zcore

A tensor library written in Zig, for Zig.

zcore aims to be Zig's answer to NumPy/PyTorch — a fast, well-documented tensor library that's as educational as it is practical. Every line is written with beginners in mind. It is the first in a planned line of data-science libraries for Zig: a dataframe (pandas equivalent) and a plotting library.

const zcore = @import("zcore");

const allocator = std.heap.page_allocator;
var t = try zcore.Tensor(f32).zeroes(allocator, &[_]usize{ 2, 3 });
defer t.destroy();

t.fill(1.0);
t.get(&[_]usize{ 0, 1 }).* = 42.0;

Installation

Add zcore as a dependency in build.zig.zon:

zig fetch --save https://github.com/ka1rav6/zcore/archive/refs/tags/v0.1.0.tar.gz

Then import the module in your build.zig:

const zcore = b.dependency("zcore", .{
    .target = target,
    .optimize = optimize,
});
exe.root_module.addImport("zcore", zcore.module("zcore"));

Quick start

git clone https://github.com/ka1rav6/zcore
cd zcore
zig build test        # run the test suite
zig build example1    # run the first example

Requires Zig 0.17.0-dev.1158+1d1193aa7 (see CI for exact pinned version).

Why contribute?

• Low barrier to entry — the codebase is small, modular, and heavily commented. You don't need to be a linear algebra expert to make a meaningful PR.
• Clear roadmap — there's a detailed implementation plan with bite-sized tasks: reshape, transpose, broadcasting, element-wise ops, and more.
• Learn by doing — contribute to a real library while deepening your understanding of Zig, memory management, SIMD, and numerical computing.
• Zero deps — no build system headaches, no bloated dependencies. Just zig build test.
• Your kind of people — no AI-generated code. Just humans writing thoughtful, commented Zig.

Current state (v0.1.0)

Feature	Status
Generic Tensor(T)	Done
Shape / strides	Done
Row-major storage	Done
init, fill, zeroes	Done
Multi-dimensional indexing	Done
Memory ownership (views)	Done
Transpose, slice, resize	Done
Broadcasting	Done
0-length tensor support	Done
Element-wise arithmetic	Planned
Matrix arithmetic	Planned
Full NumPy-level API	Planned

See the implementation plan for the full roadmap.

Documentation

• Implementation Plan — phased roadmap from v0.1.0 through v1.0
• Style Guide — code conventions
• Contributing Guide — how to get involved
• Changelog — release history

How to run

Command	Description
zig build test	Run all tests
zig build run	Run the executable
zig build example1	Run the first example

License

MIT — see LICENSE.