BaseCallback

class lumopt2.utils.callbacks.BaseCallback

Base class for optimization callbacks.

Callbacks are invoked at various points during optimization to enable logging, visualization, checkpointing, or other monitoring tasks.

All hook methods default to no-ops, so subclasses only need to override the events they actually care about. Concrete callbacks typically implement either the *_iteration_* pair (for per-step bookkeeping such as the visualizer) or just on_function_eval (for everything-fires-on-every-eval loggers); both flavours coexist happily with the no-op defaults supplied here.

Hooks fire in this order:

1. on_optimization_start – once at the beginning.

2. on_function_eval – after each objective-function evaluation.

3. on_iteration_start – before each iteration (if the optimizer exposes iteration boundaries).

4. on_iteration_end – after each iteration (idem).

5. on_optimization_end – once at the end.

SciPy-driven optimizers expose both per-iteration and per-evaluation hooks; gradient-free or single-pass optimizers may only fire on_function_eval. Subclasses that only care about iteration boundaries can therefore safely leave on_function_eval at the no-op default and vice-versa.

Methods

BaseCallback.on_function_eval(project, ...)	Called after each objective function evaluation.
BaseCallback.on_iteration_end(project, ...)	Called at the end of each iteration.
BaseCallback.on_iteration_start(iteration, ...)	Called at the start of each iteration.
BaseCallback.on_optimization_end(success, ...)	Called when optimization completes.
BaseCallback.on_optimization_start(project, ...)	Called when optimization begins.