1 // Ceres Solver - A fast non-linear least squares minimizer 2 // Copyright 2015 Google Inc. All rights reserved. 3 // http://ceres-solver.org/ 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: sameeragarwal@google.com (Sameer Agarwal) 30 31 #ifndef CERES_INTERNAL_NUMERIC_DIFF_TEST_UTILS_H_ 32 #define CERES_INTERNAL_NUMERIC_DIFF_TEST_UTILS_H_ 33 34 #include "ceres/cost_function.h" 35 #include "ceres/internal/port.h" 36 #include "ceres/sized_cost_function.h" 37 #include "ceres/types.h" 38 39 namespace ceres { 40 namespace internal { 41 42 // Noise factor for randomized cost function. 43 static constexpr double kNoiseFactor = 0.01; 44 45 // Default random seed for randomized cost function. 46 static constexpr unsigned int kRandomSeed = 1234; 47 48 // y1 = x1'x2 -> dy1/dx1 = x2, dy1/dx2 = x1 49 // y2 = (x1'x2)^2 -> dy2/dx1 = 2 * x2 * (x1'x2), dy2/dx2 = 2 * x1 * (x1'x2) 50 // y3 = x2'x2 -> dy3/dx1 = 0, dy3/dx2 = 2 * x2 51 class CERES_EXPORT_INTERNAL EasyFunctor { 52 public: 53 bool operator()(const double* x1, const double* x2, double* residuals) const; 54 void ExpectCostFunctionEvaluationIsNearlyCorrect( 55 const CostFunction& cost_function, NumericDiffMethodType method) const; 56 }; 57 58 class EasyCostFunction : public SizedCostFunction<3, 5, 5> { 59 public: Evaluate(double const * const * parameters,double * residuals,double **)60 bool Evaluate(double const* const* parameters, 61 double* residuals, 62 double** /* not used */) const final { 63 return functor_(parameters[0], parameters[1], residuals); 64 } 65 66 private: 67 EasyFunctor functor_; 68 }; 69 70 // y1 = sin(x1'x2) 71 // y2 = exp(-x1'x2 / 10) 72 // 73 // dy1/dx1 = x2 * cos(x1'x2), dy1/dx2 = x1 * cos(x1'x2) 74 // dy2/dx1 = -x2 * exp(-x1'x2 / 10) / 10, dy2/dx2 = -x2 * exp(-x1'x2 / 10) / 10 75 class CERES_EXPORT TranscendentalFunctor { 76 public: 77 bool operator()(const double* x1, const double* x2, double* residuals) const; 78 void ExpectCostFunctionEvaluationIsNearlyCorrect( 79 const CostFunction& cost_function, NumericDiffMethodType method) const; 80 }; 81 82 class CERES_EXPORT_INTERNAL TranscendentalCostFunction 83 : public SizedCostFunction<2, 5, 5> { 84 public: Evaluate(double const * const * parameters,double * residuals,double **)85 bool Evaluate(double const* const* parameters, 86 double* residuals, 87 double** /* not used */) const final { 88 return functor_(parameters[0], parameters[1], residuals); 89 } 90 91 private: 92 TranscendentalFunctor functor_; 93 }; 94 95 // y = exp(x), dy/dx = exp(x) 96 class CERES_EXPORT_INTERNAL ExponentialFunctor { 97 public: 98 bool operator()(const double* x1, double* residuals) const; 99 void ExpectCostFunctionEvaluationIsNearlyCorrect( 100 const CostFunction& cost_function) const; 101 }; 102 103 class ExponentialCostFunction : public SizedCostFunction<1, 1> { 104 public: Evaluate(double const * const * parameters,double * residuals,double **)105 bool Evaluate(double const* const* parameters, 106 double* residuals, 107 double** /* not used */) const final { 108 return functor_(parameters[0], residuals); 109 } 110 111 private: 112 ExponentialFunctor functor_; 113 }; 114 115 // Test adaptive numeric differentiation by synthetically adding random noise 116 // to a functor. 117 // y = x^2 + [random noise], dy/dx ~ 2x 118 class CERES_EXPORT_INTERNAL RandomizedFunctor { 119 public: RandomizedFunctor(double noise_factor,unsigned int random_seed)120 RandomizedFunctor(double noise_factor, unsigned int random_seed) 121 : noise_factor_(noise_factor), random_seed_(random_seed) {} 122 123 bool operator()(const double* x1, double* residuals) const; 124 void ExpectCostFunctionEvaluationIsNearlyCorrect( 125 const CostFunction& cost_function) const; 126 127 private: 128 double noise_factor_; 129 unsigned int random_seed_; 130 }; 131 132 class CERES_EXPORT_INTERNAL RandomizedCostFunction 133 : public SizedCostFunction<1, 1> { 134 public: RandomizedCostFunction(double noise_factor,unsigned int random_seed)135 RandomizedCostFunction(double noise_factor, unsigned int random_seed) 136 : functor_(noise_factor, random_seed) {} 137 Evaluate(double const * const * parameters,double * residuals,double **)138 bool Evaluate(double const* const* parameters, 139 double* residuals, 140 double** /* not used */) const final { 141 return functor_(parameters[0], residuals); 142 } 143 144 private: 145 RandomizedFunctor functor_; 146 }; 147 148 } // namespace internal 149 } // namespace ceres 150 151 #endif // CERES_INTERNAL_NUMERIC_DIFF_TEST_UTILS_H_ 152