MechanicsDSL is a production-ready, enterprise-grade computational physics framework that enables physicists, researchers, and educators to define physical systems using a LaTeX-inspired syntax and automatically generates high-performance simulations.

Lagrangian and Hamiltonian Neural Ordinary Differential Equations (NODEs)

Built on peer-reviewed research accepted at IEEE DSA 2025: "Hamiltonian Neural Networks for Robust Out-of-Time Credit Scoring: Empirical Validation and Temporal Stability Analysis"

Geometric invariant probes that certify quasi-symplectic structure in neural network training. Reveals that BatchNorm breaks the underlying geometry while GroupNorm preserves it.

UC3 — Pseudo-Hamiltonian neural networks for physics-preserving dynamics (NAIC WP7)

psipy is a Python library for advanced computational physics and symbolic mathematics. It bridges symbolic representation with numerical simulation, focusing on pseudo-differential operators and spectral methods.

A collection of original supplementary articles on various Physics and Math topics (WiP)

Exact symbolic computation of Lie algebras generated by pairwise Hamiltonians under Poisson brackets, for the N-body problem. Headline N=3 sequence 3, 6, 17, 116 is universal across potentials (1/r, 1/r², log, Yukawa, …), spatial dimensions, mass ratios (incl. Tsygvintsev cases), and physical regimes (gravity, atomic, plasma, prime distributions).