API reference - jaxstro

Use this reference when you know the operation you need and want its current Python owner, records, transform behavior, failure contract, and evidence. The groups below are documentation routes; they do not create Python namespaces.

Which method owns the operation?

Find change, approximation, linear, random, and discrete-space owners by the kind of numerical operation they perform.

Which representation owns the data?

Find physical representations and scientific-data owners without hiding units, coordinates, sampling semantics, or provenance.

Which infrastructure records trust?

Find checks, configuration, scientific contracts, evidence, provenance, and testing owners.

Surface	Status	Documentation rule
Owner-qualified module APIs	Current	Canonical reference path
Flat `jaxstro.numerics` callable re-exports	Current legacy inventory	Remain importable until the Project 2 hard cut
Proposed or delegated capabilities	Planned or ecosystem-owned	Remain outside API reference until an owner and contracts exist

Canonical import policy¶

Import a coherent module and qualify the call:

from jaxstro.numerics import rootfinding

result = rootfinding.safeguarded_bracketed_root(...)

Or import a symbol directly from the module that owns it:

from jaxstro.numerics.rootfinding import safeguarded_bracketed_root

Flat callable re-exports from jaxstro.numerics are legacy inventory awaiting Project 2. They remain importable for now, but they are not the canonical documentation path. This reference does not change runtime exports.

Method owners¶

Change and constraints: Automatic differentiation products, Rootfinding, Universal-variable Kepler propagation, Optimization helpers, and Fixed-step differential equations.
Approximation and integration: One-dimensional interpolation, Regular-grid interpolation, B-splines, and jaxstro.quad. The sampled-integration and fixed-quadrature pages document temporary compatibility paths, not current owners.
Linear structure: Linear algebra, Compensated arithmetic, Linear operators, and Special functions and bases.
Randomness: Probability distributions, PRNG key mechanics, Random streams and resampling, Sampling primitives, and Stable statistics helpers.
Discrete space: Grids and conservative binning, Structured meshes, and Spatial indexing and pairs.

Representation and data owners¶

Physical representations: Physical constants, Unit systems, Dimensional quantities, Coordinate transformations, Geometry, Astrometric constants, and Parameter bridges.
Scientific data: Spectral representations and Atmosphere libraries.

Research infrastructure owners¶

See Numerical checks, Numerical type aliases, JAX configuration, Scientific contract registry, Computational evidence, Runtime provenance, and Validation and provenance tooling.

Only current importable surfaces appear here. Proposed capabilities remain in the Methods, Representations, Workflows, and Project roadmaps until runtime owners and executable contracts exist.