Source code for reproducing the numerical results of “On the metric field description of finite plasticity”
Description
Source code for reproducing the numerical results of the paper “On the metric field description of finite plasticity”. This repository contains the supplementary code and materials accompanying “On the metric field description of finite plasticity.” It supports the numerical implementation and symbolic verification of the material-metric formulation developed in the paper. The numerical component provides a C++ Abaqus UMAT for finite-strain J₂ plasticity with double-Voce hardening, eleven paired element-level loading tests, and reference and UMAT indentation cases. Input files, execution scripts, and post-processing tools enable comparisons with a conventional Abaqus plasticity model through documented Linux and Windows workflows. The symbolic component contains a Cadabra notebook and independent SymPy scripts for verifying relations among the lattice-frame connection, dislocation density, and Riemannian curvature. A unified supplementary document is supplied as LaTeX source and a compiled PDF, together with comparison figures, summary data, and reproduction instructions. Running the numerical analyses requires a licensed Abaqus installation and the dependencies specified in the documentation; raw solver outputs are not included. The independent SymPy checks require neither Abaqus nor Cadabra.
Files
Steps to reproduce
1. Extract the archive Extract the complete ZIP and preserve its directory structure. The package root contains README.md, supplementary/, umat/, and dislocation/. Numerical analyses and symbolic checks can be reproduced independently. Raw Abaqus solver outputs are not included. 2. Prepare the numerical environment The Windows workflow requires Abaqus 2025 Standard and CAE with a valid licence, x64 MSVC Build Tools with Windows SDK, Eigen headers, and external Python 3.11 or later with NumPy and Matplotlib. Intel Fortran is not required. Abaqus Python is used separately for ODB access. From umat/, configure the paths as described in windows/README.md. Set EIGEN_INC to the directory containing Eigen/Dense. Use UMAT_PYTHON to select external Python, and ABAQUS_COMMAND and UMAT_VCVARS if the Abaqus launcher or compiler setup cannot be discovered. Install Python dependencies with the selected interpreter: python -m pip install -r windows/requirements.txt 3. Run the numerical workflow From umat/, run in PowerShell: .\windows\run.ps1 all --parallel 2 --cpus 6 --memory "6 gb" This builds the standalone UMAT, runs eleven reference element tests and eleven UMAT counterparts, solves both indentation cases, and extracts and plots their results. Alternatively, run the build, ut, case, and post stages separately. Do not repeat these stages after all unless a rerun is intended. Input datacheck alone does not reproduce the simulations. For the Linux workflow, follow umat/README.md, including compiler and Eigen configuration, both element-test configurations, indentation runs, and post-processing. 4. Inspect the numerical results Element histories are written to umat/UT/results/. Windows element-comparison figures are in umat/windows/results/element_figures/. Indentation outputs are in each case's results/ directory. Confirm successful solver completion and inspect logs and extraction summaries. Compare the regenerated results with the supplementary figures, summary CSV, and case RESULTS.md files. Small numerical differences may occur across platforms or parallel runs; identical file hashes are not required. 5. Run the independent symbolic checks From dislocation/, run: python -m pip install -r python/requirements.txt python python/run_checks.py The tested dependency is SymPy 1.14.0. Successful execution ends with “All independent Python checks passed.” Do not disable Python assertions with -O or -OO. 6. Run the Cadabra derivation With Cadabra 2.5.14 installed, run from dislocation/: cadabra2-cli -q -w cadabra/dislocation_connection_curvature_nye.cdb Alternatively, open the corresponding .cnb notebook and execute its input cells in order. All thirteen checks should pass. 7. Compile the supplement The compiled document is supplied as supplementary/main.pdf. To rebuild it, run “pdflatex main.tex” three times from supplementary/. Keep the sections/ and Figures/ directories alongside main.tex.
Institutions
- Tsinghua UniversityBeijing, Beijing