Fast quantum circuit simulation using hardware accelerated general purpose libraries

TL;DR

This paper demonstrates that using general purpose GPU libraries like CuPy can significantly accelerate quantum circuit simulation, achieving up to 22x speedup over traditional C++ simulators, while maintaining ease of use in Python.

Contribution

The paper introduces a method leveraging CuPy, a general purpose CUDA library, to enhance quantum circuit simulation performance within the Python ecosystem.

Findings

2x speedup for supremacy circuits

22x speedup for quantum multipliers

Ease of use in Python with significant performance gains

Abstract

Quantum circuit simulators have a long tradition of exploiting massive hardware parallelism. Most of the times, parallelism has been supported by special purpose libraries tailored specifically for the quantum circuits. Quantum circuit simulators are integral part of quantum software stacks, which are mostly written in Python. Our focus has been on ease of use, implementation and maintainability within the Python ecosystem. We report the performance gains we obtained by using CuPy, a general purpose library (linear algebra) developed specifically for CUDA-based GPUs, to simulate quantum circuits. For supremacy circuits the speedup is around 2x, and for quantum multipliers almost 22x compared to state-of-the-art C++-based simulators.