omnibias.struct.select
Differentiable dynamic programs for sequences and parses, structured distributions, and finite-temperature bounds.
import omnibias.struct.selectModule contract & mathematical conventions
Certified selection (measure-mode collapse): the backend-agnostic public surface.
Re-exports the numpy certificate surface of omnibias.struct._core.select at the struct
root (mirroring how omnibias.struct.certify_soft_dp sits alongside the DP layers). The
differentiable soft-selection ops and Gibbs moments live in the backend twins
omnibias.struct.torch.select / omnibias.struct.jax.select.
This is the beta -> inf *feasibility / temperature* collapse of a Gibbs law onto its mode --
**not** the founding delta -> 0 bias collapse; the derivative tower is only the exact engine
that differentiates lse_beta (do not conflate the two axes).
Shared formulas do not imply universal floating-point identity. Verify dtype, shape, derivative-order support, and the stated numerical assumptions for the path you use.
Imports & aliases
These names share a definition with another public module. Follow a name for its full contract, parameters, and examples.
| Import from this module | Canonical reference |
|---|---|
SelectionCertificate | omnibias.struct.SelectionCertificate |
argmax_stability_margin | omnibias.struct.argmax_stability_margin |
beta_for_confidence | omnibias.struct.beta_for_confidence |
certify_argmax | omnibias.struct.certify_argmax |
mass_concentration_bound | omnibias.struct.mass_concentration_bound |
seal_selection_certificate | omnibias.struct.seal_selection_certificate |