Tensor Networks for Quantum-Inspired Simulations

Abstract

Quantum algorithms have the potential to accelerate computation and reduce memory requirements on advanced quantum computers. However, current hardware limitations hinder their application to complex problems. In this work, we investigate a promising approach that bypasses the need for quantum hardware by leveraging tensor networks to simulate quantum algorithms on classical computers. We assess the performance of quantum-inspired simulators relative to classical methods in terms of memory, runtime, and accuracy. Our results demonstrate that quantum-inspired simulators can surpass their classical counterparts in accuracy while using less than half the memory. Additionally, we show that operators based on higher-precision approximations can reduce errors in quantum-inspired simulations without compromising memory requirements. Finally, we explore the capability of quantum-inspired simulators to address memory-intensive problems beyond the reach of conventional algorithms.