Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Conda is a package management system that can be used to install software and create Python (and R) development environments on many different platforms. You can install it on your own device, and it is available at many High Performance Computing centers.

conda (and mamba) are command-line tools run from the terminal (shell).

Mamba

It’s common that installing packages using Conda is slow or fails because Conda is unable to resolve dependencies. To get around this, we suggest the use of Mamba.

Mamba is a drop-in replacement for Conda that is generally faster and better at resolving dependencies.

In the examples below, we show the syntax for Mamba, but you can simply replace mamba with conda if for some reason you prefer that (e.g., if you haven’t installed mamba on your own machine).

Environments

Conda will install all of your packages into your default environment. However there are various situations in which you may want to create a separate environment, These include (1) if you are working on multiple projects simultaneously, (2) if you want to be able to eventually share your environment with a collaborator, and (3) if some package versions conflict with others in your default environment.

Creating

To create a new environment:

mamba create --name myproject2

You can also create new environments with an entirely different version of python:

mamba create --name myproject2 python=3.12

Note that under some situations your environment might make use of the Python executable or some Python packages from outside the environment. We have details and tips on when this happens and how to create a completely isolated environment.

Activating

To enter into (activate) your newly created environment, you can use source activate or mamba activate.

Here we illustrate using source as that is recommended on the SCF as discussed below.

gandalf:~$ source activate myproject2

The shell prompt will be updated to reflect the active environment name.

(myproject2) gandalf:~$

To deactivate the environment:

(myproject2) gandalf:~$ source deactivate
gandalf:~$

In some cases you may default to the base environment so you might see:

(myproject2) gandalf:~$ source deactivate
(base) gandalf:~$

mamba/conda activate versus source activate

You can also use mamba activate and mamba deactivate, but there are some issues to be aware of.

When you first try to use mamba activate, Mamba will prompt you to run mamba init to initialize Mamba for the shell you are using. You can do this, but note that it will modify your .bashrc so that Mamba commands are available whenever you log in. As part of this, Mamba will put you in the base Mamba environment automatically when a new shell starts. On the SCF, this will prevent you from accessing the Python version and related packages that the SCF provides. To avoid this, we recommend running mamba config --set auto_activate_base False if you run mamba init on the SCF.

Using source activate (and source deactivate) instead of conda activate (and conda deactivate) are deprecated but still work, and you may wish to use them in some situations (such as on the SCF) to avoid the behavior discussed above.

Installing packages

In general, you’ll need to install all the packages you need in your environment, though in some cases your environment might make use of packages from outside the environment.

You can search for available packages.

(myproject2) gandalf:~> mamba search jax
Getting repodata from channels...

conda-forge/linux-64                                        Using cache
conda-forge/noarch                                          Using cache
 Name Version Build                     Channel     Subdir
───────────────────────────────────────────────────────────
 jax  0.7.2   pyhd8ed1ab_0              conda-forge noarch
 jax  0.7.1   pyhd8ed1ab_0              conda-forge noarch
 jax  0.7.0   pyhd8ed1ab_0              conda-forge noarch
 jax  0.6.2   pyhd8ed1ab_0              conda-forge noarch
 jax  0.6.0   pyhd8ed1ab_0              conda-forge noarch
<snip>

Here we have searched for jax and mamba found several versions of it.

You can now install packages you need, including specific versions if desired.

