Gaussian Models to Non-Gaussian Realms of Quantum Photonic Simulators

Quantum photonic simulators have emerged as indispensable tools for modeling and optimizing quantum photonic circuits, bridging the gap between theoretical models and experimental implementations. This review explores the landscape of photonic quantum simulation, focusing on the transition from Gaussian to non-Gaussian models and the computational challenges associated with simulating large-scale photonic systems. Gaussian states and operations, which enable efficient simulations through covariance matrices and phase-space representations, serve as the foundation for photonic quantum computing. However, non-Gaussian states crucial for universal quantum computation introduce significant computational complexity, requiring advanced numerical techniques such as tensor networks and high-performance GPU acceleration. We evaluate the leading photonic quantum simulators, including Strawberry Fields, Piquasso, QuTiP SimulaQron, Perceval, and QuantumOPtics.jl analyzing their capabilities in handling continuous-variable (CV) and discrete-variable (DV) quantum systems. Special attention is given to hardware-accelerated methods, including GPU-based tensor network approaches, machine learning integration, and hybrid quantum-classical workflows. Furthermore, we investigate noise modeling techniques, such as photon loss and dark counts, and their impact on simulation accuracy. As photonic quantum computing moves toward practical implementations, advancements in high-performance computing (HPC) architectures, such as tensor processing units (TPUs) and system-on-a-chip (SoC) solutions, are accelerating the field. This review highlights emerging trends, challenges, and future directions for developing scalable and efficient photonic quantum simulators.