The Rust ecosystem is full of a data structure implementations that may be useful in machine learning. You can find multi-dimensional arrays, matrices, graphs, and a large list of tree libraries.
num has emerged as a de facto choice for numerical types, and a few other crates listed here show promise, but most ML frameworks have opted to implement their own data structures, so it’s too early to answer affirmatively to “Are we numpy yet?”
The list below is still lacking, so for more specific data structures don’t forget to search crates.io.
An n-dimensional array for general elements and for numerics. Lightweight array views and slicing; views support chunking and splitting.
Graph data structure library. Provides graph types and graph algorithms.
A bioinformatics library for Rust. This library provides implementations of many algorithms and data structures that are useful for bioinformatics, but also in other fields.
A collection of numeric types and traits for Rust, including bigint, complex, rational, range iterators, generic integers, and more!
A sparse matrix library
dataframe structure and operations
A directed acyclic graph data structure library. It is Implemented on top of petgraph's Graph data structure and attempts to follow similar conventions where suitable.
A fast BVH using SAH
N-dimensional matrix class for Rust
High-level bindings for Faiss, the vector similarity search engine
Rust Data Science
N-dimensional dense arrays.
A roulette wheel selection collection, used in genetic algorithm for fitness proportionate selection.