Large-scale quantum device benchmarking via LXEB with particle-number-conserving random quantum circuits

Linear cross-entropy benchmarking (LXEB) with random quantum circuits is a standard method for evaluating quantum computers. However, LXEB requires classically simulating the ideal output distribution of a given quantum circuit with high numerical precision, which becomes infeasible beyond approximately 50 qubits, even on state-of-the-art supercomputers. As a result, LXEB cannot be directly applied to evaluate large-scale quantum devices, which now exceed 100 qubits and continue to grow rapidly in size. To address this limitation, we introduce a constraint known as particle-number conservation into the random quantum circuits used for benchmarking. This restriction significantly reduces the size of the Hilbert space for a fixed particle number, enabling classical simulations of circuits with over 100 qubits when the particle number is <a:math xmlns:a="http://www.w3.org/1998/Math/MathML"> <a:mrow> <a:mi>O</a:mi> <a:mo>(</a:mo> <a:mn>1</a:mn> <a:mo>)</a:mo> </a:mrow> </a:math> . Furthermore, we propose a modified version of LXEB, called MLXEB, which enables fidelity estimation under particle-number-conserving dynamics. Through numerical simulations, we investigate the conditions under which MLXEB provides accurate fidelity estimates.