IBM made three significant quantum computing announcements at CES this week. One was introduction of IBM Q System One; it’s really the integration of IBM’s existing 20-quibit platform into a more robust, self-contained “package” embodying what will be required as quantum computers move from the lab to the workplace. System One’s glass enclosure is not only cool-looking but also a Faraday cage. Big Blue also announced expansion of the IBM Q Network of quantum collaborators – ExxonMobil, CERN, and Argonne National Laboratory are among the new members – and plans to open the first IBM Q Quantum Computation Center in Poughkeepsie, NY, this year.
Rolling out the QC news at the Consumer Electronics Show may seem an odd choice, but then quantum computing is a little odd, and CES seems to be broadening from a consumer gadget extravaganza into a more expansive IT showcase. Regardless, the latest quantum news reflects Big Blue’s steady long-term effort not only to advance quantum computing research but also to push quantum computing towards commercialization and practical use. IBM is calling the new system, the ‘world’s first integrated quantum computer.’
“This project was begun over a year ago. We wanted to take a systems design approach to building a quantum computer. The goal of System One was to build a machine with a software stack that was automated in a way that we would do on a traditional machine, and we have all of the means of self-calibration, and the special purpose-built electronics to control the qubits and read them out,” said Bob Wisnieff, CTO quantum computing, IBM Research.
“You want to keep the overall system performing as stably as possible. [Final assembly and testing occurred] in October and November and it is it online now. Users have been using it for the last three or four weeks,” he said. This Q System One is located at the Thomas J Watson Research Center in Yorktown Heights, NY. A second will be installed at the new Poughkeepsie center.
IBM Q System One is comprised of a number of custom components that work together including:
- Quantum hardware designed to be stable and auto-calibrated to give repeatable and predictable high-quality qubits;
- Cryogenic engineering that delivers a continuous cold and isolated quantum environment;
- High precision electronics in compact form factors to tightly control large numbers of qubits;
- Quantum firmware to manage the system health and enable system upgrades without downtime for users; and
- Classical computation to provide secure cloud access and hybrid execution of quantum algorithms.
One challenge being tackled is the need to maintain the quality of qubits used to perform quantum computations. As noted in the official announcement, “Powerful yet delicate, qubits quickly lose their special quantum properties, typically within 100 microseconds (for state-of-the-art superconducting qubits), due in part to the interconnected machinery’s ambient noise of vibrations, temperature fluctuations, and electromagnetic waves. Protection from this interference is one of many reasons why quantum computers and their components require careful engineering and isolation.”
Q System One’s new ‘package’ includes a nine-foot-tall, nine-foot-wide case of half-inch thick borosilicate glass “forming a sealed, airtight enclosure that opens effortlessly using “roto-translation,” a motor-driven rotation around two displaced axes engineered to simplify the system’s maintenance and upgrade process while minimizing downtime.”
Wisnieff noted, “We chose was a laminated glass such that the glass itself is able to absorb RF and at the top of the case there is metal so that it acts as an ideal Faraday cage with a ground plane above. You can think of it as we are creating a quality of space where we want to control all of the aspects that matter so the qubit can operate with the maximum success possible.”
Like all current quantum systems except for D-Wave’s, which can be purchased and located on-premise, IBM System Q One is meant to be accessed via the cloud. Wisnieff said the cloud paradigm is likely to be the dominant delivery mechanism for quantum computing for the foreseeable future. System Q One currently runs IBM’s fourth-generation, 20-quibit processor. Even though the system is designed to be more robust, and therefore a more reliable resource for IBM’s Q Network collaborators, Wisnieff suggests that keeping the system in-house will facilitate making upgrades in all areas.
In building the web-platform around System Q One, IBM has focused heavily on usability. Debugging, for example, is a particularly thorny issue because you can’t measure the state of a quantum system without changing it. Prior simulation of ‘quantum code’ run on traditional machines is necessary and may require specialized compute resources.
“You run a simulation of the algorithm on a conventional machine, [where] it is perfectly legal for me know to probe the state of the system to understand exactly what is going on. We’ve learned there’s a ten-to-one simulation to actual quantum run [typically] required,” he said. “We have a number of specialized simulators available. We refer to them as different back ends. So when users submit their codes, they specify the back end that they want to run on, depending upon what aspect of the system they are interested in testing.”
It’s hoped that eventually much of the underlying complexity of quantum computing (system behavior, non-intuitive algorithms, quantum coding) can be abstracted and hidden from domain scientists.
“We’re certainly not ready to do that across the board yet,” said Wisnieff. “One of the things we have already done is in quantum chemistry where we have allowed people to use the data files and data formats that typically would have been submitted to conventional tools to do quantum chemistry calculations. They can submit the same job to a conventional tool like a supercomputer or take that job and move it onto a quantum computer.
