What is a virtual environment?¶

Beginner

Prerequisites: What is a Python package

Why you need one¶

Suppose you have one Python on your machine and every pip install lands in it. Eventually:

Project A needs pandas==1.5.
Project B needs pandas==2.2 (the API changed; 1.5 code won't run).
A system utility script needs pandas==1.3.

One Python cannot hold all three at once — this is dependency conflict. The fix: give each project its own isolated "virtual environment", each with its own package list.

Analogy: a per-project MATLAB working directory + userpath that pins different toolbox versions.

Two mainstream options¶

Option A: conda (recommended for power-systems engineers)¶

Anaconda / Miniconda is the most popular distribution in scientific computing. Bundles Python + virtualenv management + a compiler toolchain. One install is enough.

We recommend Miniconda (lightweight, ~100 MB):

# Create a fresh environment named pylectra-env with Python 3.11
conda create -n pylectra-env python=3.11

# Activate it
conda activate pylectra-env

# Install pylectra inside the environment (this pip only touches it)
pip install pylectra

# Leave when done
conda deactivate

After activation the shell prompt becomes (pylectra-env) C:\Users\...>, marking you as inside an isolated environment — every pip / python invocation now affects only this one.

Option B: venv (built into Python, no extra install)¶

# Create an environment called .venv in the current directory
python -m venv .venv

# Activate
.venv\Scripts\activate           # Windows
source .venv/bin/activate        # macOS / Linux

# Install packages
pip install pylectra

# Exit
deactivate

venv produces a plain folder (the .venv name is convention; the leading dot makes it hidden). To delete the environment, just delete the folder.

Which one when?¶

Situation	Recommendation
Multiple Python versions (3.10 / 3.11 / 3.12)	conda
Installing `pandapower`, `scipy`, anything with C/Fortran extensions but no compiler on hand	conda
Corporate machine without admin rights	conda (Miniconda user install)
Python already installed; just want lightweight isolation	venv

The rest of this documentation uses conda in examples because it's the most common in power-systems circles.

A common mistake¶

Running pip install without activating an environment — the package lands in the "system Python" or the previously-activated env, and the next time you activate the right one import raises ModuleNotFoundError.

How to verify:

# Where is the current python?
where python                     # Windows
which python                     # macOS / Linux

# Where will pip install to?
pip -V

If the output contains your env's name (e.g. pylectra-env), you're correct.

FAQ¶

Q: Will environments fill up my disk?¶

Each is a few hundred MB to 1 GB (with numpy/scipy/matplotlib). One or two are fine; for dozens, share a "general science env" and only spin up dedicated ones when conflicts force it.

Q: My environment broke. How do I fix it?¶

Don't fix it — conda env remove -n pylectra-env (or rm -rf .venv) and recreate. Rebuilding is faster than debugging.

Q: Mixing conda and pip — safe?¶

Generally yes, but a rule of thumb: install binary-heavy science packages with conda (numpy, scipy, pandapower are all on conda), install pure-Python packages with pip (pylectra). Run conda install ... first, then pip install ....

Q: Can I use Jupyter Notebook with this?¶

Yes. After conda activate pylectra-env, run pip install jupyterlab, then jupyter lab. Notebooks pick up whichever Python is currently active.

Next steps¶

Install Pylectra — apply what you just learned and install pylectra inside an isolated environment.