If we ask mamba to install version 0.6.2, it will check to make sure that it can proceed, and will tell you what it will do:

(myproject) gandalf:~$ mamba install jax=0.6.2
Pinned packages:

  - python=3.13


Transaction

  Prefix: /scratch/users/paciorek/conda/envs/myproject2

  Updating specs:

   - jax=0.6.2


  Package                  Version  Build                Channel          Size
────────────────────────────────────────────────────────────────────────────────
  Install:
────────────────────────────────────────────────────────────────────────────────

  + c-ares                  1.34.5  hb9d3cd8_0           conda-forge     207kB
  + importlib-metadata       8.7.0  pyhe01879c_1         conda-forge      35kB
  + jax                      0.6.2  pyhd8ed1ab_0         conda-forge       2MB
  + jaxlib                   0.6.2  cpu_py313h9430eff_1  conda-forge      67MB
  + libabseil           20250127.1  cxx17_hbbce691_0     conda-forge       1MB
  + libblas                 3.11.0  1_h4a7cf45_openblas  conda-forge      18kB
  + libcblas                3.11.0  1_h0358290_openblas  conda-forge      18kB
  + libgfortran             15.2.0  h69a702a_7           conda-forge      29kB
  + libgfortran5            15.2.0  hcd61629_7           conda-forge       2MB
  + libgrpc                 1.71.0  h8e591d7_1           conda-forge       8MB
  + liblapack               3.11.0  1_h47877c9_openblas  conda-forge      18kB
  + libopenblas             0.3.30  pthreads_h94d23a6_4  conda-forge       6MB
  + libprotobuf             5.29.3  h7460b1f_2           conda-forge       3MB
  + libre2-11           2025.06.26  hba17884_0           conda-forge     212kB
  + ml_dtypes                0.5.1  py313h08cd8bf_1      conda-forge     294kB
  + numpy                    2.3.5  py313hf6604e3_0      conda-forge       9MB
  + opt_einsum               3.4.0  pyhd8ed1ab_1         conda-forge      62kB
  + re2                 2025.06.26  h9925aae_0           conda-forge      27kB
  + scipy                   1.16.3  py313h11c21cd_1      conda-forge      17MB
  + zipp                    3.23.0  pyhd8ed1ab_0         conda-forge      23kB

  Summary:

  Install: 20 packages

  Total download: 115MB

────────────────────────────────────────────────────────────────────────────────


Confirm changes: [Y/n]

Typing Y will let mamba proceed.

You could also avoid the interactivity and let mamba proceed without asking by including -y in your mamba install invocation.

You can list installed packages like this:

(myproject2) gandalf:~> mamba list jax
List of packages in environment: "/scratch/users/paciorek/conda/envs/myproject2"

  Name    Version  Build                Channel    
─────────────────────────────────────────────────────
  jax     0.6.2    pyhd8ed1ab_0         conda-forge
  jaxlib  0.6.2    cpu_py313h9430eff_1  conda-forge

Environment locations

If you have a scratch directory on the SCF, your Conda environments by default will be stored in /scratch/users/<username>/conda/envs. If you don’t have a scratch directory, they will be in the usual ~/.conda/envs directory. The purpose of this is to reduce backup sizes, as scratch it not backed up and Conda environments can be reproduced based on saving an environment recipe using mamba env export.

Source packages that Conda caches for later use will be stored in /tmp/pkgs on each machine.

Reproducing an environment

You can create a snapshot of your current environment that can then be reproduced by collaborators, or even just yourself if you want to setup the same environment on another device. To export an environment:

source activate myproject2
mamba env export --from-history > environment.yml

Transmit the environment.yml file to the other person or device. One can then create a new environment from that file:

mamba env create -f environment.yml

Note that installing on a different operating system can easily fail because the requirements file includes build-specific versioning that only works on the specific operating system, so using on a different OS can require one to pass additional flags, such as the --from-history flag above.

Isolating your environment from the SCF Python installation

When creating a Conda environment, if you do not specify the version of Python, your environment will use our default Python version via the SCF default Python executable, and with all the SCF-installed Python packages available to you. That has the advantage that you may not need to install a bunch of packages.

gandalf:~$ mamba create -y -n myenv
gandalf:~$source activate myenv
(myenv) gandalf:~$ type python
python is /usr/local/linux/miniforge-3.13/bin/python
(myenv) gandalf:~$ python -c "import numpy; print(numpy.__version__)"
2.1.3

The downside is that your environment is tied to our defaults and is harder to reproduce. Also, in many cases, when you install a package in the environment, updated versions of dependencies (potentially including Python itself) may be installed.

