# Copyright 2021 IRT Saint Exupéry, https://www.irt-saintexupery.com
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License version 3 as published by the Free Software Foundation.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
# Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser General Public License
# along with this program; if not, write to the Free Software Foundation,
# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
# Contributors:
# INITIAL AUTHORS - initial API and implementation and/or
# initial documentation
# :author: Francois Gallard
# OTHER AUTHORS - MACROSCOPIC CHANGES
"""Common tools for testing opt libraries."""
from __future__ import annotations
from typing import TYPE_CHECKING
from typing import Any
import numpy as np
from gemseo.algos.opt.opt_factory import OptimizersFactory
from gemseo.problems.optimization.power_2 import Power2
from gemseo.problems.optimization.rastrigin import Rastrigin
from gemseo.problems.optimization.rosenbrock import Rosenbrock
if TYPE_CHECKING:
from numpy._typing import NDArray
from gemseo.algos.opt.optimization_library import OptimizationLibrary
from gemseo.algos.opt_problem import OptimizationProblem
[docs]
class OptLibraryTestBase:
"""Main testing class."""
[docs]
@staticmethod
def relative_norm(x: NDArray[float], x_ref: NDArray[float]) -> float:
"""Compute a relative Euclidean norm between to vectors.
Args:
x: The vector.
x_ref: The reference vector.
Returns:
The relative Euclidean norm between two vectors.
"""
xr_norm = np.linalg.norm(x_ref)
if xr_norm < 1e-8:
return np.linalg.norm(x - x_ref)
return np.linalg.norm(x - x_ref) / xr_norm
[docs]
@staticmethod
def norm(x: NDArray[float]) -> float:
"""Compute the Euclidean norm of a vector.
Args:
x: The vector.
Returns:
The Euclidean norm of the vector.
"""
return np.linalg.norm(x)
[docs]
@staticmethod
def generate_one_test(
opt_lib_name: str, algo_name: str, **options: Any
) -> OptimizationLibrary:
"""Solve the Power 2 problem with an optimization library.
This optimization problem has equality constraints.
Args:
opt_lib_name: The name of the optimization library.
algo_name: The name of the optimization algorithm.
**options: The options of the optimization algorithm.
Returns:
An optimization library after the resolution of the Power 2 problem.
"""
problem = OptLibraryTestBase().get_pb_instance("Power2")
opt_library = OptimizersFactory().create(opt_lib_name)
opt_library.execute(problem, algo_name=algo_name, **options)
return opt_library
[docs]
@staticmethod
def generate_one_test_unconstrained(opt_lib_name, algo_name, **options):
"""Solve the Rosenbrock problem with an optimization library.
This optimization problem has no constraints.
Args:
opt_lib_name: The name of the optimization library.
algo_name: The name of the optimization algorithm.
**options: The options of the optimization algorithm.
Returns:
An optimization library after the resolution of the Rosenbrock problem.
"""
problem = OptLibraryTestBase().get_pb_instance("Rosenbrock")
opt_library = OptimizersFactory().create(opt_lib_name)
opt_library.execute(problem, algo_name=algo_name, **options)
return opt_library
[docs]
@staticmethod
def generate_error_test(opt_lib_name, algo_name, **options):
"""Solve the Power 2 problem with an optimization library.
This optimization problem has constraints.
This problem raises an error when calling the objective.
Args:
opt_lib_name: The name of the optimization library.
algo_name: The name of the optimization algorithm.
**options: The options of the optimization algorithm.
Returns:
An optimization library after the resolution of the Rosenbrock problem.
"""
problem = Power2(exception_error=True)
opt_library = OptimizersFactory().create(opt_lib_name)
opt_library.execute(problem, algo_name=algo_name, **options)
return opt_library
[docs]
@staticmethod
def run_and_test_problem(
problem: OptimizationProblem, opt_library: str, algo_name: str, **options: Any
) -> str | None:
"""Run and test an optimization algorithm.
Args:
problem: The optimization problem.
opt_library: The name of the optimization library.
algo_name: The name of the optimization algorithm.
**options: The options of the optimization algorithm.
Returns:
The error message if the optimizer cannot find the solution,
otherwise ``None``.
"""
opt = opt_library.execute(problem, algo_name=algo_name, **options)
x_opt, f_opt = problem.get_solution()
x_err = OptLibraryTestBase.relative_norm(opt.x_opt, x_opt)
f_err = OptLibraryTestBase.relative_norm(opt.f_opt, f_opt)
if x_err > 1e-2 or f_err > 1e-2:
pb_name = problem.__class__.__name__
return (
"Optimization with "
+ algo_name
+ " failed to find solution of problem "
+ pb_name
+ " after n calls = "
+ str(len(problem.database))
)
return None
[docs]
@staticmethod
def create_test(
problem: OptimizationProblem,
opt_library: str,
algo_name: str,
options: dict[str, Any],
):
"""Create a function to run and test an optimization algorithm.
Args:
problem: The optimization problem.
opt_library: The name of the optimization library.
algo_name: The name of the optimization algorithm.
options: The options of the optimization algorithm.
Returns:
The error message if the optimizer cannot find the solution,
otherwise ``None``.
"""
def test_algo(self=None) -> None:
"""Test the algorithm.
Raises:
RuntimeError: When the algorithm cannot find the solution.
"""
msg = OptLibraryTestBase.run_and_test_problem(
problem, opt_library, algo_name, **options
)
if msg is not None:
raise RuntimeError(msg)
return msg
return test_algo
[docs]
@staticmethod
def get_pb_instance(
pb_name: str,
pb_options: dict[str, Any] | None = None,
) -> Rosenbrock | Power2 | Rastrigin:
"""Return an optimization problem.
Args:
pb_name: The name of the optimization problem.
pb_options: The options to be passed to the optimization problem.
Raises:
ValueError: When the problem is not available.
"""
if pb_options is None:
pb_options = {}
if pb_name == "Rosenbrock":
return Rosenbrock(2, **pb_options)
if pb_name == "Power2":
return Power2(**pb_options)
if pb_name == "Rastrigin":
return Rastrigin(**pb_options)
msg = f"Bad pb_name argument: {pb_name}"
raise ValueError(msg)
[docs]
def generate_test(self, opt_lib_name, get_options=None, get_problem_options=None):
"""Generates the tests for an opt library Filters algorithms adapted to the
benchmark problems.
Args:
opt_lib_name: The name of the optimization library.
get_options: A function to get the options of the algorithm.
get_problem_options: A function to get the options of the problem.
Returns!
The test methods to be attached to a unitest class.
"""
tests = []
factory = OptimizersFactory()
if factory.is_available(opt_lib_name):
opt_lib = OptimizersFactory().create(opt_lib_name)
for pb_name in ["Rosenbrock", "Power2", "Rastrigin"]:
if get_problem_options is not None:
pb_options = get_problem_options(pb_name)
else:
pb_options = {}
problem = self.get_pb_instance(pb_name, pb_options)
algos = opt_lib.filter_adapted_algorithms(problem)
for algo_name in algos:
# Reinitialize problem between runs
problem = self.get_pb_instance(pb_name, pb_options)
if get_options is not None:
options = get_options(algo_name)
else:
options = {"max_iter": 10000}
test_method = self.create_test(problem, opt_lib, algo_name, options)
name = "test_" + opt_lib.__class__.__name__ + "_" + algo_name
name += "_on_" + problem.__class__.__name__
name = name.replace("-", "_")
test_method.__name__ = name
tests.append(test_method)
return tests
