How to install SciPy in a Python virtual environment with pip

Scientific Python projects often need SciPy without changing the operating system's base Python packages. A project-local venv gives pip a private site-packages directory for SciPy, NumPy, and any later analysis dependencies.

The official SciPy install page lists pip as an environment-based workflow after creating and activating a venv environment. Running pip through the active python interpreter keeps the install tied to that environment instead of a different pip executable on PATH.

A POSIX shell with python3 already installed is assumed here, and .venv is created in the current project directory. On Windows, use py -m venv .venv and the activation script under .venv\Scripts, and install missing venv support through the platform package manager before using pip.

1. Change to the project directory that should own the environment.

   $ cd ~/scipy-demo

   Keeping .venv beside the project files makes the interpreter and installed packages easy to pair with that project.

2. Create the virtual environment.

   $ python3 -m venv .venv

   On success, venv normally prints no output and bootstraps pip into the environment unless --without-pip is used.

   If Debian or Ubuntu reports that ensurepip is unavailable, install the matching python3-venv package and rerun the command.

   Related: How to create a virtual environment in Python

3. Activate the virtual environment in the current shell.

   $ source .venv/bin/activate

   Use .\.venv\Scripts\Activate.ps1 in Windows PowerShell or .venv\Scripts\activate.bat in Command Prompt.

   Related: How to activate a Python virtual environment

4. Confirm that pip belongs to the virtual environment.

   $ python -m pip --version
   pip 26.1.2 from /home/user/scipy-demo/.venv/lib/python3.14/site-packages/pip (python 3.14)

   The path after from should point inside the project .venv directory before installing SciPy.

5. Install SciPy into the active environment.

   $ python -m pip install scipy
   Collecting scipy
   Collecting numpy<2.8,>=2.0.0 (from scipy)
   ##### snipped #####
   Installing collected packages: numpy, scipy
   
   Successfully installed numpy-2.5.0 scipy-1.18.0

   pip installs a compatible SciPy wheel and the required NumPy dependency when wheels are available for the active Python and platform. Exact versions and wheel tags vary over time.

   Related: How to install third-party packages in Python using pip

6. Show the installed SciPy metadata.

   $ python -m pip show scipy
   Name: scipy
   Version: 1.18.0
   Summary: Fundamental algorithms for scientific computing in Python
   ##### snipped #####
   Location: /home/user/scipy-demo/.venv/lib/python3.14/site-packages
   Requires: numpy

   The Location field should point inside the active virtual environment.

7. Run a SciPy import and function smoke test.

   $ python -c "import scipy; from scipy import special; print(scipy.__version__); print(special.expit(0))"
   1.18.0
   0.5

   Look for the installed SciPy version followed by 0.5 from scipy.special.expit(0).

8. Check dependency metadata after the install.

   $ python -m pip check
   No broken requirements found.

   If pip check reports conflicts, reinstall the project's tested dependency set instead of keeping a partially broken environment.