ic
/
magnum
mirror of https://github.com/mosra/magnum.git
6
0
Fork 0
Code Issues Projects Releases Wiki Activity
Browse Source

Math: remove std::tuple usage from Algorithms::svd(). 

There's Containers::Triple now for this instead. Printing a message to
Error in case of a failure might have made sense back in 2010, but now
it absolutely doesn't, so it's additionally wrapped in an Optional now.

Also looks like the actual use without "convenient" std::tie() is a lot
less verbose. Haha.
pull/168/head
Vladimír Vondruš 3 years ago
parent
90cf45ddc7
commit
162f4fd6ec
5 changed files with 33 additions and 38 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 5
      doc/changelog.dox
  2. 8
      doc/snippets/MagnumMathAlgorithms.cpp
  3. 12
      src/Magnum/Math/Algorithms/Svd.h
  4. 44
      src/Magnum/Math/Algorithms/Test/SvdTest.cpp
  5. 2
      src/singles/MagnumMath.hpp

5
doc/changelog.dox
Unescape View File

@ -1480,6 +1480,11 @@ See also:
for implicit conversions from/to a @ref std::string and/or @ref std::pair,
but in some cases you may be forced to change the code that uses those
APIs.
- @ref Math::Algorithms::svd() that used to return a @ref std::tuple and
printed a message to @relativeref{Magnum,Error} if it didn't converge now
returns a @ref Containers::Tuple wrapped in an @ref Containers::Optional.
As the optional makes this a breaking change, no backwards compatibility
header for tuple conversion is included.
- @ref Image, @ref ImageView and @ref Trade::ImageData now look for a
@cpp pixelFormatSize() @ce API via ADL instead of @cpp pixelSize() @ce. In
case you were passing a custom pixel format enum to the image classes, you

8
doc/snippets/MagnumMathAlgorithms.cpp
Unescape View File

@ -62,11 +62,9 @@ enum: std::size_t { cols = 3, rows = 4 };
/* [svd] */
Math::RectangularMatrix<cols, rows, Double> m;
Math::RectangularMatrix<cols, rows, Double> uPart;
Math::Vector<cols, Double> wDiagonal;
Math::Matrix<cols, Double> v;
std::tie(uPart, wDiagonal, v) = Math::Algorithms::svd(m);
auto svd = Math::Algorithms::svd(m);
Math::RectangularMatrix<cols, rows, Double> uPart = svd->first();
Math::Vector<cols, Double> wDiagonal = svd->second();
// Extend U
Math::Matrix<rows, Double> u{Math::ZeroInit};

12
src/Magnum/Math/Algorithms/Svd.h
Unescape View File

@ -29,7 +29,8 @@
* @brief Function @ref Magnum::Math::Algorithms::svd()
*/
#include <tuple>
#include <Corrade/Containers/Optional.h>
#include <Corrade/Containers/Triple.h>
#include "Magnum/Math/Functions.h"
#include "Magnum/Math/Matrix.h"
@ -65,7 +66,7 @@ on given matrix where @p rows >= `cols`: @f[
Returns first @p cols column vectors of @f$ \boldsymbol{U} @f$, diagonal of
@f$ \boldsymbol{\Sigma} @f$ and non-transposed @f$ \boldsymbol{V} @f$. If the
solution doesn't converge, returns zero matrices.
solution doesn't converge, returns @ref Containers::NullOpt.
Full @f$ \boldsymbol{U} @f$, @f$ \boldsymbol{\Sigma} @f$ matrices and original
@f$ \boldsymbol{M} @f$ matrix can be reconstructed from the values as
@ -81,7 +82,7 @@ for an example. Implementation based on *Golub, G. H.; Reinsch, C. (1970).
@see @ref qr(), @ref Matrix3::rotationShear(), @ref Matrix4::rotationShear()
*/
/* The matrix is passed by value because it is changed inside */
template<std::size_t cols, std::size_t rows, class T> std::tuple<RectangularMatrix<cols, rows, T>, Vector<cols, T>, Matrix<cols, T>> svd(RectangularMatrix<cols, rows, T> m) {
template<std::size_t cols, std::size_t rows, class T> Containers::Optional<Containers::Triple<RectangularMatrix<cols, rows, T>, Vector<cols, T>, Matrix<cols, T>>> svd(RectangularMatrix<cols, rows, T> m) {
static_assert(rows >= cols, "Unsupported matrix aspect ratio");
static_assert(T(1)+TypeTraits<T>::epsilon() > T(1), "Epsilon too small");
constexpr T tol = Implementation::smallestDelta<T>()/TypeTraits<T>::epsilon();
@ -255,8 +256,7 @@ template<std::size_t cols, std::size_t rows, class T> std::tuple<RectangularMatr
/* Exceeded iteration count, done */
} else if(iteration >= maxIterations-1) {
Error{} << "Magnum::Math::Algorithms::svd(): no convergence";
return std::make_tuple(RectangularMatrix<cols, rows, T>{}, Vector<cols, T>{}, Matrix<cols, T>{ZeroInit});
return {};
}
/* Shift from bottom 2x2 minor */
@ -316,7 +316,7 @@ template<std::size_t cols, std::size_t rows, class T> std::tuple<RectangularMatr
}
}
return std::make_tuple(m, q, v);
return Containers::triple(m, q, v);
}
}}}

