Research

The Cloud Quantum Computing Platform (CQCP) Group at the Beijing Academy of Quantum Information Sciences (BAQIS) was established in early 2022, led by Researcher Heng Fan. Our team is dedicated to theoretical and experimental research in superconducting quantum computing (SQC), alongside the development and application of CQCPs. Our goals include increasing the number of qubits in superconducting quantum chips, establishing measurement and control systems for SQC, and leveraging these systems to advance quantum computing, quantum simulation, CQCPs, and quantum artificial intelligence. We also aim to promote the transformation and industrialization of quantum computing technologies. We emphasize the simulation of quantum phenomena using quantum computing methods, including quantum many-body systems, quantum phase transitions, quantum dynamics, and quantum chemistry. Our quantum computing research focuses on implementing diverse quantum algorithms, integrating quantum computing with machine learning and artificial intelligence, optimizing quantum logic gates, and realizing quantum error correction codes. Recently, our efforts have concentrated on two key areas:

◇ Theoretical and Experimental Research in SQC

◇ Advancing CQCPs and Integrating Quantum Computing Systems with Supercomputing and Intelligent Computing Systems

Below are some of our research directions and representative scientific achievements:

[1] Design and Fabrication of Novel Multi-Qubit Superconducting Quantum Chip (Supported by the Institute of Physics, Chinese Academy of Sciences (IOP, CAS))

The fabrication of SQC chips is pivotal to establishing an internationally advanced quantum cloud computing center, encompassing both chip design and processing. The effort is led by Associate Researcher Zhong-Cheng Xiang (part-time), with team members including Associate Researcher Xiao-Hui Song and over ten graduate students such as Gui-Han Liang, Si-Lu Zhao, and Zheng-Yang Mei. Over the years, the team has gained extensive experience in superconducting quantum chip design and fabrication, utilizing equipment and resources from the IOP, CAS, and public platforms to independently design and produce high-quality superconducting quantum chips in various configurations.

For chip design, the team is proficient in simulation technologies using software such as AnsysEM, Sonnet, and Comsol, enabling independent design of devices including Transmon qubits, C-shunt qubits, tunable couplers, tunable central cavities, Purcell filters, Josephson Parametric Amplifiers (JPAs), and Traveling Wave Parametric Amplifiers (TWPAs). We have developed the Transmon multi-qubit tunable coupling configuration independently, while most other device designs are also original, building on established principles. Future efforts will focus on designing larger-scale qubit chips and exploring short- and long-range coupling between qubits across different chips.

For chip fabrication, The team has mastered the use and maintenance of microfabrication equipment and techniques, including electron beam evaporation, laser direct writing, magnetron sputtering, electron beam lithography, reactive ion etching, wet etching, thermal evaporation, flip-chip bonding, wire bonding, profilometers, ozone cleaners, electron microscopes, X-ray diffractometers, Physical Property Measurement Systems (PPMS), atomic force microscopes, ion beam etching, probe stations, dicing saws, and white light interferometers. We are also exploring advanced fabrication techniques such as deep silicon etching, atomic layer deposition, and HF vapor etching.

Caption: Design, fabrication, and iterative process of SQC chips

Based on this expertise, the team has fabricated various SQC chips and published several high-impact research works, including:

◇ 10-qubit directly coupled chip for the ScQ (http://q.iphy.ac.cn and Quafu CQCP (https://quafu.baqis.ac.cn/and Sci. China Phys. Mech. 65, 110362, 2022).

◇ 20-qubit fully connected chip (Nat. Commun. 14, 1971, 2023; Phys. Rev. Lett. 128, 150501, 2022).

◇ 10-qubit tunable coupling chip (Nat. Commun. 14, 3263, 2023).

◇ 21-qubit "丰"-shaped tunable coupling flip-chip (Phys. Rev. Applied 23, 024059, 2025; arXiv:2501.18319v1, 2025).

