Verifying Quantum Advantage Experiments with Multiple Amplitude Tensor Network Contraction

Area of Science:

• Quantum computing
• Computational physics
• Supercomputing

Background:

• Quantum supremacy experiments, like Google's Sycamore, present significant classical verification challenges due to escalating computational costs.
• Direct classical verification of quantum supremacy experiments is computationally intensive, limiting its feasibility.

Purpose of the Study:

• To provide a direct classical verification of a quantum supremacy experiment.
• To demonstrate a novel simulation capability for complex quantum systems.

Main Methods:

• Utilized a new-generation Sunway supercomputer for direct verification.
• Employed a multiple-amplitude tensor network contraction algorithm exploiting classical "store-and-compute" advantages.
• Implemented a fused tensor network contraction algorithm for enhanced efficiency on heterogeneous architectures.

Main Results:

• Successfully computed 3×10^6 exact amplitudes for experimentally generated bitstrings.
• Achieved a cross-entropy benchmarking fidelity of 0.191%, closely matching the estimated value of 0.224%.
• Demonstrated a significant leap in classical simulation capability for quantum systems.

Conclusions:

• The developed methods enable efficient classical verification of quantum supremacy experiments.
• This approach has broad implications for solving quantum many-body problems, statistical problems, and combinatorial optimization.
• Highlights the potential of supercomputers in advancing quantum information science research.