The Advanced Quantum Testbed at Lawrence National Berkeley Laboratory (AQT at Berkeley Lab) is an open and collaborative research program funded by the U.S. Department of Energy’s Office of Science. AQT explores and defines the future of quantum computers based on one of the leading technologies, superconducting circuits.
Multidisciplinary teams from academia, industry, and government laboratories apply to become AQT users. Then, in collaboration with AQT’s expert team of 20+ scientists and engineers, testbed users access the hardware and software free of charge, participate in its evolution, co-publish scientific results, and ultimately advance the science enabled by quantum computing.
The user program at AQT is not a typical turnkey collaboration available through commercial cloud-based quantum computers. Instead, these user-driven research projects represent scientific collaborations that extend over weeks and months. AQT provides users with detailed access to the entire computing stack, from the processors and control systems to the software interfaces. Users can also engage with AQT specialists at the fundamental levels of hardware fabrication, customization, study, and deployment.
Since the launch of the testbed user program in late 2020, AQT has received dozens of research proposals from users across the globe. An average of 5-6 users are active at a given time, working closely with AQT staff.
Three research teams at DOE national laboratories have offered their perspectives about conducting their novel research at AQT.
Microscopic nuclear simulations: The research conducted by Lawrence Livermore National Laboratory
Sofia Quaglioni serves as the Deputy Group Leader of the Nuclear Data & Theory Group in the Nuclear and Chemical Sciences Division at Lawrence Livermore National Laboratory (LLNL). She leads a user research project at AQT, with a collaboration that includes the University of Trento, titled “Near-term Quantum Simulations for Nuclear Physics.”
High-performance classical computing has a limited capacity to simulate physical systems that display complex quantum behavior. An example of this is how nuclei and rare isotopes bind from their neutron and proton constituents and how they decay and interact.
Quaglioni’s research team develops quantum algorithms to simulate the microscopic dynamics of nucleons. In collaboration with AQT staff, the team seeks to demonstrate a hybrid co-processing scheme, which uses classical and quantum computing to simulate the scattering of two neutrons.
A common thread amongst AQT users and partners is the value of detailed information about the specific hardware and ability to work with staff who have deep knowledge of the systems.
“We have been assigned a knowledgeable and dedicated team of quantum information scientists and experimentalists that access the testbed and run the simulations on our behalf. The AQT staff keeps us up to date with the results of each experiment they perform, and we have direct access to the measurements. The ability to interact with them and analyze the data to mitigate error allows us to converge toward the most efficient implementation of our algorithm and run slightly more complicated simulations. The AQT staff helps us both interpret what we are observing and make it successful by bringing to the table their experience in error mitigation techniques,” Quaglioni said.
Kyle A. Wendt, a staff scientist in LLNL’s Nuclear Data & Theory Group, also described his perspective.
“There’s a very clear vacuum in the scientific literature about nuclear physics on quantum computers and the understanding of what’s really going to happen on a quantum machine. AQT’s main edge is the intimate access we have with the staff to understand how our simulations are running and what’s happening. This access is not available anywhere else. And it helps us try to best tune our work to that platform to maximize our success. By working with the hardware specialists at AQT, we can fill in that vacuum, have a better sense of what’s really going on in our problem, and then consider other problems to design better algorithms for our next iteration.”
In addition to working with each research team, AQT seeks to establish a user community to share lessons learned and provide feedback. This feedback helps to continually improve all aspects of the platform.
Quaglioni explained:
“A few months ago, in the context of a user meeting, we did learn about other techniques that we were not aware of from other testbed users. So this user meeting sparked new ideas for our research. At the same time, our research also provides feedback to the hardware specialists.”
Wendt added:
“There’s a co-design process where we’re growing in parallel, as opposed to other testbeds elsewhere that provide new tools, and it’s up to the scientists to figure out how to use them. With AQT, it’s growing in tandem. This long-term vision will allow for broader and deeper science to be done.”
Quaglioni, Wendt, and the research team from Livermore and Trento look forward to continuing with the AQT user program.
“I hope that the AQT program has a long and fruitful life and that we can continue interacting with them,” expressed Wendt.
Holistic benchmarking: The research conducted by Sandia National Laboratories
Timothy Proctor is a principal investigator at Sandia National Laboratories’ Quantum Performance Laboratory (QPL), leading their research into quantum computer benchmarks. Due to environmental interference, any imperfection in quantum computing hardware causes errors in quantum logic gates and corrupts results. Researchers use various benchmarks to measure these errors’ rates by running carefully designed random circuits.
