Tracking quantum computing in America: Berkeley National Laboratory Progress Report

In April 2022, Lawrence-Berkeley National Laboratory released a progress report on its Advanced Quantum Testbed (AQT). Through this article, we can understand the actual progress in the field of quantum computing in the leading national laboratories in the United States.

What is AQT?

The Advanced Quantum Testbed (AQT), a leading collaborative research platform, was established at Lawrence-Berkeley National Laboratory in 2018 and is funded by the U.S. Department of Energy (DOE) Office of Science's Advanced Scientific Computing Research Program. The first phase focuses on deploying and integrating advanced superconducting hardware, and developing new quantum benchmarking techniques; this includes verification and validation, noise detection, suppression and mitigation.

In 2020, AQT will open the experimental platform. Received numerous experimental applications from academia, national laboratories, and start-up companies; the content covers: quantum algorithms, simulation, characterization, verification, hardware control, firmware, software, and processor architecture, etc. During the three years from 2018 to 2021, AQT has conducted various demonstrations in the fields of quantum simulation, optimization, quantum chemistry and condensed matter, nuclear physics and high energy physics.

Based on the above-mentioned cooperation opportunities and experimental facilities, AOT is also committed to training the next generation of quantum computing scientists and engineers. Provide training opportunities such as mentoring, open discussion, and multi-source cooperation in the quantum field. Users of AQT are also carefully selected, and will receive full access to hardware and software systems: including data on design, performance and operation. This will also further enhance the use of quantum hardware.

What architectures does AQT include?

After a series of cutting-edge experiments across disciplines, AQT has built a full-stack processing platform suitable for joint superconducting quantum information. This platform includes technologies such as the design and manufacture of novel quantum bits, quantum processor architectures, cryogenic packaging, room temperature control chains including hardware firmware and software, quantum circuit optimization tools, and quantum characterization verification and verification tools. These are all open to platform participation and users, forming a virtuous circle related to quantum research.

追踪美国量子计算：伯克利国家实验室进展报告

1) Software stack

Contains a versatile software stack for multiple open source quantum circuit formats, equipped with a range of algorithm optimization tools for efficient compilation and error mitigation. This stack can provide users with low-level access and non-standard software access requirements.

2) Commercial and custom controls

Includes custom and internal electronic controls. The internal electronic control stack, QubiC, meets the needs of platform users. The close collaboration between users and AQT scientists provides more flexibility than existing commercial quantum computing platforms.

3) Low temperature platform

Contains a Bluefors dilution refrigerator - Blizzard, equipped with 160RF line and operating at 10mk. Developed in partnership with Berkeley startup Bleximo, Blizzard includes a lab bench that can accommodate modular and scalable sets of experiments, as well as cryogenic packaging for improved electromagnetic hygiene. The platform can control and read 128 qubits and can accommodate multiple quantum processors.

4) Superconducting quantum processor

The AQT processor portfolio includes a range of different architectures with different sets of connectivity and local gates. The current standard processor for user experiments is an 8-qubit QPU with high-coherence qubits, high-fidelity gates (two-qubit gate fidelity up to 99%). New 8-qubit processors with arbitrary dynamically reconfigurable (up to all) connections are under development.

AQT is also working with MIT Lincoln Laboratory to develop new processors for the laboratory.

Major hardware advances in 2021- QubiC

Building on Berkeley Lab's longstanding particle accelerator research, AQT has developed its own chamber control stack. One of the most important is a new type of FPGA-based control processing system - QubiC, which has been modularized and open sourced. QubiC is capable of efficiently uploading and executing quantum experiments with minimal payload, can be customized to fit the unique needs of users, and provides fast feedback.

Today, AQT scientists have used QubiC to perform automated calibration experiments for two-qubit gates.

"Newer control circuits are not suitable for quantum processors, so quantum scientists purchase more instruments to accommodate the optimization needs of quantum hardware. However, the cost of control hardware does not grow linearly or exponentially, and this is also an area we are trying to dabble in. After truly realizing the need for future integration and design, we built a more feasible and cost-effective QubiC system," said Guang Huang, leader of the QubiC R&D collaboration in Berkeley Lab's Accelerator Technology and Applied Physics Division.

追踪美国量子计算：伯克利国家实验室进展报告

Open source test platform

AQT supports the U.S. Department of Energy (DOE) science and energy programs as an open-source experimental platform and is encouraged to be used by cross-disciplinary research teams in academia, industry, and national laboratories. To this end, AQT provides easily accessible deep expertise to refine project ideas for maximum potential impact, and often even bring new ways to implement them.

In addition, AQT is building its own user community, allowing users to exchange ideas and share knowledge. In 2020-2021, more than half of the projects are working with national lab teams. Users of the testbed include industry partners such as Quantum Benchmark, Keysight and Super.tech.

In October 2021, AQT launched a second initiative to users, receiving project applications on multiple topics. Includes implementing algorithms on scientific computers, noisy intermediate-scale quantum (NISQ) hardware benchmarks, and jointly building next-generation architectures and algorithms. Today, as a rapidly developing quantum information science community laboratory, AQT continues to receive applications for scientific research projects on a rolling basis.

追踪美国量子计算：伯克利国家实验室进展报告

Research highlights

Since its inception, AQT has made extensive progress in many fields of quantum information science. The specific experimental highlights are listed as follows:

1) Hi-Fi iToffoli door

The development of NISQ devices has expanded the range of executable quantum circuits with high-fidelity single- and two-qubit gates. Equipping NISQ devices with three-qubit gates will enable more complex quantum algorithms and efficient quantum error correction protocols with reduced circuit depth. In 2021, using fixed-frequency superconducting qubits, researchers demonstrated a high-fidelity iToffoli gate based on two-qubit interactions [1], the so-called cross-resonance effect. Like Toffoli gates, this three-qubit gate can be used to perform general-purpose quantum computations with up to 98.26% fidelity, providing more efficient gate synthesis than Toffoli gates.

