Python 在Conda UnsatifiableError消息中找不到冲突

试图在Docker中的conda中安装pytorch，但出现不满意的错误。但是，我无法在错误消息中找到任何实际冲突，或者我可能无法正确理解它。使用的Docker映像是nvidia/cuda:10.1-cudnn7-devel-ubuntu18.04。如果我不指定要安装的pytorch版本，就不会有问题。但我确实需要这个版本的pytorch来编写一些遗留代码

这是conda命令和错误消息

(Python3.6) root@0cb9aad73116:/# conda install -c pytorch pytorch=0.3.1
Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: |
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed

UnsatisfiableError: The following specifications were found to be incompatible with each other:



Package libgcc-ng conflicts for:
python==3.6.9 -> libgcc-ng[version='>=7.3.0']
zlib -> libgcc-ng[version='>=7.2.0|>=7.3.0']
tk -> libgcc-ng[version='>=7.2.0|>=7.3.0']
xz -> libgcc-ng[version='>=7.2.0']
pytorch=0.3.1 -> libgcc-ng[version='>=5.4.0']
libedit -> libgcc-ng[version='>=7.2.0|>=7.3.0']
libffi -> libgcc-ng[version='>=7.2.0']
ncurses -> libgcc-ng[version='>=7.2.0|>=7.3.0']
sqlite -> libgcc-ng[version='>=7.2.0|>=7.3.0']
readline -> libgcc-ng[version='>=7.2.0|>=7.3.0']
openssl -> libgcc-ng[version='>=7.2.0|>=7.3.0']
Package libstdcxx-ng conflicts for:
ncurses -> libstdcxx-ng[version='>=7.2.0|>=7.3.0']
pytorch=0.3.1 -> libstdcxx-ng[version='>=5.4.0']
python==3.6.9 -> libstdcxx-ng[version='>=7.3.0']
libffi -> libstdcxx-ng[version='>=7.2.0']
Package xz conflicts for:
python==3.6.9 -> xz[version='>=5.2.4,<6.0a0']
Package cudnn conflicts for:
pytorch=0.3.1 -> cudnn[version='7.0.*|>=7.0.5,<=8.0a0']
Package nccl conflicts for:
pytorch=0.3.1 -> nccl[version='<2']
Package setuptools conflicts for:
wheel -> setuptools
pip -> setuptools
Package zlib conflicts for:
tk -> zlib[version='>=1.2.11,<1.3.0a0']
python==3.6.9 -> zlib[version='>=1.2.11,<1.3.0a0']
Package ca-certificates conflicts for:
openssl -> ca-certificates
Package libffi conflicts for:
python==3.6.9 -> libffi[version='>=3.2.1,<4.0a0']
Package cudatoolkit conflicts for:
pytorch=0.3.1 -> cudatoolkit[version='8.*|8.0.*']
Package _libgcc_mutex conflicts for:
libgcc-ng -> _libgcc_mutex=[build=main]
Package numpy conflicts for:
pytorch=0.3.1 -> numpy[version='>=1.11|>=1.11.3,<2.0a0']
Package openssl conflicts for:
python==3.6.9 -> openssl[version='>=1.1.1c,<1.1.2a']
Package libedit conflicts for:
sqlite -> libedit[version='>=3.1.20170329,<3.2.0a0|>=3.1.20181209,<3.2.0a0']
Package ncurses conflicts for:
python==3.6.9 -> ncurses[version='>=6.1,<7.0a0']
readline -> ncurses[version='6.0.*|>=6.0,<7.0a0|>=6.1,<7.0a0']
libedit -> ncurses[version='6.0.*|>=6.1,<7.0a0']
Package cffi conflicts for:
pytorch=0.3.1 -> cffi
Package sqlite conflicts for:
python==3.6.9 -> sqlite[version='>=3.29.0,<4.0a0']
Package * conflicts for:
pytorch=0.3.1 -> *[track_features=cuda90]
Package tk conflicts for:
python==3.6.9 -> tk[version='>=8.6.8,<8.7.0a0']
Package wheel conflicts for:
pip -> wheel
Package certifi conflicts for:
setuptools -> certifi[version='>=2016.09|>=2016.9.26']
Package mkl conflicts for:
pytorch=0.3.1 -> mkl[version='>=2018|>=2018.0.2,<2019.0a0']
Package readline conflicts for:
python==3.6.9 -> readline[version='>=7.0,<8.0a0']
Package pip conflicts for:
python==3.6.9 -> pip

如果使用Docker image

continumio/miniconda3:4.7.12

，则仍会出现类似错误

docker run -it nvidia/cuda:10.1-cudnn7-devel-ubuntu18.04 bash
apt-get update
apt install wget -y
wget https://repo.continuum.io/archive/Anaconda3-4.3.0-Linux-x86_64.sh 
bash Anaconda3-4.3.0-Linux-x86_64.sh
[Enter]
yes
[Enter]
yes
. /root/.bashrc
conda install mkl
y
conda install -c pytorch pytorch=0.3.1
y
conda list|grep pytorch

pytorch 0.3.1 py36_cuda8.0.61_cudnn7.1.2_3 pytorch

root@dacfe958940b：/#bash
（基数）root@dacfe958940b：/#python
Python 3.6.3 | Anaconda自定义（64位）|（默认值，2017年10月13日，12:02:49）
linux上的[GCC 7.2.0]
有关详细信息，请键入“帮助”、“版权”、“信用证”或“许可证”。
进口火炬
打印（火炬版本）
0.3.1.3

规范：-pytorch=0.3.1->python[version='>=2.7，=3.5，=3.6，@AAber我使用python 3.6.9，所以我想它满足了pytorch的要求？谢谢。Miniconda3-4.3.11-Linux-x86_64.sh也可以工作。我想知道原因是什么。pytorch 0.3.1软件包没有很好地开发，或者conda中有不推荐的东西？@W332 pytorch 0.3.1非常旧，所以它不支持cuda 10.1。@jodag“link”如果我使用非cuda docker映像，例如

continuMio/miniconda3:4.7.12

，将出现类似错误。因此，这可能不是由于docker提供的cuda版本造成的。

docker run -it nvidia/cuda:10.1-cudnn7-devel-ubuntu18.04 bash
apt-get update
apt install wget -y
wget https://repo.continuum.io/archive/Anaconda3-4.3.0-Linux-x86_64.sh 
bash Anaconda3-4.3.0-Linux-x86_64.sh
[Enter]
yes
[Enter]
yes
. /root/.bashrc
conda install mkl
y
conda install -c pytorch pytorch=0.3.1
y
conda list|grep pytorch