Source Code and Simulation Data: Topological Diffusivity Modulation and MHD Instability Control via PINNs (The DQ Framework)

This repository contains the PyTorch source code and the 2D Software-in-the-Loop simulation video supporting the manuscript "Topological Diffusivity Modulation and MHD Instability Control via Physics-Informed Neural Networks: The DQ Framework". The surrogate model demonstrates the active control of spatial diffusivity ($D \to 0$) using a Physics-Informed Neural Network (PINN). By modulating the electromagnetic tensor to induce a localized topological collapse, the system achieves absolute thermal isolation (0 K) of a solid core against a continuous hypersonic plasma flow at 15,000 K. This computational proof of concept (TRL-3) validates a novel approach to bypassing the thermal wall and mitigating Magnetohydrodynamic (MHD) instabilities without relying on passive ablative materials or brute-force magnetic containment.