odesign.rs

What It Is

odesign is an optimal design of experiments library written in pure Rust, aimed at finding optimal designs for (generalized) linear models. It is not a SaaS product or a general-purpose visualization tool; it is more of a low-level Rust crate for statistical modeling, experimental design, and numerical optimization research. The documentation provides links to docs.rs, the SourceHut project page, and a read-only GitHub mirror.

Core Capabilities

Based on the crawled content, odesign has three core capabilities: Feature derive, arbitrary optimalities, and optimal design solvers. Developers can define differentiable model features via #[derive(Feature)] and the FeatureFunction trait, then combine APIs such as FeatureSet, LinearModel, DOptimality, and OptimalDesign to formulate optimal design problems. The example shows how to compute a D-optimal design for a simple polynomial model 1 + x on the interval [-1,+1], producing two support points and their weights.

Language, Ecosystem, and Documentation

It is explicitly implemented in pure Rust, and the examples rely on nalgebra and num_dual, making it a good fit for users already working within the Rust numerical computing ecosystem. The documentation is provided in book form, with high-level introductions, theoretical background, use cases, and code snippets that help professional users quickly understand its abstractions. However, based on the available text, the documentation leans more toward research context and code examples; there is no clear evidence of a systematic tutorial, error-handling guide, performance benchmarks, version compatibility notes, or complete integration examples.

Pricing and Open Source

The crawled text does not mention commercial pricing, paid support, or a hosted service. Since the project points to docs.rs, SourceHut, and a GitHub mirror, it can be understood as being distributed as an open-source Rust library. That said, license details do not appear in the main text, so users should still check the repository metadata before adopting it in a rigorous setting.

Pros, Cons, and Who It’s For

Its strengths are flexible model expression, support for custom boundaries and arbitrary optimality criteria, and the use of Rust’s performance and type system. It is suitable for experimental design researchers, statistical method developers, and Rust engineering projects that need to embed optimal design capabilities. The drawbacks are a relatively high learning curve and limited accessibility for non-Rust users; there is no visible information about Python/R bindings, commercial support, SLAs, or rich ecosystem integrations.

Access from China

Availability from mainland China is unknown. docs.rs, SourceHut, and GitHub may vary in speed or accessibility depending on the network environment. There is no payment information. If your team depends on the Python/R ecosystem, it may also be worth evaluating related experimental design and statistical modeling libraries as alternatives.