romtools.workflows.inverse.vi.adaptation#

Accepted-iteration lifecycle for delayed MF-VI ROM adaptation.

Functions

activate_delayed_rom(state)

Select next iteration's model without changing the accepted estimates.

mf_vi_restart_data(state)

restore_mf_vi_state(restart, builder, ...)

Reconstruct both models; restoring a completed batch draws no samples.

stage_delayed_rom(state, builder, directory, ...)

Train only after acceptance, leaving the estimator's active ROM intact.

validate_mf_vi_restart(restart, ...[, ...])

romtools.workflows.inverse.vi.adaptation.activate_delayed_rom(state)[source]#

Select next iteration’s model without changing the accepted estimates.

romtools.workflows.inverse.vi.adaptation.restore_mf_vi_state(restart, builder, directory, dispatcher)[source]#

Reconstruct both models; restoring a completed batch draws no samples.

romtools.workflows.inverse.vi.adaptation.stage_delayed_rom(state, builder, directory, training_limit, dispatcher)[source]#

Train only after acceptance, leaving the estimator’s active ROM intact.