2) qutrit random benchmark characterization error

Ternary quantum processors exploit quantum information encoding and processing in qutrits (three-level systems), offering significant potential computational advantages over conventional qubit technology. To evaluate and compare the hardware performance of such ternary quantum processors, the researchers demonstrate an extension of the industry-standard Random Benchmark (RB) protocol, widely developed and used for qubits [2], for ternary quantum processors logic. Several correlated gates are characterized by interleaved RBs, and synchronous RBs are employed to comprehensively characterize crosstalk errors. Finally, the periodic basis is applied to the dual quantum CSUM gate and a dual qutrit process fidelity of 0.85 is obtained. This result presents and demonstrates an RB-based tool, a general approach for characterizing the performance of a qutrit processor, and diagnosing future qudit hardware control errors.

3) Random compilation using the QITE algorithm

The success of NISQ hardware shows that quantum hardware is capable of solving complex problems even without error correction. An open question is coherent errors due to the increased complexity of these devices: these errors can accumulate through circuits, making their impact on algorithms difficult to predict and mitigate. The AQT study introduced a combination of noise tailoring using random compilation and error mitigation using purification; it was also shown that round-robin benchmarks gave reliable estimates of purification. The researchers applied this method to the quantum imaginary time evolution of a transverse field existence model and reported energy estimates and ground state error rates below 1%. This study shows how noise customization and error mitigation can be combined to improve the performance of NISQ devices [3].

4) Optimized Fermion SWAP Network

The Fermion SWAP network developed by AQT is a qubit routing sequence that can be used to efficiently perform a quantum approximate optimization algorithm (QAOA). In this work, the researchers optimized the execution of QAOA's fermionic SWAP network by two techniques: first, utilizing a set of ultra-complete native hardware operations (including 150 ns controllable π/2 phase gates with fidelity up to 99.67 %), decompose the related quantum gates and SWAP networks in a way that minimizes circuit depth and maximizes gate elimination; secondly, introduces equivalent circuit averaging, which randomizes over degrees of freedom in quantum circuit compilation to reduce systematic coherence errors Impact. This technique was experimentally verified on an advanced quantum test bench by executing a QAOA circuit to find the ground states of two-node and four-node Sherrington-Kirkpatrick spin-glass models with various randomly sampled parameters. Finally, it is observed that the errors of QAOA at depth p=1 are reduced by about 60% on average over four qubits on a superconducting quantum processor [4].

5) Random compilation for scalable quantum computing

In the NISQ era, systematic misalignment, drift, and crosstalk in qubit control can lead to coherent errors without classical simulations. Coherence errors severely limit the performance of quantum algorithms in unpredictable ways, and mitigating their effects is necessary to achieve reliable quantum computing. Random compilation is a protocol designed to overcome these performance limitations by converting coherent errors into random noise, thereby greatly reducing unpredictable errors in quantum algorithms, and accurately predicting algorithm performance through error rates measured by periodic benchmarks.

In this work, the researchers demonstrate a significant performance boost for random compilation of a four-qubit quantum Fourier transform algorithm and a variable-depth random circuit on a superconducting quantum processor; and using experimentally measured error rates that are accurate Predict algorithm performance. The results show that stochastic compilation can be used to exploit and predict the capabilities of modern noisy quantum processors, paving the way for scalable quantum computing [5].

6) Effectively improve the performance of noisy quantum computers

The researchers developed and experimentally validated two efficient error mitigation protocols, called "noise-free output extrapolation" and "Pauli error cancellation," to greatly improve the performance of quantum circuits composed of gate noise loops. By combining popular mitigation strategies such as probabilistic error cancellation and noise amplification with efficient noise reconstruction methods, this protocol can mitigate a variety of noisy processes that do not satisfy the assumptions of existing mitigation protocols, including non-local and gate-dependent processes . The researchers tested the protocol on a four-qubit superconducting processor on an advanced quantum testbed and observed significant improvements in the performance of both structured and random circuits, with up to 86 percent improvement in the variation distance of the untuned output. This experiment demonstrates the effectiveness of the research protocols and their practicality on current hardware platforms [6].

追踪美国量子计算：伯克利国家实验室进展报告

Nurturing the next generation of quantum workforce

Since its inception, AQT has recruited a large number of scientists from different backgrounds and disciplines to form a cross-country and multi-field joint communication community while cultivating a quantum workforce.

AQT trains undergraduate students to carry out one-year thesis projects, and also provides graduate students with laboratory training and project opportunities to participate in quantum algorithm applications. In AQT's training programs, students and postdocs collaborate with national laboratories and companies to learn a broader range of quantum knowledge.

AQT also provides additional educational opportunities for high school and undergraduate students. For example, Kasra Nowrozih, director of the hardware department, has led UC Davis students over the past few years in the National Science Foundation's "Research Experience for Undergraduates" (REU) summer program; AQT has also collaborated with Berkeley Lab's K -12 STEM "Education and Outreach Program", Occupational Interview with Ravi Naik, Director of Organization and Measurement.

AQT students and past graduates are employed in laboratories and top institutions around the world, including Google, Alice & Bob, Korea Institute of Science and Technology, Yale Quantum Institute, University of Rochester, and more.

追踪美国量子计算：伯克利国家实验室进展报告