biophysics_fitting ❭ optimizer ❭ start_run

start_run

biophysics_fitting.optimizer.start_run(db_setup, n, pop=None, client=None, continue_cp=False, offspring_size=1000, eta=10, mutpb=0.3, cxpb=0.7, max_ngen=600, satisfactory_boundary_dict=None)

Start an optimization run as specified in db_setup.

Parameters:

  • db_setup (data_base.DataBase) – a DataBase containing the setup of the optimization. It must include:

    • params … this is a pandas.DataFrame with the parameternames as index and the columns min_ and max_

    • get_Simulator … function, that returns a biophysics_fitting.simulator.Simulator object

    • get_Evaluator … function, that returns a biophysics_fitting.evaluator.Evaluator object.

    • get_Combiner … function, that returns a biophysics_fitting.combiner.Combiner object

    get_Simulator, get_Evaluator, get_Combiner accept the db_setup data_base as argument. This allows, that e.g. the Simular can depend on the data_base. Therefore it is e.g. possible, that the path to the morphology is not saved as absolute path. Instead, fixed parameters can be updated accordingly.

  • n (int) – The run ID. This is used to create a sub-database in db_setup, where the results of the optimization are saved.

  • pop (list of deap.Individuals | None) – The previous population if the optimization is continued. None if a new optimization is started.

  • client (distributed.Client | None) – A distributed client. If None, the optimization is run on the local machine.

  • continue_cp (bool) – If True, the optimization is continued. If False, a new optimization is started.

  • offspring_size (int) – The number of individuals in the offspring.

  • eta (int) – The number of parents selected for each offspring.

  • mutpb (float) – The mutation probability.

  • cxpb (float) – The crossover probability.

  • max_ngen (int) – The maximum number of generations.

  • satisfactory_boundary_dict (dict | None) – A dictionary with the boundaries for the objectives. If a model is found, that has all objectives below the boundary, the optimization is stopped.

For an exemplary setup of a Simulator, Evaluator and Combiner object, see biophysics_fitting.hay_complete_default_setup.

For on examplary setup of a complete optimization project see getting_started/tutorials/1. neuron models/1.3 Generation.ipynb

You can also have a look at the test case in tests.test_biophysics_fitting.optimizer_test.py