Quantum Computing with Amazon Braket In this talk, I describe some fundamental principles of quantum computing including qu-bits, superposition, and entanglement. I will demonstrate how to perform secure quantum computing tasks across many Quantum Processing Units (QPUs) using Amazon Braket, IAM, and S3. AI and Machine Learning, Quantum Computing, Amazon Braket, QPU

1. Chris Fregly
Developer Advocate
AI and Machine Learning
@AWS
Quantum Computing with
Amazon Braket

2. Me: Chris Fregly
Developer Advocate
AI and Machine Learning @ AWS
(Based in San Francisco)
Co-Author of the O'Reilly Book,
"Data Science on AWS.”
https://www.datascienceonaws.com
Founder of the Advanced
Kubeflow Meetup (Global)
https://meetup.com/Advanced-Kubeflow
github.com/data-science-on-aws
@cfregly
linkedin.com/in/cfregly

3. Data Science on AWS – Book and Workshop Outline
https://www.datascienceonaws.com/

4. Agenda
• What is Quantum Computing?
• Use Cases for Quantum Computing
• QPUs, Qubits, Superposition, Entanglement, Teleportation
• Programming QPUs
• DEMO: Quantum Computing with Amazon Braket

5. What is Quantum Computing?
• “Quantum computers are no longer theoretical devices.”
-- Programming Quantum Computers
by Johnson, Harrigan, Gimeno-Segovia
• Based on Quantum Mechanics (1930’s)
• Quantum Mechanics is “Nature’s Operating System”
• Einstein Called Quantum Mechanics ”Spooky”
“No reasonable definition of reality could be expected to
permit quantum mechanics.”

6. Why Quantum Now?
• Enormous Long-term Potential
• Learn & Build Expertise Now
• Develop New Algos and IP
• Start the Quantum Flywheel!
• Quantum Today = Deep Learning 10-15 Years Ago
Growth
Users
Hardware
capabilities
Use cases and
algorithms
Hardware
developers
Betteruser
experience
Better
understanding
of QC use cases

7. Quantum Computing Eraserrorrate(quality)
number qubits (quantity)
Eras:
1. Classically Simulatable
2. Noisy Intermediate-Scale Quantum (NISQ) ç Today
3. Error-Corrected Quantum Computing ç 5-10 Years?
>50, <100k Qubits:
Today’s Moderate-Useful Apps
Quantum Supremacy
(we just crossed over)
“Supremacy”: We can no longer track what a
quantum computer is doing.
(Huge milestone achieved in 2019.)
more qubits
betterqubits
Quantum Advantage
(not yet crossed, find useful apps)

8. Amazon Braket puts quantum computing in
the hands of every developer and scientist
Managed development
environments
High-performing
circuit simulators
Hybrid workloads on
secure, on-demand
quantum hardware

9. Amazon Braket provides secure, on-demand access to
different quantum computing technologies

10. Use Cases for Quantum Computing
• Cryptography
• Factor 2048 digit RSA key: quadrillion years (digital), 3 mins (quantum)
• Classic Machine Learning Algos are Limited on Quantum
• Linear Regression, PCA, SVM, Fast Fourier Transform, System of Eqns
• Exponential Speed-ups for Certain Classes of Algorithms
• ie. From O(n) to O(log n)
• Financial, Chemical, Material Science Simulations

11. Quantum Software Industry
https://www.creativedestructionlab.com/streams/quantum

12. Customer: Fidelity
• Financial and Market Simulations
• Creating Synthetic Securities
• Options Pricing
https://www.fcatalyst.com/projects/aug2020/amazon_braket
https://www.fcatalyst.com/projects/aug2020/amazon_braket

13. Customer: Volkswagen Group
• Automotive

14. Customer: Amgen
• Biotech and Drug Discovery

15. Customer: Qu & Co
• Chemical Simulations and Material Science

16. QPUs
• Quantum Processing Unit (QPU)
• Hybrid Quantum, Co-Processors Similar to GPUs
• Short Bursts of QPU Computations, Coordinated by CPU
• QPUs Defined by Number of Qubits (Quantum bits)
• Today’s Quantum Computers have <100 Qubits
• Today, Some Qubits Needed for Redundancy & Error Correction

17. Qubits
• Qubit: Quantum Bits, Qunibble: 4 Qubits, Qubyte: 8 Qubits
• Qubits are Ultimately 0’s and 1’s Just Like Digital Bits (?!)
• There is only 1 photon per “shot”
• Value of Qubit is Probabilistic Until READ (73% 0, 27% 1)
• READ Forces Qubit Value to 0 or 1; Destroys Quantumness
• Probabilistic State is called Superposition
1 Shot

18. Superposition
• ie. Rolling Dice are in Superposition
• Value of Dice is Probabilistic Until READ
• Humans READ Dice When Dice Stop Rolling
• Superposition Represented in Circle Notation
• Probability of Reading a 1 (vs. 0) from Superposition
No Superposition
Superposition
Normalized to
50% 0, 50% 1

19. Circle Notation
• 2^n “Circles”; n=# of Qubits
• 32 Qubits: 4,294,967,296 Circles
• 300 Qubits: More Circles Than Atoms in the Universe!

20. Qubits Needed for Use Cases

21. Superposition Amplitude
• Measured in Magnitude (Radius) + Relative Phase (Angle)
• Phase Manipulation is Critical in Quantum Programming
• Put Convention Data into Superposition with HAD Gate
Quantum Random
Number Generator

22. Entanglement
• “Bell Pair”, John Bell, Irish Physicist
• READ One Qubit, Another Qubit Reveals the Same Value
• ie. 2 Dice Always Show Same Value…Across the Universe
• Einstein Called This “Spooky Action at a Distance”
• Extends to Multiple Qubit Groups (ie. Qu-nibbles, Qu-bytes)

23. Teleportation
• “Quantum Ethernet Cable”
• Requires 2 (or More) Entangled Qubits in a Bell Pair
• Transport State of Qubit A to Qubit B; Destroys Qubit A
• Currently, Only Transports Short Distances within a QPU
• Today, Longest Teleporation is 1,400km (870mi)
• Note: Not Transporting Matter – Only Information

24. Programming QPUs: Quantum Computing Simulators
• Limited Qubit States Due to Digital RAM Constraints
• 30 Qubits == 16GB, 40 Qubits = 16TB, 50 Qubits = 16PB

25. Build: Managed Jupyter Lab Environments
Fully managed infrastructure
Pre-built environments
Learning resources
Build and test algorithms

26. Test: Hosted Circuit Simulators
Serverless execution
Optimized performance
Individual circuits and hybrid
jobs
Schrödinger
simulator
Tensor network
simulator
…

27. Run: Managed Execution of Hybrid Quantum
Algorithms
Users
Local
IDE
Fully managed
execution
Pay only for what you
use
Change backend with a
single line of code
Amazon
Braket
Classical
compute
Simulator
Amazon S3
Amazon CloudWatch
QPU

28. Programming QPUs – Random Number Generator
circuit = Circuit()
a = circuit.h(0) # Put data into Qubit Superposition
b = a.cnot(0, 1) # Entangle Qubits
task = device.run(b, s3_prefix, shots=100000)
task.result().measurement_counts
Counter({'11': 50084, '00': 49916})

29. Programming QPUs - Teleportation
• Step 1: WRITE Conventional Data from CPU into QPU
• Step 2: Put Data into Superposition using HAD Operation
• Step 3: Entangle Qubits using CNOT Operation
• Step 4: Perform PHASE Ops on Qubits in Superposition
• Step 5: READ Qubits from QPU back into CPU

30. © 2020, Amazon Web Services, Inc. or its affiliates. All rights reserved.
DEMO: Amazon Braket

31. Looking Ahead
• Quantum Research Improves Conventional Algos
• ie. Recommendation Systems
• Error-Correcting Qubits
• Enables Deeper Circuits Across More Qubits
• Quantum Now = Deep Learning 10-15 Years Ago
• Limited by Hardware Availability … Until Now!

32. AWS Center for Quantum Computing (Opens 2021)
Near-term applications
Error correction & programming
models
Quantum hardware and
technologies

33. Amazon Quantum Solutions Lab (Available Now)
Quantum
Computing
Machine learning
HPC
Build and optimize deep learning approaches on AWS
Build and optimize classical and quantum-inspired
approaches on AWS
Target use case
Build quantum computing prototypes
Develop new algorithms and approaches
Benchmarking Results
Collaboration Education Solutions

34. Technology and consulting partners provide software
and service solutions on Amazon Braket

35. Thank you!
© 2020, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Chris Fregly
github.com/data-science-on-aws
@cfregly
linkedin.com/in/cfregly
https://www.datascienceonaws.com
https://aws.amazon.com/quantum-solutions-lab/