blob: fd329a9aa61b6b2f99b47efdefaedf760b89b7f6 [file] [log] [blame]
// Ceres Solver - A fast non-linear least squares minimizer
// Copyright 2015 Google Inc. All rights reserved.
// http://ceres-solver.org/
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
// * Redistributions of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
// * Neither the name of Google Inc. nor the names of its contributors may be
// used to endorse or promote products derived from this software without
// specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
// POSSIBILITY OF SUCH DAMAGE.
//
// Author: sameeragarwal@google.com (Sameer Agarwal)
#include "ceres/trust_region_preprocessor.h"
#include <array>
#include <map>
#include "ceres/internal/config.h"
#include "ceres/ordered_groups.h"
#include "ceres/problem_impl.h"
#include "ceres/sized_cost_function.h"
#include "ceres/solver.h"
#include "gtest/gtest.h"
namespace ceres {
namespace internal {
TEST(TrustRegionPreprocessor, ZeroProblem) {
ProblemImpl problem;
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem, &pp));
}
TEST(TrustRegionPreprocessor, ProblemWithInvalidParameterBlock) {
ProblemImpl problem;
double x = std::numeric_limits<double>::quiet_NaN();
problem.AddParameterBlock(&x, 1);
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_FALSE(preprocessor.Preprocess(options, &problem, &pp));
}
TEST(TrustRegionPreprocessor, ParameterBlockBoundsAreInvalid) {
ProblemImpl problem;
double x = 1.0;
problem.AddParameterBlock(&x, 1);
problem.SetParameterUpperBound(&x, 0, 1.0);
problem.SetParameterLowerBound(&x, 0, 2.0);
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_FALSE(preprocessor.Preprocess(options, &problem, &pp));
}
TEST(TrustRegionPreprocessor, ParameterBlockIsInfeasible) {
ProblemImpl problem;
double x = 3.0;
problem.AddParameterBlock(&x, 1);
problem.SetParameterUpperBound(&x, 0, 1.0);
problem.SetParameterLowerBound(&x, 0, 2.0);
problem.SetParameterBlockConstant(&x);
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_FALSE(preprocessor.Preprocess(options, &problem, &pp));
}
class FailingCostFunction : public SizedCostFunction<1, 1> {
public:
bool Evaluate(double const* const* parameters,
double* residuals,
double** jacobians) const override {
return false;
}
};
TEST(TrustRegionPreprocessor, RemoveParameterBlocksFailed) {
ProblemImpl problem;
double x = 3.0;
problem.AddResidualBlock(new FailingCostFunction, nullptr, &x);
problem.SetParameterBlockConstant(&x);
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_FALSE(preprocessor.Preprocess(options, &problem, &pp));
}
TEST(TrustRegionPreprocessor, RemoveParameterBlocksSucceeds) {
ProblemImpl problem;
double x = 3.0;
problem.AddParameterBlock(&x, 1);
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem, &pp));
}
template <int kNumResiduals, int... Ns>
class DummyCostFunction : public SizedCostFunction<kNumResiduals, Ns...> {
public:
bool Evaluate(double const* const* parameters,
double* residuals,
double** jacobians) const override {
for (int i = 0; i < kNumResiduals; ++i) {
residuals[i] = kNumResiduals * kNumResiduals + i;
}
if (jacobians == nullptr) {
return true;
}
std::array<int, sizeof...(Ns)> N{Ns...};
for (size_t i = 0; i < N.size(); ++i) {
if (jacobians[i] != nullptr) {
MatrixRef j(jacobians[i], kNumResiduals, N[i]);
j.setOnes();
j *= kNumResiduals * N[i];
}
}
return true;
}
};
class LinearSolverAndEvaluatorCreationTest : public ::testing::Test {
