""" Copyright, the CVXPY authors Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. """ import unittest import numpy as np import cvxpy as cp from cvxpy.reductions.solvers.defines import INSTALLED_SOLVERS, QP_SOLVERS from cvxpy.reductions.solvers.qp_solvers.osqp_qpif import OSQP from cvxpy.tests.base_test import BaseTest class TestParamQuadProg(BaseTest): def setUp(self) -> None: self.solvers = [x for x in QP_SOLVERS if x in INSTALLED_SOLVERS] # Overridden method to assume lower accuracy. def assertItemsAlmostEqual(self, a, b, places: int = 2) -> None: super(TestParamQuadProg, self).assertItemsAlmostEqual(a, b, places=places) # Overridden method to assume lower accuracy. def assertAlmostEqual(self, a, b, places: int = 2) -> None: super(TestParamQuadProg, self).assertAlmostEqual(a, b, places=places) def test_param_data(self) -> None: for solver in self.solvers: np.random.seed(0) m = 30 n = 20 A = np.random.randn(m, n) b = np.random.randn(m) x = cp.Variable(n) gamma = cp.Parameter(nonneg=True) gamma_val = .5 gamma_val_new = .1 objective = cp.Minimize(gamma * cp.sum_squares(A @ x - b) + cp.norm(x, 1)) constraints = [1 <= x, x <= 2] # Solve from scratch (directly new parameter) prob = cp.Problem(objective, constraints) self.assertTrue(prob.is_dpp()) gamma.value = gamma_val_new data_scratch, _, _ = prob.get_problem_data(solver) prob.solve(solver=solver) x_scratch = np.copy(x.value) # Canonicalize problem with parameter values (solve once) prob = cp.Problem(objective, constraints) gamma.value = gamma_val data_param, _, _ = prob.get_problem_data(solver) prob.solve(solver=solver) # Get data with new parameter gamma.value = gamma_val_new data_param_new, _, _ = prob.get_problem_data(solver) prob.solve(solver=solver) x_gamma_new = np.copy(x.value) # Check if data match np.testing.assert_allclose(data_param_new['P'].todense(), data_scratch['P'].todense()) # Check if solutions match np.testing.assert_allclose(x_gamma_new, x_scratch, rtol=1e-02, atol=1e-02) def test_qp_problem(self) -> None: for solver in self.solvers: m = 30 n = 20 A = np.random.randn(m, n) b = np.random.randn(m) x = cp.Variable(n) gamma = cp.Parameter(nonneg=True) gamma.value = .5 objective = cp.Minimize(cp.sum_squares(A @ x - b) + gamma * cp.norm(x, 1)) constraints = [0 <= x, x <= 1] # Solve from scratch problem = cp.Problem(objective, constraints) problem.solve(solver=solver) x_full = np.copy(x.value) # Restore cached values solving_chain = problem._cache.solving_chain solver = problem._cache.solving_chain.solver inverse_data = problem._cache.inverse_data param_prog = problem._cache.param_prog # Solve parametric data, solver_inverse_data = solving_chain.solver.apply(param_prog) inverse_data = inverse_data + [solver_inverse_data] raw_solution = solver.solve_via_data( data, warm_start=False, verbose=False, solver_opts={}) problem.unpack_results(raw_solution, solving_chain, inverse_data) x_param = np.copy(x.value) np.testing.assert_allclose(x_param, x_full, rtol=1e-2, atol=1e-02) # TODO: Add derivatives and adjoint tests def test_var_bounds(self) -> None: """Test that lower and upper bounds on variables are propagated.""" # Create a solver instance where bounded variables are disabled. solver_instance = OSQP() solver_instance.name = lambda: "Custom OSQP, no bounded variables" solver_instance.BOUNDED_VARIABLES = False lower_bounds = -10 upper_bounds = np.arange(6).reshape((3, 2)) x = cp.Variable((3, 2), bounds=[lower_bounds, upper_bounds]) problem = cp.Problem(cp.Minimize(cp.sum(x))) data, _, _ = problem.get_problem_data(solver=solver_instance) param_quad_prog = data[cp.settings.PARAM_PROB] assert param_quad_prog.lower_bounds is None assert param_quad_prog.upper_bounds is None # Create a solver instance where bounded variables are enabled. solver_instance = OSQP() solver_instance.name = lambda: "Custom OSQP, bounded variables" solver_instance.BOUNDED_VARIABLES = True lower_bounds = -10 upper_bounds = np.arange(6).reshape((3, 2)) x = cp.Variable((3, 2), bounds=[lower_bounds, upper_bounds]) problem = cp.Problem(cp.Minimize(cp.sum(x))) data, _, _ = problem.get_problem_data(solver=solver_instance) param_quad_prog = data[cp.settings.PARAM_PROB] assert np.all(param_quad_prog.lower_bounds == lower_bounds) param_upper_bound = np.reshape(param_quad_prog.upper_bounds, (3, 2), order="F") assert np.all(param_upper_bound == upper_bounds) @unittest.skipUnless(cp.DAQP in INSTALLED_SOLVERS, 'DAQP is not installed.') def test_daqp_var_bounds(self) -> None: """Testing variable bounds problem with DAQP.""" x1 = cp.Variable(bounds=[-1,1]) x2 = cp.Variable(bounds=[-.5,1]) x3 = cp.Variable() objective = (x1**2 + x2**2)/2 + x1 + x2 + x3 constraints = [-3<=x1+x2, x1+x2<=3, -4<=x1-x2, x1-x2<=4, x3>=-2] prob = cp.Problem(cp.Minimize(objective), constraints) data, _, _ = prob.get_problem_data(solver=cp.DAQP) param_quad_prog = data[cp.settings.PARAM_PROB] assert np.all(param_quad_prog.lower_bounds == np.array([-1, -.5, -np.inf])) assert np.all(param_quad_prog.upper_bounds == np.array([1, 1, np.inf])) prob.solve(solver=cp.DAQP) assert np.isclose(x1.value, -1) assert np.isclose(x2.value, -.5) assert np.isclose(x3.value, -2)