// Ceres Solver - A fast non-linear least squares minimizer
// Copyright 2010, 2011, 2012 Google Inc. All rights reserved.
// http://code.google.com/p/ceres-solver/
//
// 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: keir@google.com (Keir Mierle)
//
// Tests shared across evaluators. The tests try all combinations of linear
// solver and num_eliminate_blocks (for schur-based solvers).
#include "ceres/evaluator.h"
#include "ceres/casts.h"
#include "ceres/cost_function.h"
#include "ceres/crs_matrix.h"
#include "ceres/evaluator_test_utils.h"
#include "ceres/internal/eigen.h"
#include "ceres/internal/scoped_ptr.h"
#include "ceres/local_parameterization.h"
#include "ceres/problem_impl.h"
#include "ceres/program.h"
#include "ceres/sized_cost_function.h"
#include "ceres/sparse_matrix.h"
#include "ceres/stringprintf.h"
#include "ceres/types.h"
#include "gtest/gtest.h"
namespace ceres {
namespace internal {
// TODO(keir): Consider pushing this into a common test utils file.
template<int kFactor, int kNumResiduals,
int N0 = 0, int N1 = 0, int N2 = 0, bool kSucceeds = true>
class ParameterIgnoringCostFunction
: public SizedCostFunction<kNumResiduals, N0, N1, N2> {
typedef SizedCostFunction<kNumResiduals, N0, N1, N2> Base;
public:
virtual bool Evaluate(double const* const* parameters,
double* residuals,
double** jacobians) const {
for (int i = 0; i < Base::num_residuals(); ++i) {
residuals[i] = i + 1;
}
if (jacobians) {
for (int k = 0; k < Base::parameter_block_sizes().size(); ++k) {
// The jacobians here are full sized, but they are transformed in the
// evaluator into the "local" jacobian. In the tests, the "subset
// constant" parameterization is used, which should pick out columns
// from these jacobians. Put values in the jacobian that make this
// obvious; in particular, make the jacobians like this:
//
// 1 2 3 4 ...
// 1 2 3 4 ... .* kFactor
// 1 2 3 4 ...
//
// where the multiplication by kFactor makes it easier to distinguish
// between Jacobians of different residuals for the same parameter.
if (jacobians[k] != NULL) {
MatrixRef jacobian(jacobians[k],
Base::num_residuals(),
Base::parameter_block_sizes()[k]);
for (int j = 0; j < Base::parameter_block_sizes()[k]; ++j) {
jacobian.col(j).setConstant(kFactor * (j + 1));
}
}
}
}
return kSucceeds;
}
};
struct EvaluatorTestOptions {
EvaluatorTestOptions(LinearSolverType linear_solver_type,
int num_eliminate_blocks,
bool dynamic_sparsity = false)
: linear_solver_type(linear_solver_type),
num_eliminate_blocks(num_eliminate_blocks),
dynamic_sparsity(dynamic_sparsity) {}
LinearSolverType linear_solver_type;
int num_eliminate_blocks;
bool dynamic_sparsity;
};
struct EvaluatorTest
: public ::testing::TestWithParam<EvaluatorTestOptions> {
Evaluator* CreateEvaluator(Program* program) {
// This program is straight from the ProblemImpl, and so has no index/offset
// yet; compute it here as required by the evalutor implementations.