“Longterm, that’s exactly the way that you want that hierarchical abstraction to exist so that researchers feel they are using this as a resource and interchangeable to a certain degree in terms of how you submit jobs. Quantum chemistry turns out to be a great place for us to begin experiment with how we might do that,” he said.
Nearer-term, the goal everyone is chasing – besides just developing better and bigger (more qubits) machines – is quantum advantage; that’s the ability use quantum computers do something sufficiently better than classical machines to make the effort worthwhile. “We think there’s a high probability that next several years we are going to begin to find algorithms that we can implement on machines that we can build that will provide some advantage,” said Wisnieff.
In the meantime, the number of collaborators signing on with IBM to develop and use quantum computers is growing. The other portion of IBM’s CES announcements dealt with expansion of the IBM Q Network which IBM describes as “the world’s first community of Fortune 500 companies, startups, academic institutions and research labs working with IBM to advance quantum computing and explore practical applications for business and science.”
Organizations joining the IBM Q Network include:
- ExxonMobil, the first energy company to join the network, “will explore how quantum computing may address computationally challenging problems across a variety of applications. Quantum computing could more effectively solve large systems of linear equations, which will accelerate the development of more realistic simulations. Potential applications include optimizing a country’s power grid, more predictive environmental and highly accurate quantum chemistry calculations to enable the discovery of new materials for more efficient carbon capture.”
- CERN, the European Laboratory for Particle Physics, “will work with IBM to explore how quantum computing may be used to advance scientific understanding of the universe. The project will bring together IBM and CERN scientists to investigate how to apply quantum machine learning techniques to classify collisions produced at the Large Hadron Collider, the world’s largest and most powerful particle accelerator.”
“These organizations will work directly with IBM scientists, engineers and consultants to explore quantum computing for specific industries. They will have cloud-based access to IBM Q systems, as they work to discover real-world problems that may be solved faster or more efficiently with a quantum computer versus a classical computer,” said Bob Sutor, vice president, IBM Q Strategy and Ecosystem.
The IBM Q Network provides its member organizations with quantum expertise and resources, quantum software and developer tools, as well as cloud-based access to IBM’s scalable commercial universal quantum computing systems available.
A subset of sorts to the IBM Q Network is IBM Q Hubs organization – you can see how hard IBM is working the IBM Q brand for its entire quantum ‘product’ line. The hubs are part of the IBM Q Network and have access to the IBM Q commercial systems, over the cloud, and focus on quantum computing education, research, development, and implementation.
“Many of our Hubs are government labs and universities. Part of their mission, too, is to partner with industry. The Hub at Keio University was the first to add members in 2017,” said IBM spokesman Chris Nay. The Oak Ridge National Laboratory’s IBM Q Hub, announced in 2017, recently added Argonne, Lawrence Berkeley, and Fermilab. “This group has a more government/academic flavor although IBM encourages its members to work with industry as well,” said Nay.
Here’s IBM’s description of current ORNL hub members and their areas of focus:
- Argonne National Laboratory “will develop quantum algorithms to help tackle challenges in chemistry and physics. New algorithms will also be used to model and simulate quantum network architectures and develop hybrid quantum-classical architectures, which combine the power of quantum processors with Argonne’s world-class supercomputing resources. Membership in the IBM Q hub will enable Argonne researchers to leverage their expertise in scalable algorithms across a broad set of multidisciplinary scientific applications and explore the impact of quantum computing on key areas including quantum chemistry and quantum materials.”
- “Fermilab “will use quantum computers for machine learning to classify objects in large cosmology survey applications, as well as optimization techniques to better understand the results of hadron collisions, and quantum simulation to research the potential of studying neutrino-nucleon cross-sections.”
- “Lawrence Berkeley National Laboratory “will use IBM Q systems as part of its quantum information science research to develop and simulate a variety of algorithms for studying strong correlation, environmental coupling, and excited state dynamics in molecular complexes and materials; novel error mitigation and circuit optimization techniques; and theories resembling the standard model in high-energy physics.”
IBM reports ORNL will use quantum computers along with high-performance supercomputers to benchmark new methods for studying strongly correlated dynamics in quantum materials, chemistry, and nuclear physics.
In addition, to the IBM Q Network and Hub, IBM also offers the no-cost and publicly available IBM Q Experience now supports more than 100,000 users, who have run more than 6.7 million experiments and published more than 130 third-party research papers. Developers have also downloaded Qiskit, a full-stack, open-source quantum software development kit, more than 140,000 times to create and run quantum computing programs.
Link to IBM Q System One announcement: https://newsroom.ibm.com/2019-01-08-IBM-Unveils-Worlds-First-Integrated-Quantum-Computing-System-for-Commercial-Use
Slide show on IBM Q System One: https://www.research.ibm.com/ibm-q/system-one/