public:
void SetUp() final {
x_ = 1.0;
y_ = 1.0;
z_ = 1.0;
problem_.AddResidualBlock(
new DummyCostFunction<1, 1, 1>, nullptr, &x_, &y_);
problem_.AddResidualBlock(
new DummyCostFunction<1, 1, 1>, nullptr, &y_, &z_);
}
void PreprocessForGivenLinearSolverAndVerify(
const LinearSolverType linear_solver_type) {
Solver::Options options;
options.linear_solver_type = linear_solver_type;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_EQ(pp.options.linear_solver_type, linear_solver_type);
EXPECT_EQ(pp.linear_solver_options.type, linear_solver_type);
EXPECT_EQ(pp.evaluator_options.linear_solver_type, linear_solver_type);
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
}
protected:
ProblemImpl problem_;
double x_;
double y_;
double z_;
};
TEST_F(LinearSolverAndEvaluatorCreationTest, DenseQR) {
PreprocessForGivenLinearSolverAndVerify(DENSE_QR);
}
TEST_F(LinearSolverAndEvaluatorCreationTest, DenseNormalCholesky) {
PreprocessForGivenLinearSolverAndVerify(DENSE_NORMAL_CHOLESKY);
}
TEST_F(LinearSolverAndEvaluatorCreationTest, DenseSchur) {
PreprocessForGivenLinearSolverAndVerify(DENSE_SCHUR);
}
#if !defined(CERES_NO_SPARSE)
TEST_F(LinearSolverAndEvaluatorCreationTest, SparseNormalCholesky) {
PreprocessForGivenLinearSolverAndVerify(SPARSE_NORMAL_CHOLESKY);
}
#endif
#if !defined(CERES_NO_SPARSE)
TEST_F(LinearSolverAndEvaluatorCreationTest, SparseSchur) {
PreprocessForGivenLinearSolverAndVerify(SPARSE_SCHUR);
}
#endif
TEST_F(LinearSolverAndEvaluatorCreationTest, CGNR) {
PreprocessForGivenLinearSolverAndVerify(CGNR);
}
TEST_F(LinearSolverAndEvaluatorCreationTest, IterativeSchur) {
PreprocessForGivenLinearSolverAndVerify(ITERATIVE_SCHUR);
}
TEST_F(LinearSolverAndEvaluatorCreationTest, MinimizerIsAwareOfBounds) {
problem_.SetParameterLowerBound(&x_, 0, 0.0);
Solver::Options options;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_EQ(pp.options.linear_solver_type, options.linear_solver_type);
EXPECT_EQ(pp.linear_solver_options.type, options.linear_solver_type);
EXPECT_EQ(pp.evaluator_options.linear_solver_type,
options.linear_solver_type);
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
EXPECT_TRUE(pp.minimizer_options.is_constrained);
}
TEST_F(LinearSolverAndEvaluatorCreationTest, SchurTypeSolverWithBadOrdering) {
Solver::Options options;
options.linear_solver_type = DENSE_SCHUR;
options.linear_solver_ordering = std::make_shared<ParameterBlockOrdering>();
options.linear_solver_ordering->AddElementToGroup(&x_, 0);
options.linear_solver_ordering->AddElementToGroup(&y_, 0);
options.linear_solver_ordering->AddElementToGroup(&z_, 1);
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_FALSE(preprocessor.Preprocess(options, &problem_, &pp));
}
TEST_F(LinearSolverAndEvaluatorCreationTest, SchurTypeSolverWithGoodOrdering) {
Solver::Options options;
options.linear_solver_type = DENSE_SCHUR;
options.linear_solver_ordering = std::make_shared<ParameterBlockOrdering>();
options.linear_solver_ordering->AddElementToGroup(&x_, 0);
options.linear_solver_ordering->AddElementToGroup(&z_, 0);
options.linear_solver_ordering->AddElementToGroup(&y_, 1);
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_EQ(pp.options.linear_solver_type, DENSE_SCHUR);
EXPECT_EQ(pp.linear_solver_options.type, DENSE_SCHUR);
EXPECT_EQ(pp.evaluator_options.linear_solver_type, DENSE_SCHUR);
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
}
