STABSim: A Parallelized Clifford Simulator with Features Beyond Direct Simulation (2507.03092v1)

Abstract: Efficient stabilizer-based computation is a key task in quantum computing as it provides a classically tractable insight into quantum algorithms, which otherwise quickly become impossible at scale on a classical device. Formulations in quantum error correction, hardware verification, hybrid computing, and more, all benefit from the ability of Clifford gates to be accurately evaluated by classical computers. However, current stabilizer simulators have significant limitations in scaling and use-case beyond direct simulation. To address this, we present a GPU-accelerated tableau stabilizer simulator that scales efficiently in direct circuit simulation, and supports a range of stabilizer-based tasks. We show how the simulator is used to quickly calculate Pauli commutation groupings between Pauli strings to reduce the number of circuit runs in quantum chemistry problems. We also present a Clifford+T circuit transpiler based on STABSim, which uses the simulator framework to track relations in non-Clifford rotations as Clifford gates are absorbed into the measurement step of the circuit. The T rotation layers left behind are then further optimized in the simulator structure to create a reduced circuit containing only the minimum non-Clifford rotations, improving the time to simulate and reducing the cost to run on real quantum devices.