Skip to main content
Callbacks can be used to monitor, log, or modify the optimization behavior. All hook methods have default no-op implementations, so subclasses only need to override the methods they care about.

​
Methods

​
on_optimization_start

Called once before the optimization loop begins. Parameters
​
context
CallbackContext
required
Context containing initial system state and minimizer reference.

​
on_step_end

Called after each optimization step. This is the main hook for monitoring and logging during optimization. All callbacks execute even if early stopping is triggered. Parameters
​
context
CallbackContext
required
Context containing current step information, systems, and metrics.

​
on_optimization_end

Called once after the optimization loop completes. Parameters
​
context
CallbackContext
required
Context containing final system state and minimizer reference.

​
Example

class MyCallback(Callback):
    def on_step_end(self, context: CallbackContext) -> None:
        print(f"Step {context.step}: energy = {context.metrics['system_energy']}")