Description
Doing matrix math on Python lists means nested loops that are slow to run, hard to read, and fall over once the data gets big.
NumPy adds a fixed-size, single-type ndarray multidimensional array to Python, backed by compiled code, so an operation over hundreds of thousands of values fits on one line. Linear algebra, Fourier transforms, statistics and random simulation come as ready-made functions, and much of the Python data-science stack is built on top of it. It's open source and fine for commercial use.
ndarray: same-type data stored contiguously and operated on as a block, no Python loops required.
Linear algebra: matrix products, eigenvalues and eigenvectors, and linear system solvers out of the box.
Fourier transforms: built-in FFT routines for signal analysis.
Random numbers and statistics: a modern generator that draws from normal and other distributions, plus the usual summary statistics.
Slicing and broadcasting: slice by rows and columns, assign in bulk, and combine arrays of different shapes without reshaping by hand.
Multiple Python versions: prebuilt Windows wheels for 32-bit and 64-bit, installable with a single pip command.
NumPy adds a fixed-size, single-type ndarray multidimensional array to Python, backed by compiled code, so an operation over hundreds of thousands of values fits on one line. Linear algebra, Fourier transforms, statistics and random simulation come as ready-made functions, and much of the Python data-science stack is built on top of it. It's open source and fine for commercial use.
Features
ndarray: same-type data stored contiguously and operated on as a block, no Python loops required.
Linear algebra: matrix products, eigenvalues and eigenvectors, and linear system solvers out of the box.
Fourier transforms: built-in FFT routines for signal analysis.
Random numbers and statistics: a modern generator that draws from normal and other distributions, plus the usual summary statistics.
Slicing and broadcasting: slice by rows and columns, assign in bulk, and combine arrays of different shapes without reshaping by hand.
Multiple Python versions: prebuilt Windows wheels for 32-bit and 64-bit, installable with a single pip command.
