EVOLVING QUANTUM COMPUTERS: Harnessing a Vast Hidden Reality

• 1 •
Quantum Computing:
Accessing Vast Untapped Resources
Jonthebeach Conference May 2017
Eric Ladizinsky
Co-Founder, Chief Scientist D-Wave Systems Inc.

3 © Copyright 2012-2013 D-Wave Systems Inc.• 3 •
PROBLEMS BEYOND OUR UNDERSTANDING

Staggering Complexity in All Areas of Human Endeavor
Interactions between drugs
and human gene products
Financial Network
Social Network
Energy Distribution Network
Vast Inefficiencies and Potential for System Crashes

Augmenting Human Intelligence
+ +
+
Machine Learning / Neural Nets
Not Enough to solve the
Scale of Problems we face?
Networked Brains Massive Classical Computing

6 © Copyright 2012-2013 D-Wave Systems Inc.6 © Copyright 2007 D-Wave Systems, Inc.
Rapidly approaching atomic
dimensions - natural limit
1.E+00
1.E+04
1.E+08
1.E+12
1.E+16
1.E+20
1.E+24
1.E+28
1.E+32
1.E+36
1.E+40
1 10 100 1000
Scale of Problem
10
40
10
32
10
24
10
16
10
8
1
RequiredNumber
ofCalculations
2N
N5
Intractable
Tractable
Certain Problems Will Remain Beyond
The Reach of any Classic Computers
• Simulation of quantum systems
• Complex combinatorial optimization
 Probabilistic Machine Learning – True AI
• Factoring large numbers - code breaking

Discovering New Ways of Harnessing Nature
Leads to Radical Leaps In Human Capability

8 © Copyright 2012-2013 D-Wave Systems Inc.confidential• 8 •
Quantum Theory: The Explanation of our World
Matter Radiation
The 1st Quantum Revolution

Superposition
Entanglement
Quantum Tunneling
Quantum Effects and the 2nd Quantum Revolution

Hugh Everett – Many Worlds
Feynman
“There’s Plenty of Room at the Bottom”
A Brief History of Quantum Computing
Peter Shor – Shor’s Algorithm

10
0110
0
00010111
1
Qubits
Possibilities Realized
Increases Exponentially
O OOO
Building Quantum Computers from Binary Elements
2n numbers
Factoring a 200 digit number
• classical - 3 billion MIPS-years
• quantum - one second
Shor’s Algorithm

De-Coherence: Why Quantum Computers are Hard to Build
Environmental interactions can pull the different streams
of reality apart destroying their collaboration …

Schrodinger’s Cat Lives! : A Macroscopic QuBit
Quantum Mechanics on a Chip?

Different Physical Platforms and Computing Models
 Superconducting Qubits and
control circuitry are macroscopic
and scale easily using standard
semiconductor techniques
 Adiabatic QC Model more robust
against noise
 Well suited to Optimization
Scalable D-Wave
Architecture
Prototype Lab Systems

The New Paradigm of Innovation:
Small Companies taking on Big Problems
 Single Shared Vision: Purpose Driven
 Do or die – willing to take big risks
 Attracts best in class teams ..
 Fast .. Rapid Prototyping .. Lack of bureaucracy
 Leveraging best available existing resources
 Nimble and innovative
 Lean … creative with limited resources

World’s only Commercial Quantum Computer
Company
What we Make
Cloud ready quantum
computing systems
C$200M capital raised;
Software ecosystem forming;
Other QC efforts adopting
annealing approach
Business Model
Quantum system sales &
quantum cloud services
149+ US patents;
90+ scientific papers;
Ranked 4th in “patent
power” by IEEE
Premier
First Customers
Temporal Defense
Systems

Supported by World Class Investors
International Investment
and Underwriting

Different Approaches to Quantum Computing
•Different models: Gate, One-way, Topological, Adiabatic
(annealing)
• Different basis: Superconducting Circuits, ion traps,
photons, solid state

Qubits: Gate vs Adiabatic QC
Atom as a logic Gate
0 1
0
1
Stable Ground States representing 0, 1

Classical annealing gets trapped
In local minimums
Quantum tunneling allows intermediate
superposition states and evolution to
lower lying ground state
Classical Annealing Quantum Annealing
Using the Evolution of a Physical System to Compute

D-Wave’s Processor Architecture
A System of Interacting (Artificial) Spins
H t  i t 
i1
28
 x
i
 hi t 
i1
28
 z
i
 Jij t z
i
z
j
i, j E
28

Finding the lowest Energy
Configuration of an interacting
spin network is a really hard
problem (NP-Complete) …
Solving this is equivalent to
many other hard, high value
problems

