IgANet-Elasticity

This repository contains elasticity example applications built on top of the IgANet core.

The core repository provides the spline, collocation, multipatch, constraint, and network infrastructure. This repository adds concrete elasticity examples, example configurations, reference runs, result export, and visualization scripts.

This README is written for newcomers. It explains:

• where the important files are
• how to install the required dependencies
• how to configure and build the examples
• how to run the executables
• how to visualize the generated results

Repository Layout

iganet-elasticity/
├── CMakeLists.txt
├── README.md
├── filedata/
│   ├── bone_simplified.xml
│   ├── five_patch_cantilever.xml
│   ├── two_cube_parametric.xml
│   └── two_patch_smoke.xml
├── results/
├── src/
│   ├── headers/
│   │   ├── lin_elasticity_multipatch_net.hpp
│   │   ├── lin_elasticity_net.hpp
│   │   └── lin_elasticity_utils.hpp
│   ├── utils/
│   │   ├── config.hpp
│   │   └── paths.hpp
│   ├── examples2D/
│   │   ├── singlePatch/
│   │   └── multiPatch/
│   └── examples3D/
│       ├── singlePatch/
│       └── multiPatch/
└── std_collocation_python/

Main Files

Example sources

• src/examples2D/singlePatch/iganet_lin_elasticity_2D.cxx
• src/examples2D/singlePatch/iganet_lin_elasticity_2D_optimized.cxx
• src/examples2D/multiPatch/iganet_lin_elasticity_2D_multipatch_parametric.cxx
• src/examples3D/singlePatch/iganet_lin_elasticity_3D.cxx
• src/examples3D/singlePatch/iganet_lin_elasticity_3D_optimized.cxx
• src/examples3D/multiPatch/iganet_lin_elasticity_3D_multipatch_bone.cxx
• src/examples3D/multiPatch/iganet_lin_elasticity_3D_multipatch_parametrized.cxx

Example configs

• src/examples2D/singlePatch/sim_config_2D_single_patch.json
• src/examples2D/multiPatch/sim_config_2D_multi_patch.json
• src/examples3D/singlePatch/sim_config_3D_single_patch.json
• src/examples3D/multiPatch/sim_config_3D_multi_patch_bone.json
• src/examples3D/multiPatch/sim_config_3D_multi_patch_parametrized.json

Visualization scripts

• src/examples2D/singlePatch/show_2D_single_patch.py
• src/examples2D/multiPatch/show_2D_multi_patch.py
• src/examples3D/singlePatch/show_3D_single_patch.py
• src/examples3D/multiPatch/show_3D_multi_patch_bone.py

Shared elasticity headers

• src/headers/lin_elasticity_net.hpp
• src/headers/lin_elasticity_multipatch_net.hpp
• src/headers/lin_elasticity_utils.hpp

Reference solver

• std_collocation_python/run_std_coll.py

Some examples call the Python reference solver automatically and append its data to the same JSON result file.

Relationship To The Core Repository

This repository depends on iganet.

The core is fetched with CMake FetchContent from GitHub. By default, CMake uses the repository and branch configured in CMakeLists.txt.

You can override the branch, tag, or commit at configure time:

cmake -S . -B build \
  -DTorch_DIR=$HOME/libtorch/share/cmake/Torch \
  -DIGANET_CORE_GIT_TAG=your_branch_or_commit

A local ../iganet checkout is not used automatically.

Prerequisites

You need:

• a C++20 compiler
• CMake >= 3.24
• Python
• LibTorch

For visualization, install these Python packages:

pip install numpy splinepy gustaf vedo

Installing LibTorch

Follow the same procedure as in the IgANet core README. The essential point is: download a LibTorch build that matches your compiler ABI and then point CMake to its Torch_DIR.

Example for Linux CPU builds, analogous to the core README:

wget https://download.pytorch.org/libtorch/cpu/libtorch-cxx11-abi-shared-with-deps-2.7.1%2Bcpu.zip -O libtorch.zip
unzip libtorch.zip -d $HOME/

Then configure with:

cmake -S . -B build -DTorch_DIR=$HOME/libtorch/share/cmake/Torch

If you use a different LibTorch install location, adjust Torch_DIR accordingly.

