Software & Projects :: Axect's Blog — Mathematics, Physics and Computations

Open-source software and research projects by Tae-Geun Kim (Axect).

🚀 Active Projects

Peroxide

Rust numerical computing library Stars Crates.io Downloads

Comprehensive numerical computing library for Rust, providing functionality comparable to NumPy/SciPy in Python. Core infrastructure for scientific computing research.

Key Features:

Linear algebra with BLAS/LAPACK integration
Optimization algorithms (Gradient Descent, Levenberg-Marquardt, etc.)
Numerical integration & ODE/PDE solvers
Statistical distributions & special functions
DataFrame with multiple I/O formats (CSV, NetCDF, JSON, Parquet)
FFI support for C, Fortran, and Python

Tech Stack: SIMD, BLAS, LAPACK, proc-macro metaprogramming

Quick Install:

cargo add peroxide

Links: GitHub | Crates.io | Documentation

Radient

Rust automatic differentiation library Stars Version Downloads

Experimental reverse-mode automatic differentiation library with arena-based memory management for research and prototyping.

Key Features:

Efficient reverse-mode AD with tape-based gradient computation
Arena allocation for memory safety and performance
Integration with Peroxide for scientific computing workflows

Quick Install:

cargo add radient

Links: GitHub | Crates.io | Documentation

DeeLeMa

Deep Learning for Mass Estimation

Deep learning framework for particle physics mass estimation using missing information search. Published in Physical Review Research.

Research Impact:

Paper: Phys. Rev. Research 5, 043186 (2023)
Preprint: arXiv:2212.12836

Tech Stack: PyTorch, PyTorch Lightning, Wandb (hyperparameter tuning), Tensorboard

Links: GitHub | Paper | arXiv

Neural Hamilton

Can AI Understand Hamiltonian Mechanics? Stars

Official implementation of operator learning for Hamiltonian mechanics. Explores whether AI can truly understand physical dynamics through four neural architectures (DeepONet, TraONet, VaRONet, MambONet).

Research Impact:

Paper: arXiv:2410.20951 - Tae-Geun Kim, Seong Chan Park (2024)
MambONet consistently outperformed RK4 solvers and competing architectures

Key Features:

Novel algorithm for generating physically plausible potentials using Gaussian Random Fields
Four neural network architectures for operator learning
Multi-language implementation (Python, Rust, Julia)
CUDA-compatible for GPU acceleration

Tech Stack: PyTorch, CUDA, Rust (numerical backend), Julia (visualization)

Links: GitHub | arXiv

PyTorch Template

Flexible PyTorch template for ML experiments Stars

Modular PyTorch project template designed for reproducible machine learning experiments with YAML-based configuration.

Key Features:

YAML-based experiment configuration for easy setup
Multiple random seed support for robust experimentation
Device selection (CPU/GPU) and learning rate scheduling
Clean modular structure for extensibility

Links: GitHub

Quantum Algorithms

Quantum computing algorithm implementations Stars

Comprehensive collection of quantum algorithms implemented in multiple frameworks (Pennylane, RustQIP, Qiskit, Cirq) with interactive Jupyter notebooks.

Tech Stack: Pennylane, RustQIP, Qiskit, Cirq, Jupyter

Links: GitHub