D-Wave’s Scalable, Quantum Processor Architecture
A System of Programmable Interacting Qubits
2 State Quantum Object:
Qubit
D-Wave Fully Programmable
Quantum Spin Computer
Structure like Neural Network

World’s Most Advanced Quantum Processor

Searching through a Vast Landscape of Possibilities
© 2014 D-Wave Systems Inc. All Rights Reserved 4
500 Qubits can explore 2500 possiblites

25 © Copyright 2012-2013 D-Wave Systems Inc.25 © Copyright 2009 D-Wave Systems Inc.
Isolating a QC Processor from the Environment preserving “Quantumness”

77K
4K
1K
300mK
20mK
Processor
Quantum
Processor
Housing

• Shielded room excludes RF
• Magnetic field reduced 50,000X
• Low Power ~ 15kWatts
•150x colder than interstellar space
Product: D-Wave TwoTM

D-Wave Two Installed at NASA Ames
USC-Lockheed Martin
Quantum Computing CenterApplications
Search for exo planets
Software Validation
 Mission planning
 Training Neural Nets
 Pattern Recognition
 New Physics of large
scale quantum system
 Autonomy/Robotics

Generating Meaning from Big Complex Data: Machine Learning
Optimizing Connections in a Neural Net to Create Intelligence
Image Recognition
Speech Recognition
Machine Autonomy
Discoveries Transcending
Human Understanding

“The Wright Flyer barely got off the ground,
but it foretold a revolution .. Their airplane
took a trajectory through the air ..
The D-Wave processor took a trajectory
through parallel universes to get to the
solution … it is literally that.
This is an amazing, somewhat historical
event. It has worked in principle.
The thing flew … “
Hartmut Neven
Head of Quantum AI
at Google

Graph
Coloring
Finding Ramsey
Numbers
Diagnosis
Constraint
Satisfaction
Monte
Carlo
Financial
Modeling
SAT Filters
Discrete Sampling
Complexity
Scaling
Quantum Error
Correcting Codes
Quantum
Simulation
Quantum
Research
Discrete
Optimization
Radiotherapy
Optimization
Deep Learning
Variational
Autoencoders
Boltzmann
Machines
Many Uses of D-Wave Technology Demonstrated
Optimization Sampling
Structured
Prediction
Trading
Trajectory
Optimization
Detecting Market
Instability

D-Wave Poised to Accelerate Machine Learning
(especially Probabilistic Graphical Models)
Computational Performance
Time
Classical
Quantum
• Enables ML models
that are too costly to
train by other means
at large scales
• Key to unsupervised
ML
Machine Learning
Sampling
Optimization

Generating Intractable Probability Distributions
for Generative Machine Learning
E
A
Z

Enabling Advanced AI: Probabilistic Machine Learning
•Discovering hidden structure
– Revealing the latent structure behind observations
– E.g. given descriptions of small organic molecules automatically identifying the latent features that make some molecules
effective as drugs
•Creating/generating/hypothesizing novel data
– Synthesizing novel data with the appropriate statistical structure
– filling in structure in occluded photographs, unseen frames in a video,
– Creating Novel Ideas via combinations of existing ideas
•Reducing amount of labeled data needed to train ML models
– E.g. generating synthetic labeled training data automatically from abundant/inexpensive unlabeled data allows us to reduce
the amount of human-curated/scarce/expensive data needed to train ML models
•Learn faster - incorporating background domain knowledge
– Humans learn from less data because we have effective background assumptions
– Probabilistic models can incorporate rich structured background knowledge allowing them to learn complex concepts faster

500 Qubit Vesuvius Processor
2013
Very Rapid Pace of Discovery, Innovation
300,000X Speed-Up realized in 2 years (one processor Generation)
128Qubit Rainier Processor
2011
300,000 X Speed Up
2 Years

Product Roadmap
D-Wave
Two
D-Wave
One
2,048-qubits Q4
2016
D-Wave architecture evolves 3-4 times per year

Growing Activity in Quantum Technology Globally
QUANTUM COMPUTER COMPANIES QUANTUM SOFTWARE PLATFORMS
QUANTUM-BASED SOLUTIONS
INTERNATIONAL RESEARCH
China Australia U.K. Singapore Japan E.U.
PsiCorp IonQ QuantumCircuits

The nineteenth century was known as the machine
age, the twentieth century will go down in history as
the information age. I believe the twenty-first century
will be the quantum age.
--- Paul Davies

EVOLVING QUANTUM COMPUTERS: Harnessing a Vast Hidden Reality

More Related Content

What's hot (6)

Similar to EVOLVING QUANTUM COMPUTERS: Harnessing a Vast Hidden Reality (20)

More from J On The Beach (20)

Recently uploaded (20)

EVOLVING QUANTUM COMPUTERS: Harnessing a Vast Hidden Reality