Alkahest — Python API reference

Installation

Python — install from PyPI. Supported versions: 3.9 through 3.13:

python -m pip install -U pip
pip install alkahest

That wheel includes the Gröbner solver (groebner) and egglog (egraph) by default, so alkahest.solve and related APIs work out of the box. It omits LLVM JIT, Cranelift, and parallel (numeric APIs fall back to the interpreter). There is no pip extra that swaps in a different native binary.

For Linux x86_64 opt-in wheels with local versions +jit or +full, use assets attached to GitHub Releases. Other platforms: build from source with maturin.

Rust — add alkahest-cas to your Cargo.toml:

[dependencies]
alkahest-cas = "2"
# alkahest-cas = { version = "2", features = ["groebner", "parallel", "egraph"] }

Requires libflint-dev / libgmp-dev / libmpfr-dev at build time (apt-get or brew install flint). See docs.rs/alkahest-cas for the full Rust API reference.

For optional Cargo features (jit, parallel, cuda, …) and full developer setup, see the Getting started chapter of the user guide. The groebner and egraph features are defaults and do not need to be specified explicitly.

Core

• Core API
  • ExprPool
  • Expr
  • DerivedResult
  • Context manager
  • Parsing
• Simplification API
  • Rule-based simplification
  • Utility passes
  • E-graph simplification
  • Pattern matching and rules
• Calculus API
  • Differentiation
  • Integration
  • Primitives (math functions)
  • Piecewise
• Polynomial API
  • UniPoly
  • MultiPoly
  • Sparse interpolation
  • RationalFunction
  • Polynomial utilities
• Numerics and evaluation API
  • Compiled evaluation
  • Ball arithmetic
  • GPU evaluation
• Transformation API
  • Tracing
  • Gradient
  • JIT compilation
  • TracedFn
  • PyTree utilities

Advanced

• Matrix API
  • Matrix
• ODE and DAE API
  • ODE
  • DAE
  • Sensitivity analysis
  • Acausal modeling
  • Hybrid systems
• Solver API
  • solve()
  • GroebnerBasis
• Code generation API
  • MLIR / StableHLO
  • C code emission
  • Horner form
  • Primitive registry
• Error handling API
  • Base class
  • Exception subclasses
  • Catching errors by subsystem

For the conceptual guide (kernel design, rule engine, e-graph, derivation logs) see the mdBook user guide.