GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-08-11 19:29:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

work around stupid msvc error when constructing at compile time an expression

Browse Source 
that involves a division by zero, even if the numeric type has floating point
This commit is contained in:
Benoit Jacob 2010-10-19 21:56:11 -04:00
parent e5073746f3
commit 9044c98cff
2 changed files with 17 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
Eigen/src/SVD/JacobiSVD.h
View File

@ -239,6 +239,9 @@ struct ei_qr_preconditioner_impl<MatrixType, HouseholderQRPreconditioner, Precon
* \a p is the greater dimension, meaning that it is still of the same order of complexity as the faster bidiagonalizing R-SVD algorithms. * \a p is the greater dimension, meaning that it is still of the same order of complexity as the faster bidiagonalizing R-SVD algorithms.
* In particular, like any R-SVD, it takes advantage of non-squareness in that its complexity is only linear in the greater dimension. * In particular, like any R-SVD, it takes advantage of non-squareness in that its complexity is only linear in the greater dimension.
* *
* If the input matrix has inf or nan coefficients, the result of the computation is undefined, but the computation is guaranteed to
* terminate in finite (and reasonable) time.
*
* The possible values for QRPreconditioner are: * The possible values for QRPreconditioner are:
* \li ColPivHouseholderQRPreconditioner is the default. In practice it's very safe. It uses column-pivoting QR. * \li ColPivHouseholderQRPreconditioner is the default. In practice it's very safe. It uses column-pivoting QR.
* \li FullPivHouseholderQRPreconditioner, is the safest and slowest. It uses full-pivoting QR. * \li FullPivHouseholderQRPreconditioner, is the safest and slowest. It uses full-pivoting QR.

16
test/jacobisvd.cpp
View File

@ -209,18 +209,30 @@ void jacobisvd_method()
VERIFY_IS_APPROX(m.jacobiSvd(ComputeFullU|ComputeFullV).solve(m), m); VERIFY_IS_APPROX(m.jacobiSvd(ComputeFullU|ComputeFullV).solve(m), m);
} }
// work around stupid msvc error when constructing at compile time an expression that involves
// a division by zero, even if the numeric type has floating point
template<typename Scalar>
EIGEN_DONT_INLINE Scalar zero() { return Scalar(0); }
template<typename MatrixType> template<typename MatrixType>
void jacobisvd_inf_nan() void jacobisvd_inf_nan()
{ {
// all this function does is verify we don't iterate infinitely on nan/inf values
JacobiSVD<MatrixType> svd; JacobiSVD<MatrixType> svd;
typedef typename MatrixType::Scalar Scalar; typedef typename MatrixType::Scalar Scalar;
Scalar some_inf = Scalar(1) / Scalar(0); Scalar some_inf = Scalar(1) / zero<Scalar>();
VERIFY((some_inf - some_inf) != (some_inf - some_inf));
svd.compute(MatrixType::Constant(10,10,some_inf), ComputeFullU | ComputeFullV); svd.compute(MatrixType::Constant(10,10,some_inf), ComputeFullU | ComputeFullV);
Scalar some_nan = Scalar(0) / Scalar(0);
Scalar some_nan = zero<Scalar>() / zero<Scalar>();
VERIFY(some_nan != some_nan);
svd.compute(MatrixType::Constant(10,10,some_nan), ComputeFullU | ComputeFullV); svd.compute(MatrixType::Constant(10,10,some_nan), ComputeFullU | ComputeFullV);
MatrixType m = MatrixType::Zero(10,10); MatrixType m = MatrixType::Zero(10,10);
m(ei_random<int>(0,9), ei_random<int>(0,9)) = some_inf; m(ei_random<int>(0,9), ei_random<int>(0,9)) = some_inf;
svd.compute(m, ComputeFullU | ComputeFullV); svd.compute(m, ComputeFullU | ComputeFullV);
m = MatrixType::Zero(10,10); m = MatrixType::Zero(10,10);
m(ei_random<int>(0,9), ei_random<int>(0,9)) = some_nan; m(ei_random<int>(0,9), ei_random<int>(0,9)) = some_nan;
svd.compute(m, ComputeFullU | ComputeFullV); svd.compute(m, ComputeFullU | ComputeFullV);