Care and Feeding of a Python Environment

This is a blog post that I continue to update that discusses how I setup a Python Anaconda environment. As these
processes change, and I discover new things, I will update this post. These are my notes for how I setup a new Python environment for myself. I also refer students fom my course here for more advanced Python setup.

Environments

List your environments:

conda env list

Select an environment (Linux, OS X):

source activate tensorflow

Select an environment (Windows):

activate tensorflow

Deactivate Environment (Linux, OS X)

source deactivate

Deactivate Environment (Windows, Linux, OS X)

deactivate

Create/Delete Environment

Create an environment:

conda create --name tensorflow python=3.6

Delete an environment:

conda remove --name tensorflow --all

Allow Jupyter Notebook to see Environment

activate tensorflow
python -m ipykernel install --user --name tensorflow --display-name "Python (tensorflow)"

Update/install

Update app packages:

conda update --all

Update one package:

pip install tensorflow --upgrade

Update Python to latest point version:

conda update python

Update Python to later version:

conda install python=3.6

Version of a package:

(tensorflow)Jeffreys-MacBook-Pro:present jeff$ python
Python 3.5.2 |Continuum Analytics, Inc.| (default, Jul  2 2016, 17:52:12)
[GCC 4.2.1 Compatible Apple LLVM 4.2 (clang-425.0.28)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
>>> print(tensorflow.__version__)
0.8.0
>>>

You might want to add a new environment to Jupyter notebook. By default, Jupyter will use
what ever environment you have active when you launch Jupyter. However, if you would like
to allow your new environment to be used in Jupyter execute the following:

For Mac:

source activate tensorflow
pip install ipykernel
python -m ipykernel install --user --name=tensorflow

For Windows:

activate tensorflow
pip install ipykernel
python -m ipykernel install --user --name=tensorflow

Jupyter Notebooks

It can be really handy to use Jupyter notebook to manage your different Python environments. To do this you must install Jupyter in your root Python environment (and each additional environment):

conda install jupyter

You must also install nb_conda in your root:

conda install nb_conda

You can also add support for R in Jupyter:

conda install -c r r-essentials

If you want your Anaconda environment to show up in Jupyter, you will need to add it. To add an environment named TensorFlow, use the following command:

source activate tensorflow
python -m ipykernel install --user --name tensorflow --display-name "Python (tensorflow)"

You might need to remove a conda environment from Jupyter. To list all environments, use the following command:

jupyter kernelspec list

To actually uninstall one, use the following command.

jupyter kernelspec uninstall tensorflow

Import/Export Environments

Once you create an environment (as described above), it can be useful to import and export these environments. You can use this to backup an environment, or transfer it to another machine.

To save an environment (make sure you have the environment you want to save activated), use this:

conda env export > environment.yml

To restore it onto another machine run this command. The name of this environment is specified in the first part of the .yml file. If you already have an environment named this, you will get an error. You can always rename the environment
inside of the .yml file. Or rename it with the -n option.

conda env create -f environment.yml -n tensorflow