Scicomp

Scicomp is a cross-platform scientific computing framework that provides equivalent implementations of computational physics tools across Python, MATLAB, and Mathematica. The framework covers three main areas: quantum mechanics simulations, thermal transport modeling, and physics-informed machine learning.

The quantum simulation capabilities include Bell state preparation, variational quantum eigensolver (VQE), quantum approximate optimization algorithm (QAOA), and Jaynes-Cummings model dynamics. These implementations parallel and complement the quantum computing modules in qubeml, which provides educational Jupyter notebooks using Qiskit, Cirq, and PennyLane for VQE, error mitigation, and quantum kernel methods. GPU acceleration is available through CUDA and CuPy, enabling high-throughput parameter sweeps and large-scale matrix operations.

Physics-informed neural networks (PINNs) are included for solving partial differential equations, combining neural network function approximation with physical constraint enforcement. Spectral methods provide an alternative numerical approach for problems where frequency-domain decomposition is advantageous.

Scicomp serves as a shared computational foundation for several domain-specific projects. Maglogic uses Python and MATLAB for nanomagnetic logic simulation with OOMMF and MuMax3 backends. Spincirc employs MATLAB core solvers and Python analysis tools for spintronic device modeling using equivalent-circuit methods. Qmatsim combines Python-based DFT and MD workflows for strain engineering in 2D transition metal dichalcogenides. Qubeml extends the quantum computing theme with six educational tool modules spanning Qiskit, Cirq, PennyLane, PyTorch, scikit-learn, and Kwant. Helios contributes computational astrophysics and Bayesian inference research. The repository is hosted at github.com/alawein/scicomp and is under active development in the scientific-computing domain at P2 priority.