program->SetParameterOffsetsAndIndex();
if (VLOG_IS_ON(1)) {
string report;
StringAppendF(&report, "Creating evaluator with type: %d",
GetParam().linear_solver_type);
if (GetParam().linear_solver_type == SPARSE_NORMAL_CHOLESKY) {
StringAppendF(&report, ", dynamic_sparsity: %d",
GetParam().dynamic_sparsity);
}
StringAppendF(&report, " and num_eliminate_blocks: %d",
GetParam().num_eliminate_blocks);
VLOG(1) << report;
}
Evaluator::Options options;
options.linear_solver_type = GetParam().linear_solver_type;
options.num_eliminate_blocks = GetParam().num_eliminate_blocks;
options.dynamic_sparsity = GetParam().dynamic_sparsity;
string error;
return Evaluator::Create(options, program, &error);
}
void EvaluateAndCompare(ProblemImpl *problem,
int expected_num_rows,
int expected_num_cols,
double expected_cost,
const double* expected_residuals,
const double* expected_gradient,
const double* expected_jacobian) {
scoped_ptr<Evaluator> evaluator(
CreateEvaluator(problem->mutable_program()));
int num_residuals = expected_num_rows;
int num_parameters = expected_num_cols;
double cost = -1;
Vector residuals(num_residuals);
residuals.setConstant(-2000);
Vector gradient(num_parameters);
gradient.setConstant(-3000);
scoped_ptr<SparseMatrix> jacobian(evaluator->CreateJacobian());
ASSERT_EQ(expected_num_rows, evaluator->NumResiduals());
ASSERT_EQ(expected_num_cols, evaluator->NumEffectiveParameters());
ASSERT_EQ(expected_num_rows, jacobian->num_rows());
ASSERT_EQ(expected_num_cols, jacobian->num_cols());
vector<double> state(evaluator->NumParameters());
ASSERT_TRUE(evaluator->Evaluate(
&state[0],
&cost,
expected_residuals != NULL ? &residuals[0] : NULL,
expected_gradient != NULL ? &gradient[0] : NULL,
expected_jacobian != NULL ? jacobian.get() : NULL));
Matrix actual_jacobian;
if (expected_jacobian != NULL) {
jacobian->ToDenseMatrix(&actual_jacobian);
}
CompareEvaluations(expected_num_rows,
expected_num_cols,
expected_cost,
expected_residuals,
expected_gradient,
expected_jacobian,
cost,
&residuals[0],
&gradient[0],
actual_jacobian.data());
}
// Try all combinations of parameters for the evaluator.
void CheckAllEvaluationCombinations(const ExpectedEvaluation &expected) {
for (int i = 0; i < 8; ++i) {
EvaluateAndCompare(&problem,
expected.num_rows,
expected.num_cols,
expected.cost,
(i & 1) ? expected.residuals : NULL,
(i & 2) ? expected.gradient : NULL,
(i & 4) ? expected.jacobian : NULL);
}
}
// The values are ignored completely by the cost function.
double x[2];
double y[3];
double z[4];
ProblemImpl problem;
};
void SetSparseMatrixConstant(SparseMatrix* sparse_matrix, double value) {
VectorRef(sparse_matrix->mutable_values(),
sparse_matrix->num_nonzeros()).setConstant(value);
}
TEST_P(EvaluatorTest, SingleResidualProblem) {
problem.AddResidualBlock(new ParameterIgnoringCostFunction<1, 3, 2, 3, 4>,
NULL,
x, y, z);
ExpectedEvaluation expected = {
// Rows/columns
3, 9,
// Cost
7.0,
// Residuals
{ 1.0, 2.0, 3.0 },
// Gradient
{ 6.0, 12.0, // x
6.0, 12.0, 18.0, // y
6.0, 12.0, 18.0, 24.0, // z
},
// Jacobian
// x y z
{ 1, 2, 1, 2, 3, 1, 2, 3, 4,
1, 2, 1, 2, 3, 1, 2, 3, 4,
1, 2, 1, 2, 3, 1, 2, 3, 4
}
};
CheckAllEvaluationCombinations(expected);
}
TEST_P(EvaluatorTest, SingleResidualProblemWithPermutedParameters) {
// Add the parameters in explicit order to force the ordering in the program.
problem.AddParameterBlock(x, 2);
problem.AddParameterBlock(y, 3);
problem.AddParameterBlock(z, 4);
// Then use a cost function which is similar to the others, but swap around
// the ordering of the parameters to the cost function. This shouldn't affect
// the jacobian evaluation, but requires explicit handling in the evaluators.
// At one point the compressed row evaluator had a bug that went undetected
// for a long time, since by chance most users added parameters to the problem
// in the same order that they occured as parameters to a cost function.
problem.AddResidualBlock(new ParameterIgnoringCostFunction<1, 3, 4, 3, 2>,
NULL,
z, y, x);
ExpectedEvaluation expected = {
// Rows/columns
3, 9,
// Cost
7.0,
// Residuals
{ 1.0, 2.0, 3.0 },
// Gradient
{ 6.0, 12.0, // x
6.0, 12.0, 18.0, // y
6.0, 12.0, 18.0, 24.0, // z
},
// Jacobian
// x y z
{ 1, 2, 1, 2, 3, 1, 2, 3, 4,
1, 2, 1, 2, 3, 1, 2, 3, 4,
1, 2, 1, 2, 3, 1, 2, 3, 4
}
};
CheckAllEvaluationCombinations(expected);
}
TEST_P(EvaluatorTest, SingleResidualProblemWithNuisanceParameters) {
// These parameters are not used.
double a[2];
double b[1];
double c[1];
double d[3];
// Add the parameters in a mixed order so the Jacobian is "checkered" with the
// values from the other parameters.
