All included operations work on varying data types and are implemented both for CPU and GPU.
```
To avoid the hazzle of creating [`torch.sparse_coo_tensor`](https://pytorch.org/docs/stable/torch.html?highlight=sparse_coo_tensor#torch.sparse_coo_tensor), this package defines operations on sparse tensors by simply passing `index` and `value` tensors as arguments ([with same shapes as defined in PyTorch](https://pytorch.org/docs/stable/sparse.html)).
Note that only `value` comes with autograd support, as `index` is discrete and therefore not differentiable.
**Note:** Binaries of older versions are also provided for PyTorch 1.4.0, PyTorch 1.5.0, PyTorch 1.6.0, PyTorch 1.7.0/1.7.1, PyTorch 1.8.0/1.8.1, PyTorch 1.9.0, and PyTorch 1.10.0/1.10.1/1.10.2 (following the same procedure).
For older versions, you might need to explicitly specify the latest supported version number in order to prevent a manual installation from source.
You can look up the latest supported version number [here](https://data.pyg.org/whl).
### From source
Ensure that at least PyTorch 1.7.0 is installed and verify that `cuda/bin` and `cuda/include` are in your `$PATH` and `$CPATH` respectively, *e.g.*:
If you want to additionally build `torch-sparse` with METIS support, *e.g.* for partioning, please download and install the [METIS library](https://web.archive.org/web/20211119110155/http://glaros.dtc.umn.edu/gkhome/metis/metis/download) by following the instructions in the `Install.txt` file.
Note that METIS needs to be installed with 64 bit `IDXTYPEWIDTH` by changing `include/metis.h`.
Afterwards, set the environment variable `WITH_METIS=1`.
Then run:
```
pip install torch-scatter torch-sparse
```
When running in a docker container without NVIDIA driver, PyTorch needs to evaluate the compute capabilities and may fail.
In this case, ensure that the compute capabilities are set via `TORCH_CUDA_ARCH_LIST`, *e.g.*:
static auto registry = torch::RegisterOperators().op("torch_sparse::cuda_version", &sparse::cuda_version); -> static auto registry = torch::RegisterOperators().op("torch_sparse::cuda_version", []{return sparse::cuda_version();});
pytest
```
```
## C++ API
## 参考资料
```shell
`torch-sparse` also offers a C++ API that contains C++ equivalent of python models.
https://github.com/rusty1s/pytorch_spline_conv
For this, we need to add `TorchLib` to the `-DCMAKE_PREFIX_PATH` (*e.g.*, it may exists in `{CONDA}/lib/python{X.X}/site-packages/torch` if installed via `conda`):
All included operations work on varying data types and are implemented both for CPU and GPU.
To avoid the hazzle of creating [`torch.sparse_coo_tensor`](https://pytorch.org/docs/stable/torch.html?highlight=sparse_coo_tensor#torch.sparse_coo_tensor), this package defines operations on sparse tensors by simply passing `index` and `value` tensors as arguments ([with same shapes as defined in PyTorch](https://pytorch.org/docs/stable/sparse.html)).
Note that only `value` comes with autograd support, as `index` is discrete and therefore not differentiable.
## Installation
### Anaconda
**Update:** You can now install `pytorch-sparse` via [Anaconda](https://anaconda.org/pyg/pytorch-sparse) for all major OS/PyTorch/CUDA combinations 🤗
Given that you have [`pytorch >= 1.8.0` installed](https://pytorch.org/get-started/locally/), simply run
```
conda install pytorch-sparse -c pyg
```
#### Note: Conda packages are not published for PyTorch 1.12 yet
### Binaries
We alternatively provide pip wheels for all major OS/PyTorch/CUDA combinations, see [here](https://data.pyg.org/whl).
#### PyTorch 1.12
To install the binaries for PyTorch 1.12.0, simply run
**Note:** Binaries of older versions are also provided for PyTorch 1.4.0, PyTorch 1.5.0, PyTorch 1.6.0, PyTorch 1.7.0/1.7.1, PyTorch 1.8.0/1.8.1, PyTorch 1.9.0, and PyTorch 1.10.0/1.10.1/1.10.2 (following the same procedure).
For older versions, you might need to explicitly specify the latest supported version number in order to prevent a manual installation from source.
You can look up the latest supported version number [here](https://data.pyg.org/whl).
### From source
Ensure that at least PyTorch 1.7.0 is installed and verify that `cuda/bin` and `cuda/include` are in your `$PATH` and `$CPATH` respectively, *e.g.*:
If you want to additionally build `torch-sparse` with METIS support, *e.g.* for partioning, please download and install the [METIS library](https://web.archive.org/web/20211119110155/http://glaros.dtc.umn.edu/gkhome/metis/metis/download) by following the instructions in the `Install.txt` file.
Note that METIS needs to be installed with 64 bit `IDXTYPEWIDTH` by changing `include/metis.h`.
Afterwards, set the environment variable `WITH_METIS=1`.
Then run:
```
pip install torch-scatter torch-sparse
```
When running in a docker container without NVIDIA driver, PyTorch needs to evaluate the compute capabilities and may fail.
In this case, ensure that the compute capabilities are set via `TORCH_CUDA_ARCH_LIST`, *e.g.*:
`torch-sparse` also offers a C++ API that contains C++ equivalent of python models.
For this, we need to add `TorchLib` to the `-DCMAKE_PREFIX_PATH` (*e.g.*, it may exists in `{CONDA}/lib/python{X.X}/site-packages/torch` if installed via `conda`):
