Setting Up a Scientific Python Environment on macOS (Apple Silicon)

This guide walks through creating a dedicated scientific Python environment called numerics, installing common scientific packages, and making the environment available inside Jupyter notebooks.

The instructions assume:

• macOS on Apple Silicon (M1/M2/M3)
• Python installed via Homebrew
• Terminal application available

1. Check Your Python Installation

Before creating an environment, verify that Python is installed and available from Homebrew.

python3 --version
which python3

Expected output:

Python 3.14.x
/opt/homebrew/bin/python3

If Python is not installed through Homebrew, install it first:

brew install python

2. Create a Dedicated Virtual Environment

Using a virtual environment keeps scientific packages isolated from the system Python installation.

Create a directory to store virtual environments:

mkdir -p ~/venvs

Create the environment:

python3 -m venv ~/venvs/numerics

Activate it:

source ~/venvs/numerics/bin/activate

Your shell prompt should now begin with:

(numerics)

Confirm that Python is coming from the virtual environment:

which python

Expected output:

/Users/subhodeep/venvs/numerics/bin/python

3. Upgrade Packaging Tools

Upgrade the package-management tools before installing scientific software:

pip install --upgrade pip setuptools wheel

Verify:

pip --version

4. Create a Requirements File

Create a file named:

~/numerics-requirements.txt

Populate it with the packages you want installed.

Example:

numpy
scipy
pandas
matplotlib
seaborn
numba
ipykernel
jupyterlab

5. Install the Scientific Stack

Install all packages listed in the requirements file:

pip install -r ~/numerics-requirements.txt

Depending on the number of packages and internet speed, installation may take several minutes.

Occasionally you may see messages such as:

WARNING: Cache entry deserialization failed, entry ignored

These warnings are generally harmless and can be ignored.

6. Register the Environment with Jupyter

Register the virtual environment as a Jupyter kernel so it appears as a selectable notebook kernel.

python -m ipykernel install \
    --user \
    --name numerics \
    --display-name "Python (numerics)"

After completion, Jupyter notebooks will offer:

Python (numerics)

as a kernel option.

7. Test the Installation

Launch an interactive Python session:

python

Run:

import numpy
import scipy
import pandas
import matplotlib
import seaborn
import numba

print("OK")

Expected output:

OK

Exit Python:

quit()

8. Verify Jupyter Kernel Registration

List all installed Jupyter kernels:

jupyter kernelspec list

Example output:

Available kernels:
  numerics
  julia-1.12
  wolframlanguage13.3

The exact list will depend on your machine.

9. Daily Usage

Activate the environment whenever you want to work in it:

source ~/venvs/numerics/bin/activate

Deactivate when finished:

deactivate

Install additional packages later:

pip install PACKAGE_NAME

Upgrade a specific package:

pip install --upgrade PACKAGE_NAME

Check for outdated packages:

pip list --outdated

10. Create a Reproducible Snapshot

Save the exact package versions currently installed:

pip freeze > ~/numerics-lock.txt

The generated file can later be used to recreate the environment exactly:

pip install -r ~/numerics-lock.txt

This is useful for reproducible research, sharing environments with collaborators, and recovering a working setup after system upgrades.

11. Create a Quick-Access Alias

To jump into your scientific environment from any terminal directory, add an alias to your shell configuration file.

Open your .zshrc file:

nano ~/.zshrc

Add this line to the bottom:

# Scientific Python Environment
alias numerics='source ~/venvs/numerics/bin/activate'

Save, exit, and reload:

source ~/.zshrc

Now, simply type numerics in any terminal to jump into your workspace.

Useful Commands

Show Installed Packages

pip list

Display Environment Location

which python

Display Package Details

pip show numpy

Remove the Environment Completely

rm -rf ~/venvs/numerics

Remove the Jupyter Kernel

jupyter kernelspec uninstall numerics

Requirements File (numerics-requirements.txt)

Create this file in your home directory (~/) , this file defines the full scientific stack that will be installed:

# Core & High-Performance
numpy
scipy
numba
numexpr
joblib
psutil

# Data Analysis & Storage
pandas
xarray
dask
h5py
tables
zarr

# Visualization
matplotlib
seaborn
plotly

# Math & Symbolic
sympy
uncertainties
mpmath

# Statistics & Inference
statsmodels
scikit-learn
emcee
corner
lmfit

# Domain Specific
astropy

# Interoperability
juliacall
wolframclient

# Terminal & Formatting
tabulate
prettytable
rich
tqdm

# Jupyter
jupyterlab
ipykernel