ceres-solver/internal/ceres/residual_block.cc

// 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)
//         sameeragarwal@google.com (Sameer Agarwal)

#include "ceres/residual_block.h"

#include <algorithm>
#include <cstddef>
#include <vector>

#include "ceres/blas.h"
#include "ceres/corrector.h"
#include "ceres/parameter_block.h"
#include "ceres/residual_block_utils.h"
#include "ceres/cost_function.h"
#include "ceres/internal/eigen.h"
#include "ceres/internal/fixed_array.h"
#include "ceres/local_parameterization.h"
#include "ceres/loss_function.h"

namespace ceres {
namespace internal {

ResidualBlock::ResidualBlock(const CostFunction* cost_function,
                             const LossFunction* loss_function,
                             const vector<ParameterBlock*>& parameter_blocks,
                             int index)
    : cost_function_(cost_function),
      loss_function_(loss_function),
      parameter_blocks_(
          new ParameterBlock* [
              cost_function->parameter_block_sizes().size()]),
      index_(index) {
  std::copy(parameter_blocks.begin(),
            parameter_blocks.end(),
            parameter_blocks_.get());
}

bool ResidualBlock::Evaluate(const bool apply_loss_function,
                             double* cost,
                             double* residuals,
                             double** jacobians,
                             double* scratch) const {
  const int num_parameter_blocks = NumParameterBlocks();
  const int num_residuals = cost_function_->num_residuals();

  // Collect the parameters from their blocks. This will rarely allocate, since
  // residuals taking more than 8 parameter block arguments are rare.
  FixedArray<const double*, 8> parameters(num_parameter_blocks);
  for (int i = 0; i < num_parameter_blocks; ++i) {
    parameters[i] = parameter_blocks_[i]->state();
  }

  // Put pointers into the scratch space into global_jacobians as appropriate.
  FixedArray<double*, 8> global_jacobians(num_parameter_blocks);
  if (jacobians != NULL) {
    for (int i = 0; i < num_parameter_blocks; ++i) {
      const ParameterBlock* parameter_block = parameter_blocks_[i];
      if (jacobians[i] != NULL &&
          parameter_block->LocalParameterizationJacobian() != NULL) {
        global_jacobians[i] = scratch;
        scratch += num_residuals * parameter_block->Size();
      } else {
        global_jacobians[i] = jacobians[i];
      }
    }
  }

  // If the caller didn't request residuals, use the scratch space for them.
  bool outputting_residuals = (residuals != NULL);
  if (!outputting_residuals) {
    residuals = scratch;
  }

  // Invalidate the evaluation buffers so that we can check them after
  // the CostFunction::Evaluate call, to see if all the return values
  // that were required were written to and that they are finite.
  double** eval_jacobians = (jacobians != NULL) ? global_jacobians.get() : NULL;

  InvalidateEvaluation(*this, cost, residuals, eval_jacobians);

  if (!cost_function_->Evaluate(parameters.get(), residuals, eval_jacobians)) {
    return false;
  }

  if (!IsEvaluationValid(*this,
                         parameters.get(),
                         cost,
                         residuals,
                         eval_jacobians)) {
    string message =
        "\n\n"
        "Error in evaluating the ResidualBlock.\n\n"
        "There are two possible reasons. Either the CostFunction did not evaluate and fill all    \n"  // NOLINT
        "residual and jacobians that were requested or there was a non-finite value (nan/infinite)\n"  // NOLINT
        "generated during the or jacobian computation. \n\n" +
        EvaluationToString(*this,
                           parameters.get(),
                           cost,
                           residuals,
                           eval_jacobians);
    LOG(WARNING) << message;
    return false;
  }

  double squared_norm = VectorRef(residuals, num_residuals).squaredNorm();

  // Update the jacobians with the local parameterizations.
  if (jacobians != NULL) {
    for (int i = 0; i < num_parameter_blocks; ++i) {
      if (jacobians[i] != NULL) {
        const ParameterBlock* parameter_block = parameter_blocks_[i];

        // Apply local reparameterization to the jacobians.
        if (parameter_block->LocalParameterizationJacobian() != NULL) {
          MatrixMatrixMultiply
              <Eigen::Dynamic, Eigen::Dynamic, Eigen::Dynamic, Eigen::Dynamic, 0>(
              global_jacobians[i],
              num_residuals,
              parameter_block->Size(),
              parameter_block->LocalParameterizationJacobian(),
              parameter_block->Size(),
              parameter_block->LocalSize(),
              jacobians[i], 0, 0,  num_residuals, parameter_block->LocalSize());
        }
      }
    }
  }

  if (loss_function_ == NULL || !apply_loss_function) {
    *cost = 0.5 * squared_norm;
    return true;
  }

  double rho[3];
  loss_function_->Evaluate(squared_norm, rho);
  *cost = 0.5 * rho[0];

  // No jacobians and not outputting residuals? All done. Doing an early exit
  // here avoids constructing the "Corrector" object below in a common case.
  if (jacobians == NULL && !outputting_residuals) {
    return true;
  }

  // Correct for the effects of the loss function. The jacobians need to be
  // corrected before the residuals, since they use the uncorrected residuals.
  Corrector correct(squared_norm, rho);
  if (jacobians != NULL) {
    for (int i = 0; i < num_parameter_blocks; ++i) {
      if (jacobians[i] != NULL) {
        const ParameterBlock* parameter_block = parameter_blocks_[i];

        // Correct the jacobians for the loss function.
        correct.CorrectJacobian(num_residuals,
                                parameter_block->LocalSize(),
                                residuals,
                                jacobians[i]);
      }
    }
  }

  // Correct the residuals with the loss function.
  if (outputting_residuals) {
    correct.CorrectResiduals(num_residuals, residuals);
  }
  return true;
}

int ResidualBlock::NumScratchDoublesForEvaluate() const {
  // Compute the amount of scratch space needed to store the full-sized
  // jacobians. For parameters that have no local parameterization  no storage
  // is needed and the passed-in jacobian array is used directly. Also include
  // space to store the residuals, which is needed for cost-only evaluations.
  // This is slightly pessimistic, since both won't be needed all the time, but
  // the amount of excess should not cause problems for the caller.
  int num_parameters = NumParameterBlocks();
  int scratch_doubles = 1;
  for (int i = 0; i < num_parameters; ++i) {
    const ParameterBlock* parameter_block = parameter_blocks_[i];
    if (!parameter_block->IsConstant() &&
        parameter_block->LocalParameterizationJacobian() != NULL) {
      scratch_doubles += parameter_block->Size();
    }
  }
  scratch_doubles *= NumResiduals();
  return scratch_doubles;
}

}  // namespace internal
}  // namespace ceres