◇ 30-qubit ladder configuration chip (Nat. Commun. 14, 5433, 2023; Nat. Commun. 15, 7573, 2024; PRX Quantum 6, 010325, 2025).

◇ 41-qubit directly coupled "Chuang-tzu 1.0" chip (Phys. Rev. Lett. 131, 080401, 2023, Nat. Commun., 16, 108 (2025)).

◇ 78-qubit two-dimensional tunable coupled "Chuang-tzu 2.0" chip (arXiv:2503.21553, 2025).◇ Novel tunable coupling TCCP configuration between qubits (Phys. Rev. Applied 20, 044028, 2023).

◇ Flying photon interconnect test chip (Phys. Rev. Applied 23, 024019, 2025).

Additional chips under testing or experimentation include the 48-qubit tunable coupling g-mon flip-chip, 16-qubit tunable coupling with fully connected tunable cavity flip-chip, 30-qubit tunable coupling Purcell filter flip-chip, 10-qubit tantalum film chip, C-shunt test chip, and 106-qubit heavy hexagonal chip. The team also has further chip design concepts and fabrication plans to be implemented as experimental conditions mature.

Future Strategic Roadmap, to enhance the performance of multi-qubit quantum chips, the team will focus on:

◇ Overcoming challenges in simulating larger-scale chips by integrating quantum computing power from existing superconducting chips with classical computing to efficiently simulate chip performance, addressing issues like parasitic modes and crosstalk.

◇ Developing sacrificial silicon dioxide layer airbridge technology to improve structural integrity compared to current reflow techniques.

◇ Advancing tantalum film fabrication and Purcell filter technologies to enhance chip performance, targeting average single-qubit coherence times exceeding 200 μs.

◇ Mastering modularization, through-silicon via, and multi-layer wiring technologies to support the fabrication and packaging of chips with hundreds of qubits.

The team operates multiple quantum measurement and control platforms capable of handling hundreds of qubits at the BAQIS and the IOP, CAS. Leveraging these platforms, we have achieved several significant results:

For quantum control and measurement technologies:

◇ Automated calibration technology based on neural network models (Patent: "一种超导量子比特的参数标定方法、设备和存储介质," Application No. 2024101387231, 2024).

◇ Multi-qubit readout optimization technology (Patent: "用于量子测量系统的优化方法、电子设备及存储介质," Application No. 2025100804877, 2025).

◇ Microwave correction technology (Nat. Commun. 15, 7573, 2024).

◇ Tunable coupler distortion correction technology (Phys. Rev. Applied 23, 024059, 2025).

◇ Novel three-qubit gate control technology (arXiv:2501.18319v1, 2025).

Measurement and Control Software Systems

◇ We have conducted preliminary modular packaging, optimized visualization functions, and improved waveform transmission protocols to facilitate future distributed and modular development.

For quantum simulations:

◇ Supported by strong theoretical expertise, the team excels in quantum many-body physics, non-equilibrium dynamics, quantum algorithms, and the integration of quantum computing with measurement and control experiments. Recent collaborations have produced notable outcomes:

◇ Using the 78-qubit superconducting quantum processor, Chuang-tzu 2.0, we reported the first experimental observation of the long-lived prethermal phases with tunable heating rates in many-body systems, driven by structured random protocols, characterized by the n-multipolar temporal correlations.

◇ With the University of Augsburg, Germany: Revealing dynamic signatures of weak ergodicity breaking in Stark systems (PRX Quantum 6, 010325, 2025) and spin hydrodynamics at infinite temperature (Nat. Commun. 15, 5733, 2024).

◇ With Academician Rong-Gen Cai (Institute of Theoretical Physics, CAS) and Professor Rui-Qiu Yang (Tianjin University): Experimental realization of on-chip black holes and quantum simulation of curved spacetime (Nat. Commun. 14, 3263, 2023).