To isolate your environment, be sure to specify the Python version you want, even if it is the same as the SCF’s default Python version. Of course as a result you’ll need to install all the packages you need.

gandalf:~$ mamba create -y -n myenv python=3.13
gandalf:~$ source activate myenv
gandalf:~$ type python
python is /accounts/vis/paciorek/.conda/envs/myenv/bin/python
gandalf:~$ python -c "import numpy; print(numpy.__version__)"
ModuleNotFoundError: No module named 'numpy'

One additional complication is that if you have packages installed via pip (which will be located in ~/.local in your home directory), your environment will try to use those packages, which means your environment is not fully isolated and can be hard to reproduce. In addition, if you choose to install packages for your environment via pip, you’ll generally want to install without using the --user flag. Omitting that flag will cause packages to be installed within the environment’s directory, while including the flag will cause them to be installed in ~/.local (and risk affecting your use of packages outside the environment!).

Jupyter kernels

You can create a Jupyter kernel associated with your environment for use with Jupyter notebooks.

Using Mamba/Conda as a general installer

Conda is a general package manager. Many users use it just to install Python packages, but it can be used to install software more generally, including packages that would otherwise require installation from source (e.g., using make or cmake) and potentially have complicated additional dependencies. A good option for installing a piece of software is to check if there is a Conda package for it, before you try to install from source code. Executables installed when you install a Conda package will be placed in the bin subdirectory of the active Conda environment.

R and Conda

Conda can actually be used to install R and R packages inside a Conda environment. You’re welcome to do this, but most SCF users make use of the R installation (and a wide range of R packages) that we provide at the system level and install additional packages for use with the system R.

Installing Mamba/Conda on your own computer

Using Miniforge (recommended)

For your own computer, we recommend using the Miniforge distribution as your Python installation rather than Anaconda. Miniforge by default uses the conda-forge channel, a popular channel providing a wide variety of up-to-date packages. You can optionally have it use Mamba rather than Conda.

Conda/Mamba for Mac ARM64-based machines (Apple silicon M-series chips)

With the M-series Macs, Apple is now using its own chips, referred to as Apple silicon. These have a different architecture, ARM64, than standard x86-64 chips, such as those produced by Intel and AMD.

To take full advantage of the new chips, you can use Miniforge, as it provides versions for ARM64.  You can also use Anaconda/Miniconda with ARM64 support, installing with the 64-Bit (Apple silicon) installer

Non-ARM64 programs, including non-ARM64 Miniforge installations, will work on the new Apple Silicon-based machines by automatically making use of Apple’s Rosetta2 system to translate machine code from ARM64 to x86_64, but you can expect some decrease in performance.

Using Anaconda/Miniconda

While we recommend installing Miniforge, you can also install Anaconda or a minimal installation called Miniconda.

To install Miniconda:

  1. Download Miniconda. There are installers for Linux, macOS, and Windows. Miniconda is a much smaller version of Anaconda that includes more packages out of the box as well as a graphical user interface to the installer. Miniconda is better if you don’t want to install too many unnecessary packages right at the start.

  2. Run the installer. This will prompt you a couple of times, and it is usually sensible to accept the defaults. Conda alters your shell initialization, and updates your executable path to find the base installation. For example, Miniconda will be installed into ~/miniconda3/ by default, so ~/miniconda3/bin/ will be added to your PATH variable. Conda may update your shell prompt as well with status information.

  3. Open a new terminal window. This will start up a new shell that will be aware of the new Conda tools. For example, on a Mac:

    (base) mylaptop:~$ type -a python
python is /Users/me/miniconda3/bin/python
python is /usr/bin/python
Computing in Statistics at Berkeley
Python
Computing in Statistics at Berkeley
R and RStudio