Info
[Announce] SciRS2 v0.1.2: A Comprehensive, Pure Rust Scientific Computing Ecosystem (SciPy alternative in Rust)
Hi everyone,
I am proud to announce the launch of SciRS2 (v0.1.2), a massive project aiming to provide a complete, production-ready scientific computing stack entirely in Rust.
Over the past few weeks (through v0.1.0 and v0.1.1), we have stabilized the core architecture, and with v0.1.2, we are ready for wider community adoption.
The Goal: "Pure Rust" Science
Scientific computing in Rust has often required linking against C/Fortran libraries (OpenBLAS, LAPACK, FFTW) or bridging to Python. SciRS2 changes this. We provide a 100% Pure Rust implementation for linear algebra, optimization, signal processing, and more — making deployment effortless (just cargo build) and ensuring memory safety down to the numerical core.
Key Features (v0.1.0 - v0.1.2)
SciRS2 is designed as a modular ecosystem. You can use the top-level crate or pick individual components:
scirs2-linalg): Pure Rust BLAS/LAPACK implementation. No system dependencies required.scirs2-optimize): Solvers for unconstrained/constrained problems (Nelder-Mead, BFGS, etc.), comparable to scipy.optimize.scirs2-stats, scirs2-cluster): Distributions, hypothesis testing, and clustering algorithms (K-Means, DBSCAN, Hierarchical).scirs2-signal, scirs2-ndimage): FFT-based filtering, convolution, and multi-dimensional image processing.scirs2-autograd): Production-ready automatic differentiation engine (similar to PyTorch autograd) for building custom ML models.scirs2-io): Seamless reading/writing of CSV, HDF5, and MATLAB .mat files.
Production Ready
rayon for multi-threaded performance.
Links
This is just the beginning of a larger "COOLJAPAN Ecosystem" vision. We invite you to try it out, break it, and help us build the future of scientific Rust!
1 post - 1 participant
🏷️ Rust_feed