NumPy
The fundamental package for scientific computing with Python
[Latest release: NumPy 2.2. View all releases](/news/#releases)
[NumPy 2.2.0 released!](/news)
[2024-12-08](/news)
Powerful N-dimensional arrays
Fast and versatile, the NumPy vectorization, indexing, and broadcasting concepts are the de-facto standards of array computing today.
Numerical computing tools
NumPy offers comprehensive mathematical functions, random number generators, linear algebra routines, Fourier transforms, and more.
Open source
Distributed under a liberal [BSD license](https://github.com/numpy/numpy/blob/main/LICENSE.txt), NumPy is developed and maintained [publicly on GitHub](https://github.com/numpy/numpy) by a vibrant, responsive, and diverse [community](/community/).
Interoperable
NumPy supports a wide range of hardware and computing platforms, and plays well with distributed, GPU, and sparse array libraries.
Performant
The core of NumPy is well-optimized C code. Enjoy the flexibility of Python with the speed of compiled code.
Easy to use
NumPy’s high level syntax makes it accessible and productive for programmers from any background or experience level.
Try NumPy
Use the interactive shell to try NumPy in the browser
`""" To try the examples in the browser: 1. Type code in the input cell and press Shift + Enter to execute 2. Or copy paste the code, and click on the "Run" button in the toolbar """ # The standard way to import NumPy: import numpy as np # Create a 2-D array, set every second element in # some rows and find max per row: x = np.arange(15, dtype=np.int64).reshape(3, 5) x[1:, ::2] = -99 x # array([[ 0, 1, 2, 3, 4], # [-99, 6, -99, 8, -99], # [-99, 11, -99, 13, -99]]) x.max(axis=1) # array([ 4, 8, 13]) # Generate normally distributed random numbers: rng = np.random.default_rng() samples = rng.normal(size=2500) samples`
ECOSYSTEM
=========
Scientific Domains Array Libraries Data Science Machine Learning Visualization
Nearly every scientist working in Python draws on the power of NumPy.
NumPy brings the computational power of languages like C and Fortran to Python, a language much easier to learn and use. With this power comes simplicity: a solution in NumPy is often clear and elegant.
* * [QuTiP](https://qutip.org)
* [PyQuil](https://pyquil-docs.rigetti.com/en/stable)
* [Qiskit](https://qiskit.org)
* [PennyLane](https://pennylane.ai)
* * [Pandas](https://pandas.pydata.org/)
* [statsmodels](https://www.statsmodels.org/)
* [Xarray](https://xarray.pydata.org/en/stable/)
* [Seaborn](https://seaborn.pydata.org/)
* * [SciPy](https://www.scipy.org/)
* [PyWavelets](https://pywavelets.readthedocs.io/)
* [python-control](https://python-control.org/)
* [HyperSpy](https://hyperspy.org/)
* * [Scikit-image](https://scikit-image.org/)
* [OpenCV](https://opencv.org/)
* [Mahotas](https://mahotas.rtfd.io/)
* * [NetworkX](https://networkx.org/)
* [graph-tool](https://graph-tool.skewed.de/)
* [igraph](https://igraph.org/python/)
* [PyGSP](https://pygsp.rtfd.io/)
* * [AstroPy](https://www.astropy.org/)
* [SunPy](https://sunpy.org/)
* [SpacePy](https://spacepy.github.io/)
* * [PsychoPy](https://www.psychopy.org/)
* * [BioPython](https://biopython.org/)
* [Scikit-Bio](http://scikit-bio.org/)
* [PyEnsembl](https://github.com/openvax/pyensembl)
* [ETE](http://etetoolkit.org/)
* * [PyStan](https://pystan.readthedocs.io/en/latest/)
* [PyMC3](https://docs.pymc.io/)
* [ArviZ](https://arviz-devs.github.io/arviz/)
* [emcee](https://emcee.readthedocs.io/)
* * [SciPy](https://www.scipy.org/)
* [SymPy](https://www.sympy.org/)
* [cvxpy](https://www.cvxpy.org/)
* [FEniCS](https://fenicsproject.org/)
* * [Cantera](https://cantera.org/)
* [MDAnalysis](https://www.mdanalysis.org/)
* [RDKit](https://github.com/rdkit/rdkit)
* [PyBaMM](https://www.pybamm.org/)
* * [Pangeo](https://pangeo.io/)
* [Simpeg](https://simpeg.xyz/)
* [ObsPy](https://github.com/obspy/obspy/wiki)
* [Fatiando a Terra](https://www.fatiando.org/)
* * [Shapely](https://shapely.readthedocs.io/)
* [GeoPandas](https://geopandas.org/)
* [Folium](https://python-visualization.github.io/folium)
* * [COMPAS](https://compas.dev/)
* [City Energy Analyst](https://cityenergyanalyst.com/)
* [Sverchok](https://nortikin.github.io/sverchok/)
NumPy's API is the starting point when libraries are written to exploit innovative hardware, create specialized array types, or add capabilities beyond what NumPy provides.
Array Library
Capabilities & Application areas
[Dask](https://dask.org/)
Distributed arrays and advanced parallelism for analytics, enabling performance at scale.
[CuPy](https://cupy.dev)
NumPy-compatible array library for GPU-accelerated computing with Python.
[JAX](https://jax.readthedocs.io/)
Composable transformations of NumPy programs: differentiate, vectorize, just-in-time compilation to GPU/TPU.
[Xarray](https://xarray.pydata.org/en/stable/index.html)
Labeled, indexed multi-dimensional arrays for advanced analytics and visualization.
