Installing mini-conda: Difference between revisions

From Wiki Cours
Jump to navigation Jump to search
No edit summary
No edit summary
 
Line 1: Line 1:
To use numpy library and other common Python libraries install anaconda into your home directory:
To use numpy library and other common Python libraries install anaconda into your home directory
(here we download the pyton3 latest version, you may check the repository webpage for other versions):
<pre>
<pre>
wget http://repo.continuum.io/miniconda/Miniconda-latest-Linux-x86_64.sh -O miniconda.sh
wget http://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda.sh
chmod +x miniconda.sh
chmod +x miniconda.sh
./miniconda.sh -b
./miniconda.sh -b
Line 8: Line 9:
conda create --yes -n dev pip scipy numpy matplotlib hdf5 h5py ipython mpi4py jupyter
conda create --yes -n dev pip scipy numpy matplotlib hdf5 h5py ipython mpi4py jupyter
</pre>
</pre>
'''Attention''': the "miniconda" name of the directory may vary slightly over time, check it.
'''Attention''': the "miniconda" name of the directory may vary slightly over versions, check it.


Then add this line into your ~/.bashrc file:
Then add this line into your ~/.bashrc file:
Line 22: Line 23:
source activate dev
source activate dev
</pre>
</pre>
Now, virtual environment contains its own compilers. If you need to run a Python script in parallel,
Now, virtual environment contains its own compilers.
all needs to be compiled using these compilers, unless you manage to redirect to intel compilers or mkl libraries on orlando. An example of a software which needs to be compiled in this way is the molecular dynamics simulation software LAMMPS,
which can be called from Python and run in parallel.

Latest revision as of 11:05, 22 September 2018

To use numpy library and other common Python libraries install anaconda into your home directory (here we download the pyton3 latest version, you may check the repository webpage for other versions):

wget http://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda.sh
chmod +x miniconda.sh
./miniconda.sh -b
export PATH=~/miniconda/bin:$PATH
conda update --yes conda
conda create --yes -n dev pip scipy numpy matplotlib hdf5 h5py ipython mpi4py jupyter

Attention: the "miniconda" name of the directory may vary slightly over versions, check it.

Then add this line into your ~/.bashrc file:

PATH=$HOME/miniconda/bin:$PATH

The following line activates the virtual environment named "dev", which effectively moves you to a new folder with its own binary files, compilers, etc. The line needs to be called before running a script using the virtual environment. The line can be put either into you .bashrc file or into the (.sge) submission script.

source activate dev

Now, virtual environment contains its own compilers.