1 #include "main.h" 2 #include <Eigen/MPRealSupport> 3 #include <Eigen/LU> 4 #include <Eigen/Eigenvalues> 5 #include <sstream> 6 7 using namespace mpfr; 8 using namespace Eigen; 9 10 void test_mpreal_support() 11 { 12 // set precision to 256 bits (double has only 53 bits) 13 mpreal::set_default_prec(256); 14 typedef Matrix<mpreal,Eigen::Dynamic,Eigen::Dynamic> MatrixXmp; 15 16 std::cerr << "epsilon = " << NumTraits<mpreal>::epsilon() << "\n"; 17 std::cerr << "dummy_precision = " << NumTraits<mpreal>::dummy_precision() << "\n"; 18 std::cerr << "highest = " << NumTraits<mpreal>::highest() << "\n"; 19 std::cerr << "lowest = " << NumTraits<mpreal>::lowest() << "\n"; 20 21 for(int i = 0; i < g_repeat; i++) { 22 int s = Eigen::internal::random<int>(1,100); 23 MatrixXmp A = MatrixXmp::Random(s,s); 24 MatrixXmp B = MatrixXmp::Random(s,s); 25 MatrixXmp S = A.adjoint() * A; 26 MatrixXmp X; 27 28 // Basic stuffs 29 VERIFY_IS_APPROX(A.real(), A); 30 VERIFY(Eigen::internal::isApprox(A.array().abs2().sum(), A.squaredNorm())); 31 VERIFY_IS_APPROX(A.array().exp(), exp(A.array())); 32 VERIFY_IS_APPROX(A.array().abs2().sqrt(), A.array().abs()); 33 VERIFY_IS_APPROX(A.array().sin(), sin(A.array())); 34 VERIFY_IS_APPROX(A.array().cos(), cos(A.array())); 35 36 37 // Cholesky 38 X = S.selfadjointView<Lower>().llt().solve(B); 39 VERIFY_IS_APPROX((S.selfadjointView<Lower>()*X).eval(),B); 40 41 // partial LU 42 X = A.lu().solve(B); 43 VERIFY_IS_APPROX((A*X).eval(),B); 44 45 // symmetric eigenvalues 46 SelfAdjointEigenSolver<MatrixXmp> eig(S); 47 VERIFY_IS_EQUAL(eig.info(), Success); 48 VERIFY( (S.selfadjointView<Lower>() * eig.eigenvectors()).isApprox(eig.eigenvectors() * eig.eigenvalues().asDiagonal(), NumTraits<mpreal>::dummy_precision()*1e3) ); 49 } 50 51 { 52 MatrixXmp A(8,3); A.setRandom(); 53 // test output (interesting things happen in this code) 54 std::stringstream stream; 55 stream << A; 56 } 57 } 58