Future Strategic Roadmap:

We will continue exploring machine learning-based automated control technologies for multiple qubits, developing modular and distributed quantum measurement and control software and hardware systems for heterogeneous computing architectures, and proposing quantum simulation schemes for novel heterogeneous computing systems, with validation on larger experimental platforms. Specific efforts include:

◇ Low-Temperature Wiring Innovation: Replacing coaxial cables with high-density flexible microstrip lines and all-optical wiring to reduce thermal noise and wiring complexity for scalability to thousands of qubits.

◇ Breakthroughs in Measurement and Control Electronics Hardware: Developing low-latency, dynamic feedback systems through industry collaboration.

◇ Development of Distributed, Modular Measurement and Control Software: Creating software with integrated parameter management, pulse compilation, and data visualization, supporting multi-threaded parallel control and automated task scheduling.

◇ AI-Driven High-Precision Control and Automated Calibration: Using reinforcement and adversarial learning to optimize quantum gate parameters and develop automated calibration strategies accounting for topological structures and crosstalk.

Caption: Architecture and operational logic of the CQCP

[3] Development and Application of CQCP

◇ For the construction of CQCP:

In collaboration with the IOP, CAS, and Tsinghua University, we launched the Next-Generation CQCP Quafu at the Zhongguancun Forum, recognized as a "major innovation achievement" in the March 5, 2024, the government report. In 2024, alongside the SQC team, we released a "large-scale quantum cloud computing cluster" at the Zhongguancun Forum, featuring over 590 physical qubits and single chips with up to 136 qubits, achieving internationally advanced performance and advancing practical SQC. The platform currently has over 4,500 registered users, has executed more than 3 million quantum computing tasks, and has produced over 20 high-impact research papers, laying a foundation for the quantum computing ecosystem and its applications. The platform has also achieved Level 3 information system security certification, ISO9001 quality management system certification for quantum computing software development, and CMMI Level 3 certification for software capability maturity. In collaboration with Huaxia Bank, we developed a cloud quantum finance platform based on Quafu, earning the Second Prize (First Place) in the 2023 "Golden Development Award" by the Central Bank.

Caption: Societal research on cloud platform applications

◇ For cloud quantum computing technologies:

We have collaborated with internal partners to develop high-performance quantum compilation algorithms that enable quantum algorithms to execute efficiently on target quantum processors. Benchmarked against IBM Qiskit, our compiler achieves comparable SWAP gate counts while delivering higher circuit fidelity. We have independently developed the Python-based quantum computing toolkit pyquafu and a C++-version Quantum Software Development Kit (QSDK). The Quafu Runtime system significantly reduces queuing time for repetitive tasks, while the Quafu Monitor real-time monitoring system enables comprehensive tracking of platform availability, response time, and gate fidelity metrics, with mobile alerts for critical events. Additionally, we deployed the Jupyter Hub interactive programming interface on the Quafu platform, allowing users to create, edit, and submit quantum computing tasks via the cloud (currently in internal beta).

In quantum compilation, we established a graph-based quantum circuit representation framework that integrates gate dependency and topology constraints into graph embeddings. Our compilation framework combines deep reinforcement learning and graph neural networks to enable intelligent optimization under complex constraints. We further developed an end-to-end quantum compilation optimization method tailored for NISQ devices, formalizing compilation tasks via Markov decision processes. Compared to traditional heuristic approaches, our framework demonstrates superior global optimization capabilities, adaptability to diverse hardware topologies, and enhanced generalization. The platform also features a DevOps pipeline with automated deployment, container management, and cloud orchestration. These innovations position the Quafu cloud platform to overcome international technological barriers. Related patents include: "Quantum Circuit Cutting Compilation Technique Based on Physical Hardware Constraints", "Quantum Circuit Compilation Framework Supporting Customizable Workflows and Parameter Updates" and "Quantum Program Auto-generation Method and System Using LLM-based Retrieval-Augmented Generation"

◇ For applications based on Quafu CQCP:

In collaboration with partners, we designed a hardware-efficient parameterized quantum circuit architecture and developed a quantum-compatible learning algorithm. By constructing reinforcement learning policy networks with quantum neural networks, we experimentally validated the feasibility of quantum machine learning on quantum hardware (Chin. Phys. B 33, 050301, 2024). For chemical simulations, we integrated the BiwonQ molecular simulation platform (developed by the Beijing Academy of Quantum Information Sciences) with Quafu’s superconducting quantum control system to enable precise molecular simulations on real quantum processors. In finance, we deployed a quantum-enhanced portfolio optimization system on Quafu and demonstrated accelerated risk analysis workflows for financial institutions.

User surveys reveal that two-thirds of platform users are quantum computing enthusiasts or beginners, with university students constituting a significant proportion, highlighting Quafu’s potential for education and talent development. Academic researchers leverage the platform’s computational power for advanced experiments, driving scientific progress. However, industrial adoption remains limited, indicating challenges in transitioning quantum technologies to real-world applications. Future efforts will prioritize industrial collaboration to bridge this gap and accelerate commercial deployment.

◇ Future Strategic Roadmap:

To enhance the competitiveness of the quantum computing cloud platform, the team will advance the following strategic initiatives across three dimensions: technological innovation, ecosystem expansion, and educational empowerment:

Technological Innovation: Pulse-Level Quantum Programming and Simulation: We will develop a pulse-level quantum control interface for physical hardware, enabling users to directly manipulate the microwave pulse waveforms, frequencies, and timing parameters of qubits. This will be supported by a Pulse Gate Library and a time-domain optimization toolchain, allowing researchers to customize quantum gate sequences and surpass the limitations of standard quantum gate models. For instance, on superconducting quantum chips, this will facilitate advanced control techniques such as dynamic decoupling and quantum error correction, as well as simulations of strongly correlated electron systems in condensed matter physics. Additionally, the platform will feature a pulse-level quantum circuit visualization editor with capabilities for parameter scanning and noise modeling, providing a comprehensive experimental environment for quantum device calibration and control theory research.

Ecosystem Expansion: Quantum-Supercomputing Hybrid System: Our goal is to build a quantum-supercomputing hybrid computing system that integrates quantum and classical computing resources through heterogeneous computing middleware. This system will include: A quantum-classical hybrid algorithm scheduling engine that dynamically allocates resources like quantum processors (QPUs) and graphics processors (GPUs) based on task requirements. Standardized interfaces for quantum and classical data to support mixed computing pipelines in areas such as quantum neural network training and quantum chemistry simulations. Collaboration with national supercomputing centers to deploy distributed quantum cloud nodes, creating a synergistic "center-edge" computing network. The first demonstration of this system will target high-complexity challenges, such as financial derivative pricing and climate model optimization, which require the fusion of trillions of classical computations with quantum sampling.

Educational Empowerment: Hierarchical Talent Development: We will establish a multi-tiered education system to cultivate a diverse talent pool in quantum computing: For beginners: An interactive quantum programming sandbox with animated circuit demonstrations, error diagnostics, and a suite of online courses. For advanced developers: A "Quantum Case Workshop" offering over 20 real-world experimental cases on the Quafu chip, covering quantum machine learning and optimization, with Jupyter Notebook support for online debugging. For academic partnerships: Tailored quantum computing lab courses aligned with university curricula. For industry training: A certification program for quantum engineers. Additionally, a developer community forum will be launched to foster collaboration through problem-solving, code sharing, and project showcases.

Innovation Competitions: Accelerating Application Discovery: To spur innovation, we will organize a three-tier competition framework: Algorithm Challenge: Focused on practical NISQ-era applications, such as optimizing Quantum Approximate Optimization Algorithm (QAOA) parameters and innovating error mitigation techniques. Programming Marathon: A 72-hour hackathon using the Quafu Runtime system, targeting engineering challenges like quantum machine learning frameworks and quantum EDA toolchains. Industry Application Contest: Partnering with industry leaders to tackle real-world business problems. Competitions will offer quantum computing resource credits or cash prizes, with top teams eligible for incubation funding.

Ultimately, the Quafu platform will evolve from a technical tool into a comprehensive ecosystem engine, creating an innovation value chain that spans basic research, technology development, and industrial application. This will provide critical support for China's quantum computing industrialization efforts.

Caption: Key technology R&D framework for quantum-superconducting hybrid integration

[4] Research and Development of Key Technologies for Quantum-Supercomputing Integration

With the rapid advancement of the modern information society, computing systems have become the foundational infrastructure for information processing. The growing prevalence of artificial intelligence has intensified the demand for supercomputing and intelligent computing power, inevitably leading to energy supply shortages and developmental bottlenecks. Although quantum computing offers a theoretically perfect solution to these challenges, its current stage—characterized by intermediate-scale noise—has made it clear that quantum computing cannot yet replace classical computing. Internationally, there is a strong focus on the deep integration of quantum and classical computing (referred to as "Quantum-Supercomputing Integration" or heterogeneous computing). This includes systematic efforts such as building quantum cloud computing power, designing heterogeneous computing models for quantum and classical computing centers, and developing foundational platforms like quantum operating systems.

The core of Quantum-Supercomputing Integration lies in creating a synergistic computing system that combines quantum and classical computing. Through deep hardware architecture fusion and software ecosystem interoperability, this approach aims to achieve exponential increases in computing power and efficient resource utilization. The strategic importance of this direction is multifaceted:

◇ Overcoming Current Limitations: While quantum computing shows exponential acceleration potential for specific problems, it is still constrained by noise, error correction, and scalability issues, making it unable to independently handle complex tasks. Integrating with supercomputing allows quantum computing to focus on its strengths—quantum state superposition and entanglement operations—while classical computing handles preprocessing, postprocessing, and algorithm optimization, creating a complementary advantage. For example, in quantum chemistry simulations, quantum processors can efficiently solve molecular ground state energies, while classical computing optimizes parameters and verifies results, significantly enhancing overall efficiency.

◇ Lowering Entry Barriers: Quantum-Supercomputing Integration can effectively reduce the entry barrier for quantum computing. By unifying quantum and classical computing power through cloud platforms, it empowers innovation in fields such as artificial intelligence, materials science, and financial engineering. In financial risk analysis, for instance, quantum computing can accelerate Monte Carlo simulations, while classical computing handles data cleaning and model validation, drastically shortening the computation cycle.

◇ Achieving Technological Independence: This integration is a critical path for Our country to break through international technological barriers and achieve self-reliance. Currently, global quantum computing giants like IBM and Google have established technological dominance, with increasing restrictions on our country. By independently developing Quantum-Supercomputing Integration systems, We can reduce reliance on foreign technology, build a complete quantum computing industry chain, and ensure national technological security.

Our team has been strategically positioned in this direction for some time. In 2024, we connected the quantum computer at the BAQIS's 410 laboratory to a high-performance data center via fiber optics. We have also developed a preliminary quantum-classical hybrid computing runtime framework, which, after testing, achieved over a twofold speed improvement in a Quantum Neural Network (QNN) runtime training task. In September 2024, the Ministry of Industry and Information Technology's "Advanced Computing and Emerging Software" key project highlighted a critical technology initiative aimed at researching system architectures and application algorithms for supercomputing and quantum computing integration. Our team, in collaboration with the IOP, CAS, Sugon Information Industry Co., Ltd., RIGOL Technologies, and the National Supercomputing Center in Chengdu, has systematically planned for this initiative. Our partners have extensive expertise in SQC experiments, quantum computing algorithms, quantum operating systems, CQCPS, high-performance computing, and heterogeneous computing, with leading industry positions and comprehensive coverage of all research aspects required for the Quantum-Supercomputing Integration project.