44
src/Magnum/Math/Algorithms/Test/SvdTest.cpp
Unescape View File

@ -62,37 +62,35 @@ template<class T> void SvdTest::test() {
Vector8<T>{T{ 7}, T{ 8}, T{ 3}, T{ 4}, T{ 4}, T{-1}, T{ 1}, T{ 2}}};
const Vector5<T> expected(std::sqrt(T{1248}), T{0}, T{20}, std::sqrt(T{384}), T{0});
Matrix5x8<T> u{Magnum::NoInit};
Vector5<T> w{Magnum::NoInit};
Matrix5<T> v{Magnum::NoInit};
std::tie(u, w, v) = Algorithms::svd(a);
Containers::Optional<Containers::Triple<Matrix5x8<T>, Vector5<T>, Matrix5<T>>> uwv = Algorithms::svd(a);
CORRADE_VERIFY(uwv);
/* Test composition */
Matrix8<T> u2{u[0], u[1], u[2], u[3], u[4], Vector8<T>{}, Vector8<T>{}, Vector8<T>{}};
Matrix5x8<T> w2 = Matrix5x8<T>::fromDiagonal(w);
Matrix8<T> u2{uwv->first()[0], uwv->first()[1], uwv->first()[2], uwv->first()[3], uwv->first()[4], Vector8<T>{}, Vector8<T>{}, Vector8<T>{}};
Matrix5x8<T> w2 = Matrix5x8<T>::fromDiagonal(uwv->second());
{
#ifdef CORRADE_TARGET_EMSCRIPTEN
CORRADE_EXPECT_FAIL_IF((std::is_same<T, Double>::value) && u2*w2*v.transposed() != a,
CORRADE_EXPECT_FAIL_IF((std::is_same<T, Double>::value) && u2*w2*uwv->third().transposed() != a,
"Some strange problems with Double on recent Emscripten versions "
"(1.36.5 worked fine, 1.37.1 works fine on larger optimization "
"levels, not on -O1).");
#endif
CORRADE_COMPARE(u2*w2*v.transposed(), a);
CORRADE_COMPARE(u2*w2*uwv->third().transposed(), a);
}
/* Test that V is unitary */
CORRADE_COMPARE(v*v.transposed(), Matrix5<T>{IdentityInit});
CORRADE_COMPARE(v.transposed()*v, Matrix5<T>{IdentityInit});
CORRADE_COMPARE(uwv->third()*uwv->third().transposed(), Matrix5<T>{IdentityInit});
CORRADE_COMPARE(uwv->third().transposed()*uwv->third(), Matrix5<T>{IdentityInit});
/* Test W */
{
#ifdef CORRADE_TARGET_EMSCRIPTEN
CORRADE_EXPECT_FAIL_IF((std::is_same<T, Double>::value && w != expected),
CORRADE_EXPECT_FAIL_IF((std::is_same<T, Double>::value && uwv->second() != expected),
"Some strange problems with Double on recent Emscripten versions "
"(1.36.5 worked fine, 1.37.1 worked fine on larger optimization "
"levels, not on -O1, 1.37.5 works fine again).");
#endif
CORRADE_COMPARE(w, expected);
CORRADE_COMPARE(uwv->second(), expected);
}
}
@ -105,17 +103,14 @@ void SvdTest::decomposeRotationScaling() {
Matrix4 a = Matrix4::rotationZ(35.0_degf)*Matrix4::scaling({1.5f, 2.0f, 1.0f});
Matrix3x3 u{Magnum::NoInit};
Vector3 w{Magnum::NoInit};
Matrix3x3 v{Magnum::NoInit};
std::tie(u, w, v) = Algorithms::svd(a.rotationScaling());
Containers::Triple<Matrix3x3, Vector3, Matrix3x3> uwv{*Algorithms::svd(a.rotationScaling())};
CORRADE_COMPARE(u*Matrix3x3::fromDiagonal(w)*v.transposed(), a.rotationScaling());
CORRADE_COMPARE(uwv.first()*Matrix3x3::fromDiagonal(uwv.second())*uwv.third().transposed(), a.rotationScaling());
/* V contains flipped signs for the whole matrix, use it to fix the
signs for U */
CORRADE_COMPARE(w, (Vector3{1.5f, 2.0f, 1.0f}));
CORRADE_COMPARE(Matrix4::from(u*v.transposed(), {}), Matrix4::rotationZ(35.0_degf));
CORRADE_COMPARE(uwv.second(), (Vector3{1.5f, 2.0f, 1.0f}));
CORRADE_COMPARE(Matrix4::from(uwv.first()*uwv.third().transposed(), {}), