Control and Readout Software for Superconducting Quantum Computing

Cheng Guo, Futian Liang, Jin Lin, Yu Xu, Lihua Sun, Wei-Yue Liu, Sheng‐Kai Liao, Cheng-Zhi PengIEEE Transactions on Nuclear Science 2019 · DOI · arXiv

Abstract

As critical components of superconducting quantum computers, high-speed arbitrary waveform generators (AWGs), ultra-precision DC sources, and high-speed digitizers are essential for qubit manipulation and readout. However, as the number of qubits scales, the complexity of experimental setups increases rapidly, leading to bottlenecks in instrument management, signal distortion, and data transmission efficiency. To address these challenges, we propose an instrument management software design that maps separate physical resources to a unified virtual instrument. This software features server-side signal correction, a multi-threaded AWG task mechanism, and a custom data-link protocol for digitizers to enhance efficiency. The system also provides real-time feedback capabilities via an instruction compiler, meeting the rigorous requirements of superconducting quantum computing as demonstrated in a 12-bit quantum experiment.

Introduction: The Scaling Bottleneck

Superconducting quantum processors require a sophisticated array of room-temperature electronics to function. Each qubit requires multiple channels of high-speed, precisely timed analog signals for state manipulation and sensitive digitizers for state readout. In traditional setups, managing these dozens of separate instruments manually is error-prone and slow. Furthermore, realistic hardware introduces signal imperfections, such as IQ mixer imbalances and pulse edge degradation, which can significantly lower gate fidelity.

Figure 1: Typical hardware experimental setup for superconducting qubit control and readout.

Beyond physical complexity, data bottlenecks present a major hurdle. Standard communication protocols often struggle to keep pace with the massive data volumes generated by 1 Gsps (Gigasamples per second) digitizers, leading to significant latency between experimental runs.

Proposed Software Architecture

To overcome these challenges, we developed a modular software framework based on a Client-Server (C/S) model. This architecture abstracts physical hardware into a "unified virtual instrument," allowing researchers to interact with a logical representation of the system rather than individual hardware ports.

Figure 3: The proposed Client-Server software architecture for instrument management.

The architecture is divided into two primary functional blocks:

  • Control Server: This component manages AWGs and DC sources. It utilizes a symbolic "Waveform Engine" (leveraging the GiNaC library) to define complex pulses mathematically. It applies Finite Impulse Response (FIR) filters to correct hardware defects and handles IQ mixer calibration to suppress local frequency leakage.
  • Readout Server: Dedicated to digitizer management, this server regroups captured data frames and employs homodyne algorithms to determine qubit states (|0⟩ or |1⟩) by analyzing phase and amplitude in the IQ plane.

Technical Innovations and Performance

Several key innovations drive the system's efficiency:

  1. Custom Data Link Protocol: To bypass the overhead of standard network stacks, the Readout Server uses the WinPcap library to capture raw packets directly from the Ethernet port. This allows for data transmission speeds exceeding 400 Mbps through a single Gigabit-NIC.
  2. Multi-threaded Asynchronous Communication: The Control Server maintains individual task threads for each AWG. This ensures that adding more hardware does not lead to a linear increase in communication latency, facilitating easier scaling.
  3. Low Latency Waveform Generation: The system can generate complex 6000-point waveforms in as little as 30 to 79 microseconds, depending on the pulse shape.

Conclusion

By virtualizing hardware resources and optimizing data transmission protocols, this software framework provides a scalable solution for managing the complex electronics required for superconducting quantum computing. The system successfully balances high-speed data throughput with precise signal correction, paving the way for larger-scale quantum processors and the eventual implementation of real-time feedback for fault-tolerant quantum computing.

Summary updated

Abstract

Being important parts of the superconducting quantum computer, the high-speed arbitrary waveform generator (AWG), ultraprecision dc source, and high-speed digitizer are used to manipulate the qubit. The complexity of an experimental setup increases rapidly as the number of qubits grows. Cumbersome instrument management, distortion of signals, and inefficiency of data transmission are gradually highlighted and become the bottlenecks in scaling up the number of qubits. In addition, fault-tolerant quantum computing has real-time feedback requirements of qubit states. To deal with these challenges, we propose an instrument management software design in this paper. The software maps the resources of separate instruments to a unified virtual instrument, achieving the scalability of the instruments. The processing and correction of signals are deployed on a server, which automatically corrects the distortion of the signals. By designing a multi-threaded mechanism for the AWG and a custom-defined data-link protocol for the digitizer, the efficiency of the software meets the requirements of experiments. Cooperating with the AWG and digitizer, the software provides the real-time feedback capability using an instruction compiler. Thus, the software meets the requirements of superconducting quantum computing, as demonstrated through its application to a 12-bit quantum experiment.
BibTeX
@article{guo2019control,
  author = {Cheng Guo and Futian Liang and Jin Lin and Yu Xu and Lihua Sun and Wei-Yue Liu and Sheng‐Kai Liao and Cheng-Zhi Peng},
  title = {{Control and Readout Software for Superconducting Quantum Computing}},
  journal = {IEEE Transactions on Nuclear Science},
  year = {2019},
  doi = {10.1109/tns.2019.2920337},
}
Figure