❭ API reference ❭ biophysics_fitting ❭

optimizer

Multi-objective optimization algorithm.

This module interfaces with BluePyOpt [1] and provides:

  • a start population can be defined.

  • optimizations organized in a data base.

  • execution on a distributed system using dask.

  • return of all objectives, not only the combined ones.

The top-level pipeline can be started with start_run().

Note

The evolutionary algorithm implementation (adapted from BluePyOpt) is in biophysics_fitting.optimizer_LGPL, which is licensed under LGPL-3.0-or-later.

Functions

robust_int(x)

Robustly convert something to an integer.

get_max_generation(db_run)

Returns the index of the next iteration in a database.

save_result(db_run, features, objectives)

Save the results of an optimization iteration in a database.

setup_db_run(db_setup, run)

Create a sub-database for each run of the optimization algorithm.

get_objective_function(db_setup)

Get the objective function for the optimization.

get_mymap(db_setup, db_run, c, satisfactory_boundary_dict, n_reschedule_on_runtime_error)

Get a map function for evaluating the parameters.

get_population_with_different_n_objectives(old_pop, n_objectives)

Adapt the number of objectives of individuals

start_run(db_setup, n, pop, client, continue_cp, offspring_size, eta, mutpb, cxpb, max_ngen, satisfactory_boundary_dict)

Start an optimization run as specified in db_setup.