Notes from the D-Wave Systems Quantum Computing Seminar

By: Jeffrey Cohen, Founder & President of US Advanced Computing Infrastructure, Inc.
April 18, 2019
An informative, 4-hour session, with multiple speakers and ongoing Q&A.

Rene Copeland, President, D-Wave (Government) Inc. is responsible for government and university clients.

• He mentioned that not all quantum computers work the same way, nor have the same architecture (D-Wave uses superconducting, Annealing systems (2,048 Qbits) and others use Gate systems (up to 72 Qbits).  D-Wave has annealing systems which periodically aligns the components of the system on a quantum basis (like thermal annealing aligns molecules in steel by heating the metal, working it as it cools, then heating it again).  
• D-Wave is enhancing its architecture (from Chimera to Pegasus) to add hamiltonian functions. This adds materials sciences simulation capabilities. This is one of the capabilities of gate-based quantum systems, and an important client use-case.  This is an important innovation for D-Wave Systems. They are also adding enhanced physical networking between qubits (more programming options), and increasing from 2,048 to ~5,000 qubits per QPU/system.
  
• For classical network architects, it is important to know there is no ‘flex-fabric’ nor shared networking. Qubits need to be connected via a physical network linkage to be programmed together.
• The overall system will be rated at 20Kw to 25Kw (which is largely due to the cooling infrastructure).
• Quantum Computers cannot yet outperform the most powerful classical computers (a term called quantum supremacy), but they are getting more powerful.  As I wrote in a past blog, classical HPC systems also keep improving (e.g., exascale computer being built at Argonne National Lab, 10^18 FLOPS)
• There are videos to watch that show client’s quantum experiments and experiences, including companies like Denso and Volkswagen.  You can find them on YouTube. I watched many of them in preparing this summary.

Edward (Denny) Dahl, Architect & Scientist, D-Wave Systems,

• Denny discussed three types of quantum architectures, superconducting loops (D-Wave), Trapped Ions (IBM and Google) and Topological Matter, Majorama (Microsoft).  Not sure which one Intel is pursuing.
• Error correction was discussed in detail, and it creates a very high overhead on a quantum system.  One speaker suggested up to 9 physical qubits may be required to operate a logical (and error corrected) qubit.  Sounds like legacy Computer Associates storage management models (9 copies of the data ensures 100% redundancy). It is a potential show-stopper for quantum computing reaching ‘general purpose computing’ capabilities and scale unless it is resolved.
• Denny gave us a little history.  It took D-Wave 7 years to build its first generation system (2004 – 2011), and only 9 years to build and release 4 more generations (Pegasus is a 5th generation system).  Innovation is accelerating.
• He sees the D-Wave use-cases as focusing on Optimization, Machine Learning, and Materials simulations.
• Denny gave some specifics on Pegasus networking that I will summarize by saying there is increased and more complex network connectivity between qubits, and greater qubit density in Pegasus.  The details are important once you start programming, and Denny can provide them.
• They will be targeting to run Non-Stoquastic hamiltonians.  They point us to a key publication by Ozfidan in March 2019 (1).

Bob Sorensen, VP for Research, Technology, and Chief Analyst for Quantum Computing (QC) at Hyperion Research.  A High Performance Computing (HPC) spin-off from IDC that is 100% US owned

• Hyperion Research has been requested to start a research activity around Quantum Computing.  They also organize large-scale events for the high performance computing user community. The HPC User Forum, www.hpcuserforum.com, definitely looks interesting and I asked to join.
• Bob discussed how quantum computing (QC) should be thought of in three buckets, hardware, software and algorithms.  Hardware is noisy (NISQ) and runs with errors, used for unique, special purposes, and still tweaked and adjusted for specific applications (e.g., clock speed, or qubit performance characteristics) to get a job to run better.  Systems software needs more work to write great operating systems, database management systems, compilers and networking models. Algorithm writing is the use-case for individuals and small firms. The race is on for killer applications, and just applications that solve specific client problems.  This is where the value gets created for clients.
• He spoke about algorithms and business use cases.  For oil and gas, maximizing output from petroleum reservoirs is a high-leverage activity.  A 10% improvement in yield (from 10% to 11%) would result in $170B in incremental revenue. Chemistry and Pharma are two other high-leverage industries.  
• Another use-case is to help showcase the value of quantum.  Chad Rigetti, CEO of Rigetti Systems, just announced their cloud-based, quantum - classical hybrid operating environment, named Quantum Cloud Services (QCS).  In September, 2018, Rigetti announced a $1M challenge. The prize goes to anyone that demonstrates quantum advantage on their QCS system. It has recently been improved, and they are looking for their ‘killer application.’
• Bob also spoke generally about workforce enablement, and referenced University of Wisconsin’s masters degree program in Quantum.  Universities are teaching quantum in greater numbers.

