Numerix

Numerical library for arrays, transforms, sampling, and linear algebra

What is Numerix?

Numerix is a numeric library for JavaScript and TypeScript that follows the NumPy API. It gives you arrays, arithmetic, statistics, linear algebra, signal transforms, and random sampling under one consistent namespace. Operations work on nested arrays and broadcast operands the way NumPy does, so a scalar combines with a matrix without manual looping.

Installation

The package is not published to any registry yet, so clone it from source.

git clone https://github.com/NeaByteLab/Numerix.git
cd Numerix

Usage

import Numeric from '@neabyte/numerix'

// Creates a 2x2 matrix filled with zeros.
Numeric.create.zero([2, 2])

// Adds two vectors element by element.
Numeric.math.add([1, 2, 3], [10, 20, 30]) // [11, 22, 33]

// Reshapes a flat vector into a matrix.
Numeric.shape.reshape([1, 2, 3, 4], [2, 2])

// Computes the sum of every entry.
Numeric.reduce.sum([1, 2, 3, 4]) // 10

Every method, with examples, is documented in docs/README.md.

API

Every method lives in the namespace that matches its group in the contract.

Namespace	Methods	Covers
create	14	array creation, range, mesh, diagonal, vandermonde
shape	17	reshape, transpose, broadcast, join, split, pad, roll, flip
index	8	where, select, take, put, ravel, unravel, mask, slide
math	34	arithmetic, powers, rounding, gcd, convolution, gradients
trig	16	sin, cos, tan, inverses, hyperbolic, deg, rad, unwrap
complex	4	real, imag, angle, conj
reduce	12	sum, prod, cumulative forms, min, max, argmin, argmax
stat	11	mean, average, median, quantile, variance, std, cov, corr
sort	5	order, argsort, partition, argpartition, search
set	6	unique, union, intersect, diff, xor, member
logic	16	comparisons, boolean operators, finiteness checks
bit	8	bitwise operators, shifts, popcount, pack, unpack
linalg	20	dot, matmul, einsum, norm, det, inv, solve, qr, svd, eig
fft	8	forward, inverse, real transforms, frequency, shift
random	32	28 distributions plus seed, choice, shuffle, permute
poly	19	evaluate, fit, roots, derivatives, bases, integration
window	5	bartlett, blackman, hamming, hanning, kaiser
string	26	concatenation, case, padding, searching, splitting, encoding
dtype	4	cast, check, info, endian
util	4	apply, memory, format, io

Acknowledgements

This project follows the NumPy 2.5.4 API, see NOTICE for the full attribution.

License

This project is distributed under the BSD 3-Clause license. See the LICENSE file for the full text.