problem.AddParameterBlock(a, 2);
problem.AddParameterBlock(x, 2);
problem.AddParameterBlock(b, 1);
problem.AddParameterBlock(y, 3);
problem.AddParameterBlock(c, 1);
problem.AddParameterBlock(z, 4);
problem.AddParameterBlock(d, 3);
problem.AddResidualBlock(new ParameterIgnoringCostFunction<1, 3, 2, 3, 4>,
NULL,
x, y, z);
ExpectedEvaluation expected = {
// Rows/columns
3, 16,
// Cost
7.0,
// Residuals
{ 1.0, 2.0, 3.0 },
// Gradient
{ 0.0, 0.0, // a
6.0, 12.0, // x
0.0, // b
6.0, 12.0, 18.0, // y
0.0, // c
6.0, 12.0, 18.0, 24.0, // z
0.0, 0.0, 0.0, // d
},
// Jacobian
// a x b y c z d
{ 0, 0, 1, 2, 0, 1, 2, 3, 0, 1, 2, 3, 4, 0, 0, 0,
0, 0, 1, 2, 0, 1, 2, 3, 0, 1, 2, 3, 4, 0, 0, 0,
0, 0, 1, 2, 0, 1, 2, 3, 0, 1, 2, 3, 4, 0, 0, 0
}
};
CheckAllEvaluationCombinations(expected);
}
TEST_P(EvaluatorTest, MultipleResidualProblem) {
// Add the parameters in explicit order to force the ordering in the program.
problem.AddParameterBlock(x, 2);
problem.AddParameterBlock(y, 3);
problem.AddParameterBlock(z, 4);
// f(x, y) in R^2
problem.AddResidualBlock(new ParameterIgnoringCostFunction<1, 2, 2, 3>,
NULL,
x, y);
// g(x, z) in R^3
problem.AddResidualBlock(new ParameterIgnoringCostFunction<2, 3, 2, 4>,
NULL,
x, z);
// h(y, z) in R^4
problem.AddResidualBlock(new ParameterIgnoringCostFunction<3, 4, 3, 4>,
NULL,
y, z);
ExpectedEvaluation expected = {
// Rows/columns
9, 9,
// Cost
// f g h
( 1 + 4 + 1 + 4 + 9 + 1 + 4 + 9 + 16) / 2.0,
// Residuals
{ 1.0, 2.0, // f
1.0, 2.0, 3.0, // g
1.0, 2.0, 3.0, 4.0 // h
},
// Gradient
{ 15.0, 30.0, // x
33.0, 66.0, 99.0, // y
42.0, 84.0, 126.0, 168.0 // z
},
// Jacobian
// x y z
{ /* f(x, y) */ 1, 2, 1, 2, 3, 0, 0, 0, 0,
1, 2, 1, 2, 3, 0, 0, 0, 0,
/* g(x, z) */ 2, 4, 0, 0, 0, 2, 4, 6, 8,
2, 4, 0, 0, 0, 2, 4, 6, 8,
2, 4, 0, 0, 0, 2, 4, 6, 8,
/* h(y, z) */ 0, 0, 3, 6, 9, 3, 6, 9, 12,
0, 0, 3, 6, 9, 3, 6, 9, 12,
0, 0, 3, 6, 9, 3, 6, 9, 12,
0, 0, 3, 6, 9, 3, 6, 9, 12
}
};
CheckAllEvaluationCombinations(expected);
}
TEST_P(EvaluatorTest, MultipleResidualsWithLocalParameterizations) {
// Add the parameters in explicit order to force the ordering in the program.
problem.AddParameterBlock(x, 2);
// Fix y's first dimension.
vector<int> y_fixed;
y_fixed.push_back(0);
problem.AddParameterBlock(y, 3, new SubsetParameterization(3, y_fixed));
// Fix z's second dimension.