Future Strategic Roadmap:

Looking ahead, our team is committed to establishing an internationally advanced "Quantum-Supercomputing Integration" cloud computing center that achieves deep fusion between classical and quantum computing. Specifically, we will:

Build a high-performance heterogeneous computing hardware framework. Develop automated multi-bit calibration technologies. Create an operating system for multi-quantum task scheduling and circuit compilation tailored to heterogeneous computing systems. Develop high-performance noisy quantum simulators to meet the simulation and verification needs of algorithms and applications. Conduct demonstrations and explorations of multi-qubit quantum-classical hybrid algorithms and applications via cloud platforms.

On this foundation, we will continuously improve control and measurement precision, driven by scientific research to add functionalities and advance technological development. Ultimately, we aim to bring multiple quantum systems—such as ion traps and neutral atoms—onto the cloud, realizing a multi-system Quantum-Supercomputing Integration computing framework. To achieve these goals, our team will focus on the following research areas:

◇ Development of Scalable Quantum Measurement and Control Systems for Quantum-Supercomputing Integration: We plan to collaborate with quantum measurement and control equipment manufacturers to develop a scalable quantum measurement and control system specifically designed for Quantum-Supercomputing Integration. This system will efficiently manage large-scale quantum bit arrays using a modular hardware architecture based on the PXIe platform, optimizing wiring and ensuring electromagnetic shielding effectiveness exceeds 65 dBc. A hybrid cooling solution combining air and liquid cooling will be designed to control temperature rises within 5°C. In the future, the system will support hundreds of quantum bit control channels with feedback delays under 800 ns, eventually scaling to 5,000 channels. By integrating advanced hardware and cooling technologies, we will achieve precise and efficient control of large-scale quantum bit arrays, laying the groundwork for practical quantum computing.

◇ Development of Quantum Measurement and Control Platforms and Technologies for Quantum-Supercomputing Integration: We will build a quantum measurement and control platform tailored for Quantum-Supercomputing Integration, incorporating cutting-edge technologies to achieve high-precision quantum bit manipulation. Key developments include an all-optical microwave signal transmission system capable of handling over 1,000 channels with minimal heat load and supporting machine learning-based automated calibration technologies to reduce single quantum bit calibration time to under 15 minutes. The platform will integrate high-fidelity quantum array control and readout systems, aiming for 99.8% two-qubit gate fidelity. This research will ensure the platform supports large-scale, high-precision quantum bit control and efficient interaction with classical computing resources.

◇ Development of Quantum Runtime Systems: We will develop a quantum runtime system to provide an efficient execution environment for quantum tasks. The system will optimize task scheduling and circuit compilation, reducing circuit depth by over 10% and enhancing quantum algorithm performance. By designing a robust software framework, we will facilitate seamless operation and management of quantum computing, enabling reliable and rapid execution of complex quantum tasks.

Development of a Unified Quantum-Classical Programming Framework: In collaboration with partner teams, we will develop a unified programming framework that integrates quantum and classical computing resources, supporting C/C++ and Python interfaces. This framework will allow developers to easily write hybrid quantum-classical programs, promoting algorithm development that leverage both computing paradigms. By providing a seamless programming interface, this research will accelerate the development and deployment of quantum applications across various fields.

◇ Development of Quantum-Classical Computing Resource Management and Scheduling Systems: This project will implement a management and scheduling system for quantum and classical computing resources, optimizing task allocation between quantum and classical processors. We will develop advanced scheduling algorithms to improve resource utilization efficiency by 20%, ensuring effective task distribution. The system will maximize overall computing performance through intelligent management of quantum and classical resource synergy.

◇ Development of High-Performance Quantum Computing Simulators: In collaboration with partner teams, we will develop high-performance quantum computing simulators to simulate large-scale quantum systems on classical hardware. The simulator will support noisy quantum circuit simulations for up to 20 qubits and enable distributed simulations across multiple GPUs and computing nodes. By providing accurate and scalable simulation capabilities, this research will assist researchers in designing, testing, and optimizing quantum algorithms.