1// Ceres Solver - A fast non-linear least squares minimizer 2// Copyright 2010, 2011, 2012 Google Inc. All rights reserved. 3// http://code.google.com/p/ceres-solver/ 4// 5// Redistribution and use in source and binary forms, with or without 6// modification, are permitted provided that the following conditions are met: 7// 8// * Redistributions of source code must retain the above copyright notice, 9// this list of conditions and the following disclaimer. 10// * Redistributions in binary form must reproduce the above copyright notice, 11// this list of conditions and the following disclaimer in the documentation 12// and/or other materials provided with the distribution. 13// * Neither the name of Google Inc. nor the names of its contributors may be 14// used to endorse or promote products derived from this software without 15// specific prior written permission. 16// 17// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 18// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE 21// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 22// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 23// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 24// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 25// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 26// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 27// POSSIBILITY OF SUCH DAMAGE. 28// 29// Author: keir@google.com (Keir Mierle) 30 31#ifndef CERES_INTERNAL_GRADIENT_CHECKING_COST_FUNCTION_H_ 32#define CERES_INTERNAL_GRADIENT_CHECKING_COST_FUNCTION_H_ 33 34#include <string> 35 36#include "ceres/cost_function.h" 37 38namespace ceres { 39namespace internal { 40 41class ProblemImpl; 42 43// Creates a CostFunction that checks the jacobians that cost_function computes 44// with finite differences. Bad results are logged; required precision is 45// controlled by relative_precision and the numeric differentiation step size is 46// controlled with relative_step_size. See solver.h for a better explanation of 47// relative_step_size. Caller owns result. 48// 49// The condition enforced is that 50// 51// (J_actual(i, j) - J_numeric(i, j)) 52// ------------------------------------ < relative_precision 53// max(J_actual(i, j), J_numeric(i, j)) 54// 55// where J_actual(i, j) is the jacobian as computed by the supplied cost 56// function (by the user) and J_numeric is the jacobian as computed by finite 57// differences. 58// 59// Note: This is quite inefficient and is intended only for debugging. 60CostFunction* CreateGradientCheckingCostFunction( 61 const CostFunction* cost_function, 62 double relative_step_size, 63 double relative_precision, 64 const string& extra_info); 65 66// Create a new ProblemImpl object from the input problem_impl, where 67// each CostFunctions in problem_impl are wrapped inside a 68// GradientCheckingCostFunctions. This gives us a ProblemImpl object 69// which checks its derivatives against estimates from numeric 70// differentiation everytime a ResidualBlock is evaluated. 71// 72// relative_step_size and relative_precision are parameters to control 73// the numeric differentiation and the relative tolerance between the 74// jacobian computed by the CostFunctions in problem_impl and 75// jacobians obtained by numerically differentiating them. For more 76// details see the documentation for 77// CreateGradientCheckingCostFunction above. 78ProblemImpl* CreateGradientCheckingProblemImpl(ProblemImpl* problem_impl, 79 double relative_step_size, 80 double relative_precision); 81 82} // namespace internal 83} // namespace ceres 84 85#endif // CERES_INTERNAL_GRADIENT_CHECKING_COST_FUNCTION_H_ 86