Skip to main content
Callbacks let you run custom code at defined points during an experiment simulation without modifying the solver or model. They are attached to sim.solve() and invoked automatically as the solver progresses through cycles and steps.
The Callback API is experimental — the interface may change in future releases.

​
Built-in callbacks

​
LoggingCallback

Added automatically if no LoggingCallback is present in the list passed to sim.solve(). It logs progress to pybamm.logger or an optional log file: 
import pybamm

# Log to console (default)
cb = pybamm.LoggingCallback()

# Log to a file instead
cb = pybamm.LoggingCallback(logfile="simulation.log")
LoggingCallback emits:
  • NOTICE on experiment start, each cycle start, each step start, and experiment end
  • WARNING when an experiment step is infeasible
  • ERROR on solver errors

​
CallbackList

CallbackList wraps multiple callbacks so they can be dispatched as one: 
callbacks = pybamm.callbacks.CallbackList([
    pybamm.LoggingCallback(),
    my_custom_callback,
])
You normally do not need to construct CallbackList manually — setup_callbacks() does this for you and ensures a LoggingCallback is always present.

​
Callback interface

Every callback method receives a single logs dict. The table below lists each hook and the keys available in logs at that point:
MethodWhen calledKey logs entries
on_experiment_start(logs)Before the first cycle—
on_cycle_start(logs)Before each cycle"cycle number" (current, total) tuple; "elapsed time"
on_step_start(logs)Before each step"cycle number", "step number" (current, total) tuple, "step operating conditions"
on_step_end(logs)After each step completes"stopping conditions" → {"time": ...}, "experiment time"
on_cycle_end(logs)After each cycle completes"stopping conditions" → {"capacity": ..., "voltage": ...}, "summary variables", "start capacity", "Minimum voltage [V]"
on_experiment_end(logs)After the last cycle"elapsed time"
on_experiment_error(logs)When a SolverError occurs"error"
on_experiment_infeasible_time(logs)Step hits default duration"step duration", "cycle number", "step number", "step operating conditions"
on_experiment_infeasible_event(logs)Step terminated by an event"termination", "cycle number", "step number", "step operating conditions"

​
Creating a custom callback

Subclass pybamm.callbacks.Callback and override whichever hooks you need. Methods that are not overridden default to pass.
save_callback.py
import json
import pybamm


class StepDataCallback(pybamm.callbacks.Callback):
    """
    Saves the step operating conditions and elapsed time to a JSON file
    at the end of each step.
    """

    def __init__(self, output_path="step_log.json"):
        self.output_path = output_path
        self.records = []

    def on_step_end(self, logs):
        record = {
            "step": logs["step number"][0],
            "cycle": logs["cycle number"][0],
            "experiment_time_s": logs["experiment time"],
        }
        self.records.append(record)

    def on_experiment_end(self, logs):
        with open(self.output_path, "w") as f:
            json.dump(self.records, f, indent=2)
        pybamm.logger.info(f"Step data written to {self.output_path}")

​
Using callbacks with Simulation

Pass a list of callback instances to the callbacks keyword of sim.solve():
use_callbacks.py
import pybamm

model = pybamm.lithium_ion.SPM()
param = pybamm.ParameterValues("Chen2020")

experiment = pybamm.Experiment(
    [
        "Discharge at 1C until 3.0 V",
        "Rest for 10 minutes",
        "Charge at 0.5C until 4.2 V",
    ]
)

sim = pybamm.Simulation(
    model,
    parameter_values=param,
    experiment=experiment,
)

callbacks = [
    pybamm.LoggingCallback(),
    StepDataCallback(output_path="my_run.json"),
]

sol = sim.solve(callbacks=callbacks)
A LoggingCallback is inserted automatically if you do not include one, so you only need to list callbacks you actually want in addition to the default logging.

​
Example: early-stopping callback

The following callback raises an exception to stop the simulation if the voltage drops below a threshold mid-experiment:
early_stop_callback.py
import pybamm


class EarlyStopCallback(pybamm.callbacks.Callback):
    """Stop the simulation if voltage drops below a threshold."""

    def __init__(self, min_voltage=3.0):
        self.min_voltage = min_voltage

    def on_step_end(self, logs):
        # "stopping conditions" is available after each step
        stop = logs["stopping conditions"]
        # Inspect solver-provided minimum voltage if present
        if stop.get("voltage") is not None:
            v_min = stop["voltage"][0]
            if v_min < self.min_voltage:
                raise pybamm.SolverError(
                    f"Voltage {v_min:.3f} V fell below threshold {self.min_voltage:.3f} V"
                )


sim = pybamm.Simulation(
    pybamm.lithium_ion.SPM(),
    parameter_values=pybamm.ParameterValues("Chen2020"),
    experiment=pybamm.Experiment(["Discharge at 2C until 2.5 V"]),
)
sim.solve(callbacks=[EarlyStopCallback(min_voltage=2.8)])
CallbackList delegates every on_* call to each registered callback using a callback_loop_decorator. The decorator is applied automatically at class definition time by inspecting all methods that start with on_ via inspect.getmembers. This means adding a new on_custom_event method to Callback automatically becomes dispatchable through CallbackList without any changes to CallbackList itself.

Build docs developers (and LLMs) love

Get started for freeTalk to us