Dr. Joel M. Gottlieb, Senior Presales Analyst, D-Wave Systems, who helps new customers to effectively use the D-Wave systems.  

• He explained Leap (the access model for a D-Wave system, and Ocean, a set of libraries and tools to develop quantum algorithms.  Ocean is based on the Python programming language and API (structured and abstracted), and runs on Jupyter Notebook.  https://jupyter.org/ and also a D-Wave GitHub repository.  We need to be proficient in the Jupyter Notebooks and Python to work in the environment.  For those proficient in coding, ‘PIP install dwave-ocean-sdk’ which I plan to learn and use.
• The timing on adding the Materials Sciences problem running on Leap is to be decided.  The challenge is that materials science jobs take longer to run than the free minute Leap (free) users have.  Paid users spend $2,000 per hour per month, so they can leverage Leap for materials sciences.
• He discussed how innovation is progressing very fast in terms of how to access the system and run jobs.  QB Solve is an example of a quantum-classical hybrid application, which they call D-Wave Hybrid ™. The D-Wave team is bringing more tools and technologies to market to make it easier for people to crowdsource algorithms and gain exposure to Quantum Computing.

Attorney, R. Paul Stimers, Partner, of the law firm K&L Gates (2).  

• Paul discussed the legislative process to support quantum and outlined the National Quantum Initiative Act (NQI) law that was passed on December 21, 2018.  The NQI law establishes US Federal coordination of quantum research and development efforts by civilian agencies. NQI outlined a budget and programs totaling $255M per year for five years, of existing monies appropriated or otherwise made available.  In other words, this is not new money, but a call to find and redirect existing funds to be spent on quantum.
• There will be an NQI Advisory committee, which Paul suggested would support three goals: 1) have the information on quantum be minimally classified, 2) rely on commercial technologies and providers over time, and 3) have our allies perform the development work as much as possible.  The idea is to keep quantum computing open, commercial, global and collaborative.
• This law has a parallel with the National Nanotechnology Initiative passed in 2000 (3).  Their .gov website indicates the Fiscal Year 2019 US Federal Budget has $1.4B to spend on nano technology, as compared to $255M for Quantum.
• Paul discussed the Quantum Industry Coalition: A unified business voice for US leadership in quantum computing, communications and cryptography in Washington, DC.  Their website is www.quantumindustrycoalition.com.  Intel and other large technology providers play an active role in this coalition.  It is worth a read and your consideration whether to join (open to companies working in Quantum Computing).  

Diane Carr, Director, Customer training.  

• Diane spoke briefly about D-Wave Systems efforts to help train and enable clients and other users of the platform, and I recall ~ 1,000 have been trained so far.  There will be an open training course in the Fall of 2019, and I plan to attend.

Notes:

1. Ozfidan, I & Deng, C & Y. Smirnov, A & Lanting, T & Harris, R & Swenson, Loren & Whittaker, J & Altomare, F & Babcock, M & Baron, C & Berkley, Andrew & Boothby, K & Christiani, H & Bunyk, Paul & Enderud, C & Evert, B & Hager, M & Hilton, J & Huang, S & Amin, Mohammad. (2019). Demonstration of nonstoquastic Hamiltonian in coupled superconducting flux qubits.
   
2. http://m.klgates.com/r-paul-stimers/
   
3. Nanotechnology Initiative.  www.nano.gov. and  https://www.nano.gov/about-nni/what