Skip to content

feat: own our calibration bounds instead of SYMFLUENCE's catalogue - #6

Open
DarriEy wants to merge 1 commit into
mainfrom
feat/own-calibration-bounds
Open

DarriEy wants to merge 1 commit into
mainfrom
feat/own-calibration-bounds

Conversation

@DarriEy

@DarriEy DarriEy commented Jul 28, 2026

Copy link
Copy Markdown
Collaborator

This package predates SYMFLUENCE's register_model_bounds() seam, so HECHMS's calibration bounds live in the framework's shared catalogue, reached through symfluence.optimization.core.parameter_bounds_registry — a module deprecated for removal at SYMFLUENCE 2.0. Changing our own bounds currently requires a framework release.

register() now contributes them through the seam, so get_model_bounds('HECHMS') resolves what this package registers, ahead of the framework's built-in entry. This works against current SYMFLUENCE and lets a later framework release drop the compatibility entry.

No calibration result changes. Values come from this package's own PARAM_BOUNDS, already the source for every other bounds consumer here, and verified identical to what the framework served: 14 names, matching on min, max and transform, with bounds_registration_conflicts() empty.

Part of the SYMFLUENCE service-decomposition work: the framework is removing per-model knowledge from its core so the model suite can be extracted into its own distribution.

🤖 Generated with Claude Code

This package predates SYMFLUENCE's register_model_bounds() seam, so HECHMS's
calibration bounds lived in the FRAMEWORK's shared catalogue and reached us
through symfluence.optimization.core.parameter_bounds_registry -- a module
deprecated for removal at SYMFLUENCE 2.0. Changing our own bounds therefore
required a framework release.

register() now contributes them through the seam, so get_model_bounds('HECHMS')
resolves what this package registers, ahead of the framework's built-in entry.
That works against current SYMFLUENCE and lets a later release drop the
compatibility entry.

No calibration result changes: the values come from this package's own
PARAM_BOUNDS, which were already the source for every other bounds consumer
here, and were verified identical to what the framework served -- 14 names,
matching on min, max AND transform, with no registration conflicts.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant