Introduction to Conda

Estimated time: 25 minutes

Overview

Questions

• Why do I need Conda when I already have modules?
• How do I create an isolated Python environment?
• How do I activate and deactivate environments?

Objectives

• Understand the difference between modules and conda environments
• Create and activate conda environments
• Verify which Python is being used
• Deactivate environments when done

Modules vs. Conda: When to Use Each

---

Modules load pre-installed software managed by system administrators: - Good for: Large applications, compiled binaries, cluster-wide tools - Examples: R, GAMESS, Gaussian, GCC

Conda manages Python/R packages in isolated environments: - Good for: Python libraries, package dependencies, project-specific stacks - Examples: numpy, scipy, matplotlib, scikit-learn, TensorFlow

Most workflows use both: modules for system software, conda for language-specific packages.

Your First Conda Environment

---

Step 1: Load Miniconda

Miniconda is a minimal conda distribution available as a module:

BASH

$ module load miniconda3
$ conda --version
conda 26.3.2

The miniconda3 module comes in three versions on Sagehen HPC; py313_26.3.2-2 is the default (D) — and often already loaded (L).

Step 2: Create an environment

Create a new conda environment for a specific project:

BASH

$ conda create -n myproject python=3.11

This creates an environment called “myproject” with Python 3.11. Conda asks for confirmation:

The following packages will be installed:
  ...

Proceed? [y/N] y

Press y to proceed.

Step 3: Activate the environment

BASH

$ conda activate myproject
(myproject) $

Notice the prompt changes to show (myproject). You’re now inside the environment.

Step 4: Verify Python

BASH

(myproject) $ python --version
Python 3.11.8

(myproject) $ which python
/rhome/<myusername>/.conda/envs/myproject/bin/python

Python is from your environment, not the system Python.

Step 5: Deactivate when done

BASH

(myproject) $ conda deactivate
$

The prompt returns to normal.

Listing Environments

---

See all conda environments on your account:

BASH

$ conda env list
# conda environments:
#
base                  /opt/.../miniconda3/py313_26.3.2-2
myproject             /rhome/<myusername>/.conda/envs/myproject
analysis_2024         /rhome/<myusername>/.conda/envs/analysis_2024

The base environment is miniconda3 itself; other environments are yours.

Callout

WARNING: Never Install into Base

DO NOT run conda install in the base environment without specifying -n envname.

BASH

# BAD - never do this on shared systems
$ conda install tensorflow

Instead, always create a named environment:

BASH

# GOOD
$ conda create -n ml_project tensorflow
$ conda activate ml_project

On shared clusters, installing in base can consume your disk quota and interfere with other users’ work.

Challenge

Challenge 1: Create and Activate an Environment

1. Load miniconda3
2. Create an environment called “data_sci” with Python 3.10
3. Activate it
4. Check which Python is being used
5. Deactivate

BASH

$ module load miniconda3
$ conda create -n data_sci python=3.10
$ conda activate data_sci
(data_sci) $ which python
(data_sci) $ python --version
(data_sci) $ conda deactivate

Show me the solution

After activation:

which python outputs: /rhome/<myusername>/.conda/envs/data_sci/bin/python
python --version outputs: Python 3.10.X

After deactivation, which python points to system Python or shows not found. The key point: conda activation changes your PATH to prioritize the environment’s Python.

Key Points

• Modules load system software; conda manages Python packages in isolated environments
• Create environments with conda create -n name python=version
• Use conda activate to enter an environment; conda deactivate to exit
• Most workflows use both modules and conda together
• Never install into the base environment on shared systems