vector<int> z_fixed;
z_fixed.push_back(1);
problem.AddParameterBlock(z, 4, new SubsetParameterization(4, z_fixed));
// f(x, y) in R^2
problem.AddResidualBlock(new ParameterIgnoringCostFunction<1, 2, 2, 3>,
NULL,
x, y);
// g(x, z) in R^3
problem.AddResidualBlock(new ParameterIgnoringCostFunction<2, 3, 2, 4>,
NULL,
x, z);
// h(y, z) in R^4
problem.AddResidualBlock(new ParameterIgnoringCostFunction<3, 4, 3, 4>,
NULL,
y, z);
ExpectedEvaluation expected = {
// Rows/columns
9, 7,
// Cost
// f g h
( 1 + 4 + 1 + 4 + 9 + 1 + 4 + 9 + 16) / 2.0,
// Residuals
{ 1.0, 2.0, // f
1.0, 2.0, 3.0, // g
1.0, 2.0, 3.0, 4.0 // h
},
// Gradient
{ 15.0, 30.0, // x
66.0, 99.0, // y
42.0, 126.0, 168.0 // z
},
// Jacobian
// x y z
{ /* f(x, y) */ 1, 2, 2, 3, 0, 0, 0,
1, 2, 2, 3, 0, 0, 0,
/* g(x, z) */ 2, 4, 0, 0, 2, 6, 8,
2, 4, 0, 0, 2, 6, 8,
2, 4, 0, 0, 2, 6, 8,
/* h(y, z) */ 0, 0, 6, 9, 3, 9, 12,
0, 0, 6, 9, 3, 9, 12,
0, 0, 6, 9, 3, 9, 12,
0, 0, 6, 9, 3, 9, 12
}
};
CheckAllEvaluationCombinations(expected);
}
TEST_P(EvaluatorTest, MultipleResidualProblemWithSomeConstantParameters) {
// The values are ignored completely by the cost function.
double x[2];
double y[3];
double z[4];
// Add the parameters in explicit order to force the ordering in the program.
problem.AddParameterBlock(x, 2);
problem.AddParameterBlock(y, 3);
problem.AddParameterBlock(z, 4);
// f(x, y) in R^2
problem.AddResidualBlock(new ParameterIgnoringCostFunction<1, 2, 2, 3>,
NULL,
x, y);
// g(x, z) in R^3
problem.AddResidualBlock(new ParameterIgnoringCostFunction<2, 3, 2, 4>,
NULL,
x, z);
// h(y, z) in R^4
problem.AddResidualBlock(new ParameterIgnoringCostFunction<3, 4, 3, 4>,
NULL,
y, z);
// For this test, "z" is constant.
problem.SetParameterBlockConstant(z);
// Create the reduced program which is missing the fixed "z" variable.
// Normally, the preprocessing of the program that happens in solver_impl
// takes care of this, but we don't want to invoke the solver here.
Program reduced_program;
vector<ParameterBlock*>* parameter_blocks =
problem.mutable_program()->mutable_parameter_blocks();
// "z" is the last parameter; save it for later and pop it off temporarily.
// Note that "z" will still get read during evaluation, so it cannot be
// deleted at this point.
ParameterBlock* parameter_block_z = parameter_blocks->back();
parameter_blocks->pop_back();
ExpectedEvaluation expected = {
// Rows/columns
9, 5,
// Cost
// f g h
( 1 + 4 + 1 + 4 + 9 + 1 + 4 + 9 + 16) / 2.0,
// Residuals
{ 1.0, 2.0, // f
1.0, 2.0, 3.0, // g
1.0, 2.0, 3.0, 4.0 // h
},
// Gradient
{ 15.0, 30.0, // x
33.0, 66.0, 99.0, // y
},
// Jacobian
// x y
{ /* f(x, y) */ 1, 2, 1, 2, 3,
1, 2, 1, 2, 3,
/* g(x, z) */ 2, 4, 0, 0, 0,
2, 4, 0, 0, 0,
2, 4, 0, 0, 0,
/* h(y, z) */ 0, 0, 3, 6, 9,
0, 0, 3, 6, 9,
0, 0, 3, 6, 9,
0, 0, 3, 6, 9
}
};
CheckAllEvaluationCombinations(expected);
// Restore parameter block z, so it will get freed in a consistent way.
parameter_blocks->push_back(parameter_block_z);
}
TEST_P(EvaluatorTest, EvaluatorAbortsForResidualsThatFailToEvaluate) {
// Switch the return value to failure.
problem.AddResidualBlock(
new ParameterIgnoringCostFunction<20, 3, 2, 3, 4, false>, NULL, x, y, z);
// The values are ignored.
double state[9];
scoped_ptr<Evaluator> evaluator(CreateEvaluator(problem.mutable_program()));
scoped_ptr<SparseMatrix> jacobian(evaluator->CreateJacobian());
double cost;
EXPECT_FALSE(evaluator->Evaluate(state, &cost, NULL, NULL, NULL));
}
// In the pairs, the first argument is the linear solver type, and the second
// argument is num_eliminate_blocks. Changing the num_eliminate_blocks only
// makes sense for the schur-based solvers.
//
// Try all values of num_eliminate_blocks that make sense given that in the
// tests a maximum of 4 parameter blocks are present.
