A comprehensive scientific computing and AI/ML library in pure Rust, offering SciPy-compatible APIs with 10-100x performance gains.
SciRS2 is a production-ready scientific computing and AI infrastructure library written entirely in Rust. It provides a comprehensive ecosystem for linear algebra, statistics, machine learning, signal processing, and more, while maintaining API compatibility with SciPy. The library solves the problem of performance and dependency management in scientific computing by leveraging Rust's performance, safety, and concurrency features to deliver significant speed improvements over traditional Python-based solutions without requiring external C/Fortran libraries.
Rust developers and researchers who need high-performance, memory-safe scientific computing and AI/ML capabilities, particularly those migrating from Python's SciPy/NumPy stack or building performance-critical applications in domains like data science, engineering, and machine learning.
Developers choose SciRS2 for its 100% Pure Rust implementation by default, which eliminates complex C/Fortran system dependencies and ensures easy installation and cross-platform compatibility. Its unique selling point is delivering 10-100x performance gains through ultra-optimized SIMD operations and multi-backend GPU acceleration while maintaining SciPy API compatibility for a smooth migration path.
SciRS2 - Scientific Computing and AI in Rust., providing SciPy-compatible APIs while leveraging Rust's performance, safety, and concurrency features. Unlike traditional scientific libraries
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Eliminates C/Fortran dependencies with OxiBLAS and OxiFFT, ensuring easy installation via `cargo add` and cross-platform compatibility as highlighted in the README's 'Pure Rust by Default' section.
Achieves 10-100x speed improvements through ultra-optimized SIMD operations and multi-backend GPU acceleration, backed by benchmarks showing 14.17x speedups for element-wise operations.
Spans 32 workspace crates covering linear algebra, statistics, neural networks, and more, with 25,863 tests ensuring reliability across scientific computing and AI domains.
Leverages Rust's ownership system to prevent memory leaks and data races, with zero `unwrap()` in production code for robust error handling as stated in the key features.
Maintains API compatibility with SciPy, providing a smooth migration path for Python users, which is a core part of the project's philosophy and goals.
As a v0.4.1 release, it may undergo breaking changes, with the README acknowledging ongoing development and 19 stubs remaining, making it less stable for long-term production use.
Python bindings via PyO3 are feature-gated and not enabled out-of-the-box, requiring additional setup and configuration, which complicates interoperability for mixed-language projects.
The README notes partial support on Windows with some test failures, indicating potential platform-specific bugs that could affect deployment in Windows-centric environments.
With 2.91M lines of Rust code across 32 crates, compile times can be significant, potentially slowing development iteration and increasing resource requirements.
scirs is an open-source alternative to the following products: