1 F2 F3

1 M 01 2.13

0 F 07 2.12

0 F 23 4.2

? F 99 1.2

Click or NotPrediction

Matrix expressionUsers

Model

(Click-through-rate)

（C）Recruit Communications Co., Ltd.

0

4

8

12

16

0 6 12 18

Budget Pacing

• Budget pacing is also important

• Control of budget pacing helps

advertisers to…

– Reach a wider range of audience

– Avoid a premature campaign stop / overspending

37

Too fast

Budget pacing controlled

Budg

et

spendi

ng

Target budget

Time (hours)

（C）Recruit Communications Co., Ltd.

4. Summary

• Budget pacing is important for display advertising• Formulate the problem as QUBO• Use D-Wave 2X to solve budget pacing control

optimization problem• Quantum annealing finds a better solution than the

greedy method.

38

Copyright©D-WaveSystemsInc. 39

DENSOOptimizationProjects

VideosfromCES,LasVegas,January2018

AutonomousDrivinghttps://www.youtube.com/watch?v=Bx9GLH_GklA

FactoryOptimizationhttps://www.youtube.com/watch?v=BkowVxTn6EU

Copyright©D-WaveSystemsInc. 40

Mission

Tohelpsolvethemostchallengingproblemsinthemultiverse:

• Optimization

• MachineLearning

• MonteCarlo/Sampling

• MaterialScience

41 billings7893

ORNL is managed by UT-Battelle for the US Department of Energy

42 Presentation_name

Adiabatic Quantum Programming at ORNL: Workflow Environments and HPC Integration APIs

The First:

There are currently 3 main challenges in Deep Learning

A Study of Complex Deep Learning Networks on High Performance, Neuromorphic, and Quantum Computers

QuantumMachineLearningforElection Modelling

Election2016:Casestudyinthedifficultlyof sampling

Wheredidthemodelsgo wrong?

QuantumMachineLearningforElectionModelling– MaxHenderson, 2017 44

Forecastingelectionsonaquantum computer

QuantumMachineLearningforElectionModelling– MaxHenderson, 2017 45

• Quantumcomputingresearchhasshownpotentialbenefits(speedups)intrainingvariousdeepneural networks1-3

• Coreidea:UseQC-trainedmodelstosimulateelectionresults.Potential benefits:

• Moreefficientsampling/ training• Intrinsic,tuneablestate correlations• Inclusionofadditionalerror models

1. Adachi, Steven H., and Maxwell P. Henderson. "Application of quantum annealing totraining of deep neural networks." arXiv preprint arXiv:1510.06356 (2015).2. Benedetti,Marcello,etal."Estimationofeffectivetemperaturesinquantumannealersforsamplingapplications:Acasestudywithpossibleapplicationsindeep

learning."PhysicalReviewA94.2(2016): 022308.3. Benedetti, Marcello, et al. "Quantum-assisted learning of graphical models with arbitrary pairwise connectivity." arXiv preprint arXiv:1609.02542 (2016).

Summary

QuantumMachineLearningforElectionModelling– MaxHenderson, 2017 46

• TheQC-trainednetworkswereabletolearnstructureinpollingdatatomakeelectionforecastsinlinewiththemodelsof 538

• Additionally,theQC-trainednetworksgaveTrumpamuchhigherlikelihoodofvictoryoverall,eventhoughthestate’sfirstordermomentsremained unchanged

• Ideallyinthefuture,wecouldrerunthismethodusingcorrelationsknownwithmoredetailin-housefor 538

• Finally,theQC-trainednetworkstrainedquickly,andsinceeachmeasurementisasimulation,eachiterationofthetrainingmodelproduced25,000simulations(oneforeachnationalerrormodel),whichalreadyeclipsesthe20,000simulations538performseachtimetheyreruntheir models

COPYRIGHT 2016 LOCKHEED MARTIN CORPORATION – ALL RIGHTS RESERVED47

Quantum Enabled Machine Learning

Supervised Learning: Improving Neural Network Training

Adachi, Steven H., and Maxwell P. Henderson. "Application of Quantum Annealing to Training of Deep Neural Networks." arXiv preprint arXiv:1510.06356 (2015).

