Installing and importing NumPy

Install

NumPy installs like any Python package, inside a virtual environment, as the Python course showed. Choose the tool you use:

uv init meu-projeto
cd meu-projeto
uv add numpy

NumPy distributes pre-compiled versions (wheels) for the most common systems and Python versions. Therefore, the installation takes seconds and does not require a compiler.

Import, and the convention `np`

import sys
from pathlib import Path

import numpy as np

print("versão 2 ou superior:", int(np.__version__.split(".")[0]) >= 2)
print("onde está o NumPy:", Path(np.__file__).parent.name)
print("este Python existe no disco:", Path(sys.executable).exists())

versão 2 ou superior: True
onde está o NumPy: numpy
este Python existe no disco: True

The alias np is not a language rule, but it is such a strong convention that all documentation, every tutorial, and every answer on the internet assumes it. Using another alias just makes you translate what you read.

To see how your NumPy was built (which linear algebra libraries it uses, for example), there is the show_config:

np.show_config()

When the import fails

The most common error is installing with the pip in a terminal and receiving No module named 'numpy' in the editor. Almost always this means that there are two different Pythons: the one in the terminal received the package, and the one in the editor (or the notebook) did not. To find out which Python is running and if it sees NumPy:

python -c "import sys; print(sys.executable)"
python -m pip show numpy
python -c "import numpy; print(numpy.__version__, numpy.__file__)"

The rule that saves me time

Always use python -m pip instead of pip pure. This way, the pip is necessarily the one from the interpreter you called, and the package lands in the right place.