Rasmus Munk LarsenandRasmus Munk Larsen
3006c69d4e
libeigen/eigen!2702 Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
30 lines
1.2 KiB
C++
30 lines
1.2 KiB
C++
// SPDX-FileCopyrightText: The Eigen Authors
|
|
// SPDX-License-Identifier: MPL-2.0
|
|
|
|
#include "main.h"
|
|
#include <Eigen/SVD>
|
|
|
|
EIGEN_DECLARE_TEST(bdcsvd_fastmath) {
|
|
typedef Matrix<double, 6, 6> Matrix6d;
|
|
const double kTolerance = 16 * Matrix6d::RowsAtCompileTime * NumTraits<double>::epsilon();
|
|
|
|
// A finite singular-vector coefficient grows to about 2^570. Its ordinary
|
|
// squared norm overflows, and GCC's -ffast-math assumes isfinite() is true.
|
|
Matrix6d matrix = Matrix6d::Zero();
|
|
using std::ldexp;
|
|
matrix.diagonal() << 0.0, 0.0, ldexp(1.0, -487), -1.0, 0.0, 0.0;
|
|
matrix.diagonal(1) << 0.0, ldexp(1.0, -453), -ldexp(1.0, -627), 0.0, 0.0;
|
|
|
|
BDCSVD<Matrix6d, ComputeFullU | ComputeFullV> svd;
|
|
svd.setSwitchSize(3);
|
|
svd.compute(matrix);
|
|
|
|
VERIFY(svd.info() == Success);
|
|
const Matrix6d reconstruction = svd.matrixU() * svd.singularValues().asDiagonal() * svd.matrixV().transpose();
|
|
VERIFY((reconstruction - matrix).stableNorm() <= kTolerance * matrix.stableNorm());
|
|
|
|
const Matrix6d identity = Matrix6d::Identity();
|
|
VERIFY((svd.matrixU().transpose() * svd.matrixU() - identity).stableNorm() <= kTolerance);
|
|
VERIFY((svd.matrixV().transpose() * svd.matrixV() - identity).stableNorm() <= kTolerance);
|
|
}
|