Changelog.md

Changelog

All notable Changes to the Julia package AlgorithmsInterface.jl are documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.1.1] unreleased

Added

• is_active(stopping_criterion, stopping_criterion_state), the machine-readable counterpart of get_reason, reporting whether a criterion became active during the current run. It defaults to reading the state's at_iteration and is what the generic convergence reporting is now built on, rather than !isnothing(get_reason(...)).
• get_active_stopping_criteria, reporting which criteria became active. It recurses through nested combinations, so it separates stopping on a collapsed step size from stopping on an exhausted iteration budget — a distinction indicates_convergence is too coarse for.
• Convenience two-argument get_reason(algorithm, state), and likewise for is_active, indicates_convergence and get_active_stopping_criteria, extracting the criterion and its state the way is_finished already did.
• StopReasonAction, a LoggingAction that reports get_reason at the :Stop context.
• Defaults for get_reason (nothing) and for the type-domain indicates_convergence (false), so a criterion that implements neither no longer hits a MethodError from the derived convergence reporting.
• Exports for DefaultStoppingCriterionState, StopAfterTimePeriodState and GroupStoppingCriterionState, which a downstream criterion is expected to reuse.
• DefaultState, a State with the iterate, iteration and stopping_criterion_state fields, and a data field for any additional data.
• Default initialize_state and initialize_state! methods using a DefaultState, so an algorithm only has to provide a Problem, an Algorithm and a step!.
• Default initialize_state and initialize_state! methods for any StoppingCriterion, using a DefaultStoppingCriterionState.

Changed

• indicates_convergence without a state moved to the type domain: a new criterion implements indicates_convergence(::Type{YourCriterion}), and indicates_convergence(criterion) forwards to it. StopWhenAll and StopWhenAny combine their children in the type domain as well, so a criterion that only implements the variant taking an instance is no longer accounted for in a group.
• StopAfterIteration now uses the default initialize_state and initialize_state!.

Fixed

• indicates_convergence(::StoppingCriterion, ::StoppingCriterionState) had its condition inverted, claiming convergence exactly when the criterion had not indicated to stop.
• indicates_convergence for StopWhenAll and StopWhenAny now consults the children that actually indicated to stop. Previously a StopWhenAny pairing a convergence criterion with a fallback such as StopAfterIteration could never report convergence, since the criteria-only variant is all(indicates_convergence, criteria).
• is_finished! for StopWhenAll and StopWhenAny short-circuited, skipping the children after the deciding one. Stateful criteria were starved of the current iterate and their at_iteration left unset even when they did indicate to stop; every child is now updated once per iteration.
• get_reason for StopWhenAll and StopWhenAny rendered children that had not indicated to stop as the literal text "nothing", and returned "" when no child had a message at all. It now reports only the children that triggered, and nothing when there is nothing to report.
• The non-mutating is_finished for StopWhenAll and StopWhenAny reset at_iteration at iteration 0, contradicting its contract and silently clearing a group's recorded stop.
• The non-mutating is_finished for StopAfter reported the elapsed time recorded by the last is_finished! rather than reading the clock, so it could answer "not finished" for a run long past its threshold.
• get_reason for StopAfterIteration was gated on at_iteration >= max_iterations while its Base.summary was gated on at_iteration >= 0, so the two could disagree.

[0.1.0] 2026-05-01

Initial release.