SPEAR-QCP with Multi-Fidelity on Instances

We optimize the SPEAR algorithm on QCP to demonstrate the powerful SMAC4AC facade. Algorithm and instance definition is done inside scenario file.

Moreover, we present you an alternative intensification procedure “Successive Halving”.

import logging

logging.basicConfig(level=logging.INFO)

from smac.facade.smac_ac_facade import SMAC4AC
from smac.intensification.successive_halving import SuccessiveHalving
from smac.scenario.scenario import Scenario

__copyright__ = "Copyright 2021, AutoML.org Freiburg-Hannover"
__license__ = "3-clause BSD"


if __name__ == "__main__":
    scenario = Scenario("examples/commandline/spear_qcp/scenario.txt")

    # provide arguments for the intensifier like this
    intensifier_kwargs = {
        "n_seeds": 2,  # specify the number of seeds to evaluate for a non-deterministic target algorithm
        "initial_budget": 1,
        "eta": 3,
        "min_chall": 1,  # because successive halving cannot handle min_chall > 1
    }

    smac = SMAC4AC(
        scenario=scenario,  # scenario object
        intensifier_kwargs=intensifier_kwargs,  # arguments for Successive Halving
        # change intensifier to successive halving by passing the class.
        # it must implement `AbstractRacer`.
        intensifier=SuccessiveHalving,
    )

    # Start optimization
    try:
        incumbent = smac.optimize()
    finally:
        incumbent = smac.solver.incumbent

    print("Optimized configuration %s" % str(incumbent))

previous

Restore Branin

next

Simple Multi-Objective