visiblelayer

hiddenlayer vi

sibl

e no

de 1

hidden node 1

hidden node 2

hidden node 3

hidden node 4

hidden node 5

hidden node 6

hidden node 7

hidden node 8

visi

ble

node

2

visi

ble

node

3

visi

ble

node

4

visi

ble

node

5

visi

ble

node

6

visi

ble

node

7

visi

ble

node

8

AlejandroPerdomo-OrtizSeniorResearchScientist,QuantumAILab.atNASAAmesResearchCenterandatthe

UniversitySpaceResearchAssociation,USAHonorarySeniorResearchAssociate,ComputerScienceDept.,UCL,UK

NationalHarbor,MD,September28,2017

Opportunitiesandchallengesinquantum-enhancedmachinelearninginnear-termquantumcomputers

QUBITSD-waveUserGroup2017

Funding:

Perdomo-Ortiz,Benedetti,Realpe-Gomez,andBiswas.arXiv:1708.09757 (2017).ToappearintheQuantumScienceandTechnology(QST)invitedspecialissueon“Whatwouldyoudowitha1000qubit device?”

Los Alamos National Laboratory

28-Sep-2016 | 49

Copyright©D-WaveSystemsInc. 50

Mission

Tohelpsolvethemostchallengingproblemsinthemultiverse:

• Optimization

• MachineLearning

• MonteCarlo/Sampling

• MaterialScience

Copyright©D-WaveSystemsInc. 51

Z(2)latticegaugetheoryKosterlitz-Thouless model

3Dtransverse-fieldIsing model

QuantumMaterialScience@D-Wave

R.Harris

A.King E.Dahl

TheFuture,Maybe

• 2018Predictions• Beyond

Copyright©D-WaveSystemsInc. 53

GateModelMachines- 2018

• ~50– 100qubitmodelsrunning• Nolargescaleerrorcorrection• NoisyIntermediateScaleQC’s(NISQ)*• Knowifsomeproblemswillrunwithout

errorcorrection• QuantumMaterialScience?• NoShor’sAlgorithm• Quantum“Supremacy”perhapsfor

syntheticbenchmark• Importanceoferrorcorrectionandpotentialappsbecomesclear*“QuantumComputingintheNISQeraandbeyond”,JohnPreskill,CalTech,arXiv:1801.00862

Copyright©D-WaveSystemsInc. 54

.

• 75– 100“proto-apps”on2000QD-WaveSystem

• ~Halfapproachingclassicalperformanceon

smallishproblems

• DemonstrateQuantumMaterialScience

breakthrough

• Quantum“Advantage”demonstrations

• IARPAQEOandD-Wavehighercoherence

qubitdemonstrations

• Trajectoryto4000-5000qubitsystem,better

connectivity,lowernoise

15mK

QuantumAnnealing- 2018

Copyright©D-WaveSystemsInc. 55

AndBeyond

• BiginvestmentsinQC– China $11B

– EUFlagship $1B+

– UKHubs1&2 <$1B

– Japan ChristmasDaymeeting

• U.S.andCanada– fragmented– 2019U.S.budgetproposal– DOE$100M,NSF$30M,others?

• QuantumDiversity• Moresmartpeopleworkingonappsandsoftwaretools

• Bo’sUnifiedTheoryofQuantumComputing

Copyright©D-WaveSystemsInc. 56

President’sNationalStrategicComputingInitiative

Copyright©D-WaveSystemsInc. 57

AfterNike™

QuantumComputingneedsyouto:

JustDoIt™Probably

Copyright©D-WaveSystemsInc. 58

ForMoreInformationSee

D-WaveUsersGroupPresentations:– https://dwavefederal.com/qubits-2016/– https://dwavefederal.com/qubits-2017/

LANLRapidResponseProjects:– http://www.lanl.gov/projects//national-security-education-center/information-science-technology/dwave/index.php