Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Scientific representations

Scientific representations sit between mathematical methods and research workflows. They specify what an array means before a numerical method acts on it and before a workflow interprets the result. A representation fixes conventions such as units, coordinates, sampling semantics, static metadata, and the boundary of a scientific claim.

Choosing an array dtype and shape is not the same as choosing a scientific representation. Two arrays can have identical values and shapes while representing different physical dimensions, coordinate frames, spectral densities, or parameter constraints. Jaxstro makes those distinctions explicit where the runtime supports them.

MaterialStatusRuntime boundary
Units, coordinates, spectra, atmospheres, parameters, and stateCurrentImplemented Jaxstro owners with explicit contracts
Uncertainty propagationPlannedNo jaxstro.uncertainty runtime module
Fields and discretized domainsDeferredCommon runtime awaits evidence from two consumers

How to read this section

Each implemented capability page begins with a seven-row representation contract. Read that table before the examples: it names the mathematical object, convention, runtime owner, shape and unit policy, JAX transform boundary, evidence, and downstream interpretation boundary. Planned and deferred guides instead begin with an exact status and keep their derivation, failure modes, evidence gate, and claim boundary visible.

Status labels separate implemented Jaxstro capabilities from planned propagation work and deferred field abstractions. Numerical algorithms live under Numerical methods, research workflows remain at their current published locations until their dedicated migration, and validation anchors live in Validation.