This repository contains the code and data accompanying the numerical experiments for the paper "Generalizing Riemann Curvature to Regge Metrics". The implementation is based on the open-source Finite Element library NGSolve (www.ngsolve.org). To generate the data, we used NGSolve-version NGSolve-6.2.2506-203-gf6f7fcf3e. Overview of the scripts: - generate_test_mesh.py: generates structured 3D meshes, applies a random interior perturbation, and writes the meshes plus mesh diameters. - example_manifolds.py: defines the exact metrics and curvature tensors used as reference solutions. - compute_error.py: computes the H^{-2} error norm and stores intermediate grid functions. - test_convergence.py: runs the convergence experiments and writes tables to CSV/LaTeX. - prepare_data.py: formats convergence CSV files into LaTeX tables. Data and meshes: - meshes/: pre-generated meshes used in the experiments (published to avoid reliance on random seeds). - data/: raw pickles (.pkl) and generated tables (.csv and .tex). Typical workflow: 1) (Optional) Regenerate meshes with generate_test_mesh.py. 2) Run test_convergence.py to produce convergence data in data/. 3) Run prepare_data.py to format LaTeX tables for the paper.