CPU Build (Default)

From the repository root:

cmake -S . -B build -DTorch_DIR=$HOME/libtorch/share/cmake/Torch
cmake --build build -j 16

This builds all example executables directly into build/.

If you only want one target:

cmake --build build -j 16 --target iganet_lin_elasticity_2D
cmake --build build -j 16 --target iganet_lin_elasticity_2D_optimized
cmake --build build -j 16 --target iganet_lin_elasticity_2D_multipatch_parametric
cmake --build build -j 16 --target iganet_lin_elasticity_3D
cmake --build build -j 16 --target iganet_lin_elasticity_3D_optimized
cmake --build build -j 16 --target iganet_lin_elasticity_3D_multipatch_bone
cmake --build build -j 16 --target iganet_lin_elasticity_3D_multipatch_parametrized

CUDA Build (Optional)

If you want CUDA, install a CUDA-enabled LibTorch build that matches your CUDA toolchain, then configure CMake with the matching CUDA settings.

Example:

cmake -S . -B build \
  -DTorch_DIR=/path/to/libtorch/share/cmake/Torch \
  -DCMAKE_CUDA_COMPILER=/path/to/nvcc \
  -DCMAKE_CUDA_ARCHITECTURES=75 \
  -DCUDA_TOOLKIT_ROOT_DIR=/path/to/cuda
cmake --build build -j 16

The exact paths and architecture depend on your machine. CPU build is the default and should be your starting point.

Which Config File Belongs To Which Example?

Each executable reads its own config file from the matching example folder.

• iganet_lin_elasticity_2D
  • src/examples2D/singlePatch/sim_config_2D_single_patch.json
• iganet_lin_elasticity_2D_optimized
  • src/examples2D/singlePatch/sim_config_2D_single_patch.json
• iganet_lin_elasticity_2D_multipatch_parametric
  • src/examples2D/multiPatch/sim_config_2D_multi_patch.json
• iganet_lin_elasticity_3D
  • src/examples3D/singlePatch/sim_config_3D_single_patch.json
• iganet_lin_elasticity_3D_optimized
  • src/examples3D/singlePatch/sim_config_3D_single_patch.json
• iganet_lin_elasticity_3D_multipatch_bone
  • src/examples3D/multiPatch/sim_config_3D_multi_patch_bone.json
• iganet_lin_elasticity_3D_multipatch_parametrized
  • src/examples3D/multiPatch/sim_config_3D_multi_patch_parametrized.json

In practice:

• edit the matching JSON config
• rebuild only if you changed C++ code
• if you changed only the config, rerun the executable

Running The Examples

Run the executables from the repository root:

2D single-patch

./build/iganet_lin_elasticity_2D
./build/iganet_lin_elasticity_2D_optimized

2D multi-patch

./build/iganet_lin_elasticity_2D_multipatch_parametric

3D single-patch

./build/iganet_lin_elasticity_3D
./build/iganet_lin_elasticity_3D_optimized

3D multi-patch

./build/iganet_lin_elasticity_3D_multipatch_bone
./build/iganet_lin_elasticity_3D_multipatch_parametrized

Where Results Are Written

The executables write JSON files into results/.

Typical files are:

• results/result_iganet_lin_elasticity_2D.json
• results/result_iganet_lin_elasticity_2D_optimized.json
• results/result_iganet_lin_elasticity_2D_multipatch_parametric.json
• results/result_iganet_lin_elasticity_3D.json
• results/result_iganet_lin_elasticity_3D_optimized.json
• results/result_iganet_lin_elasticity_3D_multipatch_bone.json
• results/result_iganet_lin_elasticity_3D_multipatch_parametrized.json

The exact JSON contents depend on the example. Depending on the case, a result file may contain:

• geometry and deformed control points
• displacements
• stresses or derived quantities
• collocation point data
• interface diagnostics
• optional reference data from the Python standard-collocation run

Do not assume that every result file contains every field.

