TFSolver: a numerical Python toolkit for parallel electromagnetic calculation of planar multilayer thin films at...

TFSolver is a Python toolkit for the electromagnetic calculation of planar isotropic and anisotropic multilayer thin films. This toolkit uses the PyTorch library to implement the 4 × 4 matrix method. One of the key advantages of TFSolver is its ability to perform parallel simulations for multilayer thin-film stacks across a broad spectral range and various incident angles, significantly speeding up the simulation process. Additionally, it supports GPU acceleration, further improving computational efficiency. Another notable feature is its support for automatic differentiation, which enables gradient-based optimization tasks. Furthermore, TFSolver can be integrated with deep learning algorithms, supporting emerging applications such as physics-guided inverse design. Here, we first introduce the formulation and the usage of TFSolver, compare it with existing toolkits and software, and validate its computational accuracy and efficiency. With its accelerated computational performance and automatic differentiation, TFSolver is a valuable toolkit for the characterization and design of various isotropic and anisotropic materials and multilayer thin-film devices.