hydromodpy.calibration

Calibration sub-system.

Public surface:

• CalibrationEngine: the orchestrator

• Optimizer / Objective / Evaluator: Protocol contracts

• Calibrable: Pydantic-field annotation marking a calibrable parameter

• discover_calibrable: auto-discover calibrable fields in a config tree

• build_optimizer: adapter registry lookup

Modules

adapters	Optimizer adapters: scipy, optuna, grid.
cache	Content-addressable calibration cache.
cases	Reference calibration cases ported from the legacy analysis module.
cli_runner	CLI entry point hmp calibrate <calibration.toml>.
config	Pydantic model for the [calibration] TOML section.
diagnostics	Lightweight diagnostics helpers for calibration iteration traces.
engine	CalibrationEngine - orchestrates an ask/tell loop.
lumped	Lumped catchment models used as calibration targets.
materialize	Materialise a self-contained TOML overlay for one calibration candidate.
method_config	Discriminated union for the calibration method + optimizer kwargs.
metrics	RAM metric extraction for lightweight calibration trials.
natural_observations	Natural-observation helpers for network/discharge calibration.
network_transient_truth	Truth-package helpers for network/transient calibration prototypes.
objective	Objective Protocol - computes a cost from observed vs simulated data.
objective_mapping	Parse evaluated objective points from calibration iteration histories.
optimizer	Optimizer Protocol and registry.
parameters	Calibrable parameters: annotations, space, transforms, discovery.
persistence	DuckDB persistence for calibration sessions and iterations.
programmatic_runner	Programmatic entry point hmp.calibrate(project, cfg).
promotion	Top-N promotion of calibration trials.
report	Calibration session data structures and loaders.
reporting	Reusable calibration report builders.
runners	Calibration runners (trial drivers) layer.
state	Shared state primitives reused by both calibration runners.