解决PyG (PyTorch Geometric)库安装以及import时sklearn报错等问题

解决PyG (PyTorch Geometric)库安装以及import时sklearn报错等问题

2021-3-26 By Junwu Chen at CAS IPE, Beijing

1.安装pytorch

由于我们组服务器的linux系统版本太老，glibc版本低于2.17且没有root权限，所以我通过conda的方法安装pytorch（安装PyG前需要先安装torch）。conda安装pytorch的方法可以见我的博文。

2. 安装PyG

安装方法可见官方手册。
如我输入的安装命令为：

pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.6.0+cpu.html
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-1.6.0+cpu.html
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-1.6.0+cpu.html
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.6.0+cpu.html
pip install torch-geometric

但是在最后一步（安装torch-geometric）时会报错，原因是其所依赖的numba库安装不上，我的解决方法是采用conda安装numba：

conda install numba

但是安装好后，在import torch_geometric时会报错：

Traceback (most recent call last):
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/sklearn/__check_build/__init__.py", line 44, in <module>
    from ._check_build import check_build  # noqa
ImportError: dlopen: cannot load any more object with static TLS

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/torch_geometric/__init__.py", line 2, in <module>
    import torch_geometric.nn
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/torch_geometric/nn/__init__.py", line 10, in <module>
    from .models import *  # noqa
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/torch_geometric/nn/models/__init__.py", line 2, in <module>
    from .node2vec import Node2Vec
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/torch_geometric/nn/models/node2vec.py", line 5, in <module>
    from sklearn.linear_model import LogisticRegression
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/sklearn/__init__.py", line 81, in <module>
    from . import __check_build  # noqa: F401
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/sklearn/__check_build/__init__.py", line 46, in <module>
    raise_build_error(e)
  File "/home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/sklearn/__check_build/__init__.py", line 41, in raise_build_error
    %s""" % (e, local_dir, ''.join(dir_content).strip(), msg))
ImportError: dlopen: cannot load any more object with static TLS
___________________________________________________________________________
Contents of /home/jwchen/software/miniconda3/envs/pyg/lib/python3.6/site-packages/sklearn/__check_build:
__init__.py               __pycache__               _check_build.cpython-36m-x86_64-linux-gnu.so
setup.py
___________________________________________________________________________
It seems that scikit-learn has not been built correctly.

If you have installed scikit-learn from source, please do not forget
to build the package before using it: run `python setup.py install` or
`make` in the source directory.

If you have used an installer, please check that it is suited for your
Python version, your operating system and your platform.

这原因主要来自sklearn库的问题。

解决方法1：

安装低版本的sklearn

pip uninstall scikit-learn torch-geometric
pip install scikit-learn==0.20.3
pip install torch-geometric

解决方法2：

conda安装sklearn

pip uninstall scikit-learn torch-geometric
conda install scikit-learn
pip install torch-geometric