Installation#

Requirements#

torch-diffsim requires Python 3.8 or later and the following dependencies:

PyTorch >= 2.0.0
NumPy >= 1.20.0
SciPy >= 1.7.0
meshio >= 5.0.0
polyscope >= 1.3.0
matplotlib >= 3.5.0

Install from PyPI#

The easiest way to install torch-diffsim is via pip ![PyPI](https://img.shields.io/pypi/v/torch-diffsim?style=flat-square)

pip install torch-diffsim

This will automatically install all required dependencies.

IPC Installation#

The frictionless IPC solver uses the official IPC Toolkit as an optional CPU dependency:

pip install "torch-diffsim[ipc]"

The supported backend is ipctk>=1.6,<1.7. IPCImplicitEulerSolver requires CPU torch.float64 state and mesh tensors.

Install from Source#

For development or to get the latest features, you can install from source:

# Clone the repository
git clone https://github.com/Rishit-dagli/torch-diffsim
cd torch-diffsim

# Install in editable mode
pip install -e .

Development Installation#

If you want to contribute to torch-diffsim, install the development dependencies:

# Clone the repository
git clone https://github.com/Rishit-dagli/torch-diffsim
cd torch-diffsim

# Install with development dependencies
pip install -e ".[dev]"

This will install additional tools for testing and code formatting:

pytest >= 7.0.0
black >= 22.0.0
ruff >= 0.1.0

Verifying Installation#

You can verify your installation by running:

import diffsim
print(diffsim.__version__)

This should print the version number (e.g., 0.2.0).

GPU Support#

The solvers support CUDA when the mesh, material, and simulator are created on a CUDA device. To check whether CUDA is available:

import torch
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"CUDA device count: {torch.cuda.device_count()}")

The explicit and differentiable penalty solvers support CUDA.

Troubleshooting#

Import errors: If you encounter import errors, ensure that all dependencies are installed correctly:

pip install torch numpy scipy meshio polyscope matplotlib

PyTorch installation: For specific PyTorch installation instructions (e.g., with specific CUDA versions), visit the official PyTorch website.

Polyscope rendering issues: If you encounter issues with polyscope visualization, ensure you have proper OpenGL drivers installed on your system.

Installation

Contents