INSTANTIATE_TEST_CASE_P(
LinearSolvers,
EvaluatorTest,
::testing::Values(
EvaluatorTestOptions(DENSE_QR, 0),
EvaluatorTestOptions(DENSE_SCHUR, 0),
EvaluatorTestOptions(DENSE_SCHUR, 1),
EvaluatorTestOptions(DENSE_SCHUR, 2),
EvaluatorTestOptions(DENSE_SCHUR, 3),
EvaluatorTestOptions(DENSE_SCHUR, 4),
EvaluatorTestOptions(SPARSE_SCHUR, 0),
EvaluatorTestOptions(SPARSE_SCHUR, 1),
EvaluatorTestOptions(SPARSE_SCHUR, 2),
EvaluatorTestOptions(SPARSE_SCHUR, 3),
EvaluatorTestOptions(SPARSE_SCHUR, 4),
EvaluatorTestOptions(ITERATIVE_SCHUR, 0),
EvaluatorTestOptions(ITERATIVE_SCHUR, 1),
EvaluatorTestOptions(ITERATIVE_SCHUR, 2),
EvaluatorTestOptions(ITERATIVE_SCHUR, 3),
EvaluatorTestOptions(ITERATIVE_SCHUR, 4),
EvaluatorTestOptions(SPARSE_NORMAL_CHOLESKY, 0, false),
EvaluatorTestOptions(SPARSE_NORMAL_CHOLESKY, 0, true)));
// Simple cost function used to check if the evaluator is sensitive to
// state changes.
class ParameterSensitiveCostFunction : public SizedCostFunction<2, 2> {
public:
virtual bool Evaluate(double const* const* parameters,
double* residuals,
double** jacobians) const {
double x1 = parameters[0][0];
double x2 = parameters[0][1];
residuals[0] = x1 * x1;
residuals[1] = x2 * x2;
if (jacobians != NULL) {
double* jacobian = jacobians[0];
if (jacobian != NULL) {
jacobian[0] = 2.0 * x1;
jacobian[1] = 0.0;
jacobian[2] = 0.0;
jacobian[3] = 2.0 * x2;
}
}
return true;
}
};
TEST(Evaluator, EvaluatorRespectsParameterChanges) {
ProblemImpl problem;
double x[2];
x[0] = 1.0;
x[1] = 1.0;
problem.AddResidualBlock(new ParameterSensitiveCostFunction(), NULL, x);
Program* program = problem.mutable_program();
program->SetParameterOffsetsAndIndex();
Evaluator::Options options;
options.linear_solver_type = DENSE_QR;
options.num_eliminate_blocks = 0;
string error;
scoped_ptr<Evaluator> evaluator(Evaluator::Create(options, program, &error));
scoped_ptr<SparseMatrix> jacobian(evaluator->CreateJacobian());
ASSERT_EQ(2, jacobian->num_rows());
ASSERT_EQ(2, jacobian->num_cols());
double state[2];
state[0] = 2.0;
state[1] = 3.0;
// The original state of a residual block comes from the user's
// state. So the original state is 1.0, 1.0, and the only way we get
// the 2.0, 3.0 results in the following tests is if it respects the
// values in the state vector.
// Cost only; no residuals and no jacobian.
{
double cost = -1;
ASSERT_TRUE(evaluator->Evaluate(state, &cost, NULL, NULL, NULL));
EXPECT_EQ(48.5, cost);
}
// Cost and residuals, no jacobian.
{
double cost = -1;
double residuals[2] = { -2, -2 };
ASSERT_TRUE(evaluator->Evaluate(state, &cost, residuals, NULL, NULL));
EXPECT_EQ(48.5, cost);
EXPECT_EQ(4, residuals[0]);
EXPECT_EQ(9, residuals[1]);
}
// Cost, residuals, and jacobian.
{
double cost = -1;
double residuals[2] = { -2, -2};
SetSparseMatrixConstant(jacobian.get(), -1);
ASSERT_TRUE(evaluator->Evaluate(state,
&cost,
residuals,
NULL,
jacobian.get()));
EXPECT_EQ(48.5, cost);
EXPECT_EQ(4, residuals[0]);
EXPECT_EQ(9, residuals[1]);
Matrix actual_jacobian;
jacobian->ToDenseMatrix(&actual_jacobian);
Matrix expected_jacobian(2, 2);
expected_jacobian
<< 2 * state[0], 0,
0, 2 * state[1];
EXPECT_TRUE((actual_jacobian.array() == expected_jacobian.array()).all())
<< "Actual:\n" << actual_jacobian
<< "\nExpected:\n" << expected_jacobian;
}
}
} // namespace internal
} // namespace ceres