API reference¶
The reference is built from the source using mkdocstrings. Start with the quick start for a complete example.
| Area | Public entry points |
|---|---|
| Computation | orivex.compute, orivex.list_features |
| Samples | orivex.LandscapeSample, orivex.ObjectiveSense |
| Results | orivex.result.ComputationResult, FeatureValue, ExecutionMetadata |
| Specifications | orivex.FeatureSpec, costs, invariances, and capabilities |
| PyTorch | orivex.torch.compute, TensorLandscapeSample, discovery functions |
The NumPy API is available from orivex. Import the optional tensor backend explicitly
from orivex.torch. Import result models from orivex.result and feature selection errors
from orivex.registry.
Most users need only a sample and compute; the engine, planner, and intermediate calculators
are implementation details.