Merge pull request #145 from resibots/bench_dev

Changes in benchmarks

src/benchmarks/bayesopt/testfunctions.hpp

@@ -258,11 +258,13 @@ struct SixHumpCamel {
     static constexpr size_t dim_out = 1;
     double operator()(const vectord& x) const
     {
-        double x1_2 = x(0) * x(0);
-        double x2_2 = x(1) * x(1);
+        double x1 = -3 + 6 * x(0);
+        double x2 = -2 + 4 * x(1);
+        double x1_2 = x1 * x1;
+        double x2_2 = x2 * x2;
 
         double tmp1 = (4 - 2.1 * x1_2 + (x1_2 * x1_2) / 3) * x1_2;
-        double tmp2 = x(0) * x(1);
+        double tmp2 = x1 * x2;
         double tmp3 = (-4 + 4 * x2_2) * x2_2;
         return tmp1 + tmp2 + tmp3;
     }
@@ -272,6 +274,10 @@ struct SixHumpCamel {
         matrixd sols(2, 2);
         sols <<= 0.0898, -0.7126,
             -0.0898, 0.7126;
+        sols(0, 0) = (sols(0, 0) + 3.0) / 6.0;
+        sols(1, 0) = (sols(1, 0) + 3.0) / 6.0;
+        sols(0, 1) = (sols(0, 1) + 2.0) / 4.0;
+        sols(1, 1) = (sols(1, 1) + 2.0) / 4.0;
         return sols;
     }
 };

src/benchmarks/limbo/testfunctions.hpp

@@ -230,11 +230,13 @@ struct SixHumpCamel {
     static constexpr size_t dim_out = 1;
     double operator()(const Eigen::VectorXd& x) const
     {
-        double x1_2 = x(0) * x(0);
-        double x2_2 = x(1) * x(1);
+        double x1 = -3 + 6 * x(0);
+        double x2 = -2 + 4 * x(1);
+        double x1_2 = x1 * x1;
+        double x2_2 = x2 * x2;
 
         double tmp1 = (4 - 2.1 * x1_2 + (x1_2 * x1_2) / 3) * x1_2;
-        double tmp2 = x(0) * x(1);
+        double tmp2 = x1 * x2;
         double tmp3 = (-4 + 4 * x2_2) * x2_2;
         return tmp1 + tmp2 + tmp3;
     }
@@ -244,6 +246,10 @@ struct SixHumpCamel {
         Eigen::MatrixXd sols(2, 2);
         sols << 0.0898, -0.7126,
             -0.0898, 0.7126;
+        sols(0, 0) = (sols(0, 0) + 3.0) / 6.0;
+        sols(1, 0) = (sols(1, 0) + 3.0) / 6.0;
+        sols(0, 1) = (sols(0, 1) + 2.0) / 4.0;
+        sols(1, 1) = (sols(1, 1) + 2.0) / 4.0;
         return sols;
     }
 };

src/limbo/bayes_opt/boptimizer.hpp

@@ -117,7 +117,7 @@ namespace limbo {
                     }
 
                     if (Params::bayes_opt_boptimizer::hp_period() > 0
-                        && this->_current_iteration % Params::bayes_opt_boptimizer::hp_period() == 0)
+                        && (this->_current_iteration + 1) % Params::bayes_opt_boptimizer::hp_period() == 0)
                         _model.optimize_hyperparams();
 
                     this->_current_iteration++;

src/limbo/kernel/squared_exp_ard.hpp

@@ -33,7 +33,7 @@ namespace limbo {
             {
                 //assert(Params::SquaredExpARD::k()<dim);
                 Eigen::VectorXd p = Eigen::VectorXd::Zero(_ell.size() + _ell.size() * Params::kernel_squared_exp_ard::k());
-                p.head(_ell.size()) = Eigen::VectorXd::Ones(_ell.size()) * -1;
+                p.head(_ell.size()) = Eigen::VectorXd::Ones(_ell.size());
                 this->set_h_params(p);
                 _sf2 = Params::kernel_squared_exp_ard::sigma_sq();
             }
@@ -46,7 +46,7 @@ namespace limbo {
             {
                 _h_params = p;
                 for (size_t i = 0; i < _input_dim; ++i)
-                    _ell(i) = std::exp(p(i));
+                    _ell(i) = p(i);
                 for (size_t j = 0; j < (unsigned int)Params::kernel_squared_exp_ard::k(); ++j)
                     for (size_t i = 0; i < _input_dim; ++i)
                         _A(i, j) = p((j + 1) * _input_dim + i); //can be negative

src/limbo/model/gp/kernel_lf_opt.hpp

@@ -18,11 +18,10 @@ namespace limbo {
                 template <typename GP>
                 void operator()(GP& gp) const
                 {
-                    int dim = gp.kernel_function().h_params_size();
                     KernelLFOptimization<GP> optimization(gp);
                     Optimizer optimizer;
-                    auto params = optimizer(optimization, tools::random_vector(dim), false);
-                    gp.kernel_function().set_h_params(params);
+                    auto params = optimizer(optimization, gp.kernel_function().h_params(), true);
+                    gp.kernel_function().set_h_params(-6.0 + params.array() * 7.0);
                     gp.set_lik(opt::eval(optimization, params));
                     gp.recompute(false);
                 }
@@ -36,7 +35,7 @@ namespace limbo {
                     opt::eval_t operator()(const Eigen::VectorXd& params, bool compute_grad) const
                     {
                         GP gp(this->_original_gp);
-                        gp.kernel_function().set_h_params(params);
+                        gp.kernel_function().set_h_params(-6.0 + params.array() * 7.0);
 
                         gp.recompute(false);
 

src/limbo/opt/nlopt_grad.hpp

@@ -79,6 +79,11 @@ namespace limbo {
                 }
                 catch (nlopt::roundoff_limited& e) {
                     // In theory it's ok to ignore this error
+                    std::cerr << "[NLOptGrad]: " << e.what() << std::endl;
+                }
+                catch (std::invalid_argument& e) {
+                    // In theory it's ok to ignore this error
+                    std::cerr << "[NLOptGrad]: " << e.what() << std::endl;
                 }
 
                 return Eigen::VectorXd::Map(x.data(), x.size());

src/limbo/opt/nlopt_no_grad.hpp

@@ -96,6 +96,11 @@ namespace limbo {
                 }
                 catch (nlopt::roundoff_limited& e) {
                     // In theory it's ok to ignore this error
+                    std::cerr << "[NLOptNoGrad]: " << e.what() << std::endl;
+                }
+                catch (std::invalid_argument& e) {
+                    // In theory it's ok to ignore this error
+                    std::cerr << "[NLOptNoGrad]: " << e.what() << std::endl;
                 }
 
                 return Eigen::VectorXd::Map(x.data(), x.size());

src/tests/test_kernel.cpp

@@ -29,8 +29,9 @@ BOOST_AUTO_TEST_CASE(test_kernel_SE_ARD)
 
     kernel::SquaredExpARD<Params> se(2);
     Eigen::VectorXd hp(se.h_params_size());
-    hp(0) = 0; //exp(0)=1
-    hp(1) = 0;
+    // kernel does not expect in log space anymore
+    hp(0) = 1;
+    hp(1) = 1;
 
     se.set_h_params(hp);
 
@@ -40,9 +41,9 @@ BOOST_AUTO_TEST_CASE(test_kernel_SE_ARD)
     Eigen::VectorXd v2 = make_v2(0, 1);
     double s1 = se(v1, v2);
 
-    BOOST_CHECK(std::abs(s1 - exp(-0.5 * (v1.transpose() * v2)[0])) < 1e-5);
+    BOOST_CHECK(std::abs(s1 - std::exp(-0.5 * (v1.transpose() * v2)[0])) < 1e-5);
 
-    hp(0) = 1;
+    hp(0) = std::exp(1);
     se.set_h_params(hp);
     double s2 = se(v1, v2);
     BOOST_CHECK(s1 < s2);
@@ -50,8 +51,8 @@ BOOST_AUTO_TEST_CASE(test_kernel_SE_ARD)
     Params::kernel_squared_exp_ard::set_k(1);
     se = kernel::SquaredExpARD<Params>(2);
     hp = Eigen::VectorXd(se.h_params_size());
-    hp(0) = 0;
-    hp(1) = 0;
+    hp(0) = 1;
+    hp(1) = 1;
     hp(2) = 0;
     hp(3) = 0;