TEST_F(LinearSolverAndEvaluatorCreationTest,
SchurTypeSolverWithEmptyFirstEliminationGroup) {
problem_.SetParameterBlockConstant(&x_);
problem_.SetParameterBlockConstant(&z_);
Solver::Options options;
options.linear_solver_type = DENSE_SCHUR;
options.linear_solver_ordering = std::make_shared<ParameterBlockOrdering>();
options.linear_solver_ordering->AddElementToGroup(&x_, 0);
options.linear_solver_ordering->AddElementToGroup(&z_, 0);
options.linear_solver_ordering->AddElementToGroup(&y_, 1);
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_EQ(pp.options.linear_solver_type, DENSE_QR);
EXPECT_EQ(pp.linear_solver_options.type, DENSE_QR);
EXPECT_EQ(pp.evaluator_options.linear_solver_type, DENSE_QR);
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
}
TEST_F(LinearSolverAndEvaluatorCreationTest,
SchurTypeSolverWithEmptySecondEliminationGroup) {
problem_.SetParameterBlockConstant(&y_);
Solver::Options options;
options.linear_solver_type = DENSE_SCHUR;
options.linear_solver_ordering = std::make_shared<ParameterBlockOrdering>();
options.linear_solver_ordering->AddElementToGroup(&x_, 0);
options.linear_solver_ordering->AddElementToGroup(&z_, 0);
options.linear_solver_ordering->AddElementToGroup(&y_, 1);
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_EQ(pp.options.linear_solver_type, DENSE_SCHUR);
EXPECT_EQ(pp.linear_solver_options.type, DENSE_SCHUR);
EXPECT_EQ(pp.evaluator_options.linear_solver_type, DENSE_SCHUR);
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
}
TEST(TrustRegionPreprocessorTest, InnerIterationsWithOneParameterBlock) {
ProblemImpl problem;
double x = 1.0;
problem.AddResidualBlock(new DummyCostFunction<1, 1>, nullptr, &x);
Solver::Options options;
options.use_inner_iterations = true;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem, &pp));
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
EXPECT_TRUE(pp.inner_iteration_minimizer.get() == nullptr);
}
TEST_F(LinearSolverAndEvaluatorCreationTest,
InnerIterationsWithTwoParameterBlocks) {
Solver::Options options;
options.use_inner_iterations = true;
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
EXPECT_TRUE(pp.inner_iteration_minimizer.get() != nullptr);
}
TEST_F(LinearSolverAndEvaluatorCreationTest, InvalidInnerIterationsOrdering) {
Solver::Options options;
options.use_inner_iterations = true;
options.inner_iteration_ordering = std::make_shared<ParameterBlockOrdering>();
options.inner_iteration_ordering->AddElementToGroup(&x_, 0);
options.inner_iteration_ordering->AddElementToGroup(&z_, 0);
options.inner_iteration_ordering->AddElementToGroup(&y_, 0);
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_FALSE(preprocessor.Preprocess(options, &problem_, &pp));
}
TEST_F(LinearSolverAndEvaluatorCreationTest, ValidInnerIterationsOrdering) {
Solver::Options options;
options.use_inner_iterations = true;
options.inner_iteration_ordering = std::make_shared<ParameterBlockOrdering>();
options.inner_iteration_ordering->AddElementToGroup(&x_, 0);
options.inner_iteration_ordering->AddElementToGroup(&z_, 0);
options.inner_iteration_ordering->AddElementToGroup(&y_, 1);
TrustRegionPreprocessor preprocessor;
PreprocessedProblem pp;
EXPECT_TRUE(preprocessor.Preprocess(options, &problem_, &pp));
EXPECT_TRUE(pp.linear_solver.get() != nullptr);
EXPECT_TRUE(pp.evaluator.get() != nullptr);
EXPECT_TRUE(pp.inner_iteration_minimizer.get() != nullptr);
}
} // namespace internal
} // namespace ceres