Skip to content

Releases: rust-dd/stochastic-rs

v1.1.1

15 Feb 08:35
9a2a1a1

Choose a tag to compare

v1.1.0

14 Feb 22:10
cfeea0a

Choose a tag to compare

  • Internal SIMD PRNG (xoshiro256++/xoshiro128++) for all float distributions
  • Native i32x8 ziggurat for Normal and Exp — up to 5.2x faster than rand_distr
  • ~30% improvement over v1.0.0 in Normal/Exp; composed distributions (LogNormal, Gamma, Beta, etc.) also benefit
  • FGN/FBM: zero-copy Complex construction, in-place eigenvalue multiply

Full Changelog: v1.0.0...v1.1.0

v1.0.0

13 Feb 20:56
4278fc7

Choose a tag to compare

stochastic-rs v1.0.0

Added

  • Full bindings via PyO3/maturin — 85 model classes with numpy integration, callable parameters (lambdas) for interest rate models, and custom jump distributions
  • SIMD ziggurat normal and exponential distributions for faster sampling
  • Benchmarks and visualization tooling
  • sample_with_rng for reproducible sampling

Changed

  • All distributions are now generic over T (f32/f64) instead of feature-flag based
  • Trait unification: Sampling*ProcessExt, simplified Process trait
  • Updated rand to 0.9, ndarray to 0.17, rayon and other dependencies to latest
  • statrs moved from git dependency to crates.io 0.18.0; copula sampling uses internal SIMD distributions
  • Restructured copulas module
  • Improved quant module API
  • Improved FGN and FBM performance

Fixed

  • Fractional diffusions, fractional noises, and fractional processes
  • AR models, jump processes, volatility models, CGMY
  • Numerical Malliavin, Poisson/CompoundPoisson
  • CUDA feature compiler errors
  • Clippy and formatter warnings

Removed

  • Old sampling traits and test files
  • portable-simd feature flag
  • f32/f64 feature annotations (replaced by generics)
  • impl_new macro

Full Changelog: v.0.16.0...v1.0.0

v.0.16.0

24 Jan 10:36

Choose a tag to compare

  • Rewrite of FGN CUDA kernels and make generation more optimized and faster
  • Add support for PhastFT generate fractional processes even faster on CPU

Full Changelog: v.0.15.4...v.0.16.0

v.0.15.4

31 Dec 12:36

Choose a tag to compare

Fixes:

  • Fgn CUDA random seed

Full Changelog: v.0.15.3...v.0.15.4

v.0.15.1-3

07 Dec 00:24

Choose a tag to compare

v.0.15.0

05 Nov 17:54

Choose a tag to compare

Release Notes

  • stochastic module generic over T now supports f64 (better precision) and f32 (better performance)
  • constraint bounds for Heston calibration
  • updated distribution module for more performance

Full Changelog: v.0.14.0...v.0.15.0

v.0.14.0

29 Oct 22:15

Choose a tag to compare

🚀 Release Notes

🌟 Highlights

💥 Breaking Changes

  • Rename Sampling*Sampling*Ext
    (API change; update imports/usages) (298f994)
  • Removed portable-simd feature and macOS BLAS support
    (91e22b1), (d8e6ba4), (2ce548e)

✨ Added

🧩 Fixed

Full Changelog: v.0.13.3...v.0.14.0

v.0.13.3

12 Apr 21:55

Choose a tag to compare

📢 What’s New

  • Simd Accelerated Distributions: link

Full Changelog: v.0.13.1...v.0.13.2

v.0.13.1

08 Apr 08:52

Choose a tag to compare

📢 What’s New

  • Itô Formula Numerical Solver — Now supports both diffusion and jump-diffusion processes.
  • General SDE Solver — Added support for Euler–Maruyama, Milstein, SRK2, and SRK4 (Runge–Kutta) schemes. Compatible with both ordinary and fractional noise.

🔗 Full Changelog: v.0.13.0...v.0.13.1