Proctor leads Sandia’s research project at AQT titled “Holistic and Scalable Characterization of Quantum Computers.” The project develops and demonstrates the first randomized benchmark to test computationally useful (“universal”) gate sets on large quantum computers. In close collaboration with AQT researchers, Proctor’s team extends “mirror circuit benchmarks” – a technique developed at Sandia – to the gate set of AQT’s superconducting qubit processors, including universal and continuously parameterized gates. These gates, which can be regularly adjusted, are an active area of exploration for AQT to increase the platform’s performance.
The AQT user program has enabled Proctor and his team to test and demonstrate their new technique. In addition, Proctor highlighted how the particular capabilities of AQT shaped their methods.
“At the AQT users’ meeting, we learned that certain types of non-standard –not widely available– entangling gates are particularly useful for near-term computations and that AQT had been working on calibrating these gates for another user project. Unlike existing techniques, our method is uniquely suited to benchmarking these gates. This presented us with an ideal proving ground for our methods.”
Collaborating closely with AQT staff has been valuable for further refining these benchmarking techniques.
“When we are developing new benchmarks, close collaboration with AQT’s team is really valuable. We often find that insights from experimental data allow us to improve our methods, and this is an iterative process. An ongoing collaboration with AQT allows us to explore how our methods perform on a real system, improve and perfect them, and then gather huge amounts of data to test the finalized methods. This is much harder to do with publicly accessible devices, with their much more limited access.”
Proctor looks forward to continuing to interact with the AQT program. Advances in quantum computing hardware open doors to co-design and improve other benchmarking and characterization methods.
“So far, we’ve been concentrating on developing and improving our randomized benchmarking techniques. But, in the future, we’re really excited to leverage the low-level access provided by AQT to test and improve the scalable quantum computer characterization methods that we’re developing at QPL.”
Integrations with AIDE-QC Software Architecture: The research conducted by Oak Ridge National Laboratory
Oak Ridge National Laboratory’s (ORNL) Thien Nguyen focuses on quantum computing programming and controls. Nguyen conducts programming and software experiments at AQT as part of the DOE’s AIDE-QC (Advancing Integrated Development Environments for Quantum Computing) project. Led by ORNL’s Alexander McCaskey, the team proposed a research project titled “Enabling a Co-Design Workflow via Direct Integration of the AQT and AIDE-QC Platforms.” It tests ORNL’s novel XACC quantum computing programming framework.
XACC is a system-level software infrastructure that integrates hybrid programming workflows (quantum code in tandem with classical code) regardless of the type of backend. Since quantum computers are still early-stage and prone to errors, according to the Oak Ridge team, the ability to harness the computational power of a quantum processor with XACC alongside classical computing (known as heterogeneous quantum-classical computing) could open breakthroughs for science.
“The quantum computing system alone is still small, error-prone, and quite frankly, not super useful as a standalone computing platform. Using XACC, we can build quantum-classical workloads to solve complex scientific problems, such as optimization and quantum chemistry.”
Nguyen highlighted one of the critical benefits of the user program at AQT to continue to test XACC and develop such a foundational software stack upon which new compilers and associated tools can be developed.
“We value the AQT testbed as the hardware platform that we can co-design with and develop new features for our quantum programming framework. At the moment, this is unlike any other commercial testbeds that already have a set of features that they want the users to use. We can talk with AQT staff at the fundamental levels of engineering to develop our dynamic quantum programming execution model. Indeed, what I appreciate the most is that AQT staff proactively provided insights about how a quantum computer actually works behind the scenes.”
Nguyen and the Oak Ridge team believe their experimental work at AQT will influence future software designs and implementations, and collaboration with industry, including companies such as Microsoft and Zurich Instruments. In turn, as AQT hardware continues to improve from the feedback received from its users’ program, the team at Oak Ridge can deliver novel software architectures for groundbreaking science.
“For the next few months, we want to demonstrate the execution of dynamical quantum programming mechanisms at AQT.”
Research proposals for non-proprietary work at AQT are always reviewed on a rolling basis. Learn more here: https://aqt.lbl.gov/about-aqt/collaborate-with-us/
###
Founded in 1931 on the belief that the biggest scientific challenges are best addressed by teams, Lawrence Berkeley National Laboratory and its scientists have been recognized with 14 Nobel Prizes. Today, Berkeley Lab researchers develop sustainable energy and environmental solutions, create useful new materials, advance the frontiers of computing, and probe the mysteries of life, matter, and the universe. Scientists from around the world rely on the Lab’s facilities for their own discovery science. Berkeley Lab is a multiprogram national laboratory, managed by the University of California for the U.S. Department of Energy’s Office of Science.
DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.
