PyTorch Extension Library of Optimized Scatter Operations

pytorch, scatter, segment, gather
pip install torch-scatter==2.0.4


PyTorch Scatter

PyPI Version Build Status Docs Status Code Coverage


This package consists of a small extension library of highly optimized sparse update (scatter and segment) operations for the use in PyTorch, which are missing in the main package. Scatter and segment operations can be roughly described as reduce operations based on a given "group-index" tensor. Segment operations require the "group-index" tensor to be sorted, whereas scatter operations are not subject to these requirements.

The package consists of the following operations with reduction types "sum"|"mean"|"min"|"max":

In addition, we provide the following composite functions which make use of scatter_* operations under the hood: scatter_std, scatter_logsumexp, scatter_softmax and scatter_log_softmax.

All included operations are broadcastable, work on varying data types, are implemented both for CPU and GPU with corresponding backward implementations, and are fully traceable.


Ensure that at least PyTorch 1.3.0 is installed and verify that cuda/bin and cuda/include are in your $PATH and $CPATH respectively, e.g.:

$ python -c "import torch; print(torch.__version__)"
>>> 1.3.0

$ echo $PATH
>>> /usr/local/cuda/bin:...

$ echo $CPATH
>>> /usr/local/cuda/include:...

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

export TORCH_CUDA_ARCH_LIST = "6.0 6.1 7.2+PTX 7.5+PTX"


If you are installing this on Windows specifically, you will need to point the setup to your Visual Studio installation for some neccessary libraries and header files. To do this, add the include and library paths of your installation to the path lists in as described in the respective comments in the code.

If you are running into any installation problems, please create an issue. Be sure to import torch first before using this package to resolve symbols the dynamic linker must see.


import torch
from torch_scatter import scatter_max

src = torch.tensor([[2, 0, 1, 4, 3], [0, 2, 1, 3, 4]])
index = torch.tensor([[4, 5, 4, 2, 3], [0, 0, 2, 2, 1]])

out, argmax = scatter_max(src, index, dim=-1)
tensor([[0, 0, 4, 3, 2, 0],
        [2, 4, 3, 0, 0, 0]])

tensor([[5, 5, 3, 4, 0, 1]
        [1, 4, 3, 5, 5, 5]])

Running tests

python test