[Sparse](https://sparse.pydata.org/en/latest/)
NumPy-compatible sparse array library that integrates with Dask and SciPy's sparse linear algebra.
[PyTorch](https://pytorch.org/)
Deep learning framework that accelerates the path from research prototyping to production deployment.
[TensorFlow](https://www.tensorflow.org)
An end-to-end platform for machine learning to easily build and deploy ML powered applications.
[Arrow](https://arrow.apache.org/)
A cross-language development platform for columnar in-memory data and analytics.
[xtensor](https://github.com/xtensor-stack/xtensor-python)
Multi-dimensional arrays with broadcasting and lazy computing for numerical analysis.
[Awkward Array](https://awkward-array.org/)
Manipulate JSON-like data with NumPy-like idioms.
[uarray](https://uarray.org/en/latest/)
Python backend system that decouples API from implementation; unumpy provides a NumPy API.
[tensorly](http://tensorly.org/stable/home.html)
Tensor learning, algebra and backends to seamlessly use NumPy, PyTorch, TensorFlow or CuPy.
[](/images/content_images/ds-landscape.png)
NumPy lies at the core of a rich ecosystem of data science libraries. A typical exploratory data science workflow might look like:
* **Extract, Transform, Load:** [Pandas](https://pandas.pydata.org), [Intake](https://intake.readthedocs.io), [PyJanitor](https://pyjanitor-devs.github.io/pyjanitor/)
* **Exploratory analysis:** [Jupyter](https://jupyter.org), [Seaborn](https://seaborn.pydata.org), [Matplotlib](https://matplotlib.org), [Altair](https://altair-viz.github.io)
* **Model and evaluate:** [scikit-learn](https://scikit-learn.org), [statsmodels](https://www.statsmodels.org/stable/index.html), [PyMC3](https://docs.pymc.io), [spaCy](https://spacy.io)
* **Report in a dashboard:** [Dash](https://plotly.com/dash), [Panel](https://panel.holoviz.org), [Voila](https://voila.readthedocs.io/)
For high data volumes, [Dask](https://dask.org) and [Ray](https://ray.io/) are designed to scale. Stable deployments rely on data versioning ([DVC](https://dvc.org)), experiment tracking ([MLFlow](https://mlflow.org)), and workflow automation ([Airflow](https://airflow.apache.org), [Dagster](https://dagster.io) and [Prefect](https://www.prefect.io)).
[](https://ai.googleblog.com/2016/12/open-sourcing-embedding-projector-tool.html)
_[Source: Google AI Blog](https://ai.googleblog.com/2016/12/open-sourcing-embedding-projector-tool.html)_
NumPy forms the basis of powerful machine learning libraries like [scikit-learn](https://scikit-learn.org) and [SciPy](https://www.scipy.org). As machine learning grows, so does the list of libraries built on NumPy. [TensorFlow’s](https://www.tensorflow.org) deep learning capabilities have broad applications — among them speech and image recognition, text-based applications, time-series analysis, and video detection. [PyTorch](https://pytorch.org), another deep learning library, is popular among researchers in computer vision and natural language processing.
Statistical techniques called [ensemble](https://towardsdatascience.com/ensemble-methods-bagging-boosting-and-stacking-c9214a10a205) methods such as binning, bagging, stacking, and boosting are among the ML algorithms implemented by tools such as [XGBoost](https://xgboost.readthedocs.io/), [LightGBM](https://lightgbm.readthedocs.io/en/latest/), and [CatBoost](https://catboost.ai) — one of the fastest inference engines. [Yellowbrick](https://www.scikit-yb.org/en/latest/) and [Eli5](https://eli5.readthedocs.io/en/latest/) offer machine learning visualizations.
[](https://www.fusioncharts.com/blog/best-python-data-visualization-libraries)
[](https://github.com/yhat/ggpy)
[](https://www.journaldev.com/19692/python-plotly-tutorial)
[](https://altair-viz.github.io/gallery/streamgraph.html)
[](https://seaborn.pydata.org)
[](https://docs.pyvista.org/)
[](https://napari.org)
[](https://vispy.org/gallery/index.html)
NumPy is an essential component in the burgeoning [Python visualization landscape](https://pyviz.org/overviews/index.html), which includes [Matplotlib](https://matplotlib.org), [Seaborn](https://seaborn.pydata.org), [Plotly](https://plot.ly), [Altair](https://altair-viz.github.io), [Bokeh](https://docs.bokeh.org/en/latest/), [Holoviz](https://holoviz.org), [Vispy](http://vispy.org), [Napari](https://napari.org/), and [PyVista](https://docs.pyvista.org/), to name a few.
NumPy’s accelerated processing of large arrays allows researchers to visualize datasets far larger than native Python could handle.
CASE STUDIES
============
[
First Image of a Black Hole
How NumPy, together with libraries like SciPy and Matplotlib that depend on NumPy, enabled the Event Horizon Telescope to produce the first ever image of a black hole
](/case-studies/blackhole-image)[
Detection of Gravitational Waves
In 1916, Albert Einstein predicted gravitational waves; 100 years later their existence was confirmed by LIGO scientists using NumPy.
](/case-studies/gw-discov)[
Sports Analytics
Cricket Analytics is changing the game by improving player and team performance through statistical modelling and predictive analytics. NumPy enables many of these analyses.
](/case-studies/cricket-analytics)[
Pose Estimation using deep learning
DeepLabCut uses NumPy for accelerating scientific studies that involve observing animal behavior for better understanding of motor control, across species and timescales.
](/case-studies/deeplabcut-dnn)
[](#)