Visualizing Results

Each show script already has a matching default result file path. In the normal case you can just run the script directly without passing an argument.

2D single-patch

python src/examples2D/singlePatch/show_2D_single_patch.py

2D multi-patch

python src/examples2D/multiPatch/show_2D_multi_patch.py

3D single-patch

python src/examples3D/singlePatch/show_3D_single_patch.py

3D multi-patch

python src/examples3D/multiPatch/show_3D_multi_patch_bone.py

If you explicitly want to visualize a different JSON file, you can still pass the result path manually, for example:

python src/examples2D/singlePatch/show_2D_single_patch.py \
  results/result_iganet_lin_elasticity_2D_optimized.json

The show scripts also write PNG screenshots into results/.

Suggested First Steps

If you are new to the repository, start in this order:

1. iganet_lin_elasticity_2D_optimized
2. iganet_lin_elasticity_2D
3. iganet_lin_elasticity_2D_multipatch_parametric
4. iganet_lin_elasticity_3D_optimized
5. iganet_lin_elasticity_3D
6. iganet_lin_elasticity_3D_multipatch_parametrized
7. iganet_lin_elasticity_3D_multipatch_bone

That keeps the early workflow simple and makes debugging easier.

Typical Newcomer Workflow

1. Build one small example

cmake --build build -j 16 --target iganet_lin_elasticity_2D_optimized
./build/iganet_lin_elasticity_2D_optimized
python src/examples2D/singlePatch/show_2D_single_patch.py \
  results/result_iganet_lin_elasticity_2D_optimized.json

2. Inspect the config

Open:

• src/examples2D/singlePatch/sim_config_2D_single_patch.json

Typical things to change:

• material parameters
• spline settings
• optimizer settings
• boundary conditions

3. Move to multipatch

cmake --build build -j 16 --target iganet_lin_elasticity_2D_multipatch_parametric
./build/iganet_lin_elasticity_2D_multipatch_parametric
python src/examples2D/multiPatch/show_2D_multi_patch.py

4. Move to 3D

cmake --build build -j 16 --target iganet_lin_elasticity_3D_optimized
./build/iganet_lin_elasticity_3D_optimized
python src/examples3D/singlePatch/show_3D_single_patch.py \
  results/result_iganet_lin_elasticity_3D_optimized.json

Notes About XML Geometry

Some examples support:

• geometry.mode = "parametric"
• geometry.mode = "xml"

In XML mode, geometries are typically loaded from filedata/, for example:

• filedata/two_cube_parametric.xml
• filedata/five_patch_cantilever.xml
• filedata/bone_simplified.xml

Common Places To Edit

If you want to change...

example-specific settings

Edit the matching sim_config_*.json.

one concrete example

Edit the corresponding .cxx file in src/examples2D/... or src/examples3D/....

shared single-patch elasticity logic

Edit:

• src/headers/lin_elasticity_net.hpp

shared multi-patch elasticity logic

Edit:

• src/headers/lin_elasticity_multipatch_net.hpp

shared config parsing

Edit:

• src/utils/config.hpp

visualization logic

Edit the matching show_*.py.

Troubleshooting

CMake cannot fetch the core repository

Check that the machine has network access and that IGANET_CORE_GIT_REPOSITORY and IGANET_CORE_GIT_TAG point to an existing repository and branch, tag, or commit.

CMake cannot find LibTorch

Check that Torch_DIR points to:

.../share/cmake/Torch

inside your LibTorch installation.

A target is not known

Re-run CMake configure:

cmake -S . -B build -DTorch_DIR=$HOME/libtorch/share/cmake/Torch

Then build again.

A show script cannot open the result file

Check that:

• the matching executable has been run
• the expected JSON file exists in results/
• you are using the correct show script for that example

Visualization fails because of missing Python packages

Install:

pip install numpy splinepy gustaf vedo

Final Remark

The shortest path into this repository is still:

• build iganet_lin_elasticity_2D_optimized
• inspect src/examples2D/singlePatch/sim_config_2D_single_patch.json
• run src/examples2D/singlePatch/show_2D_single_patch.py

That gives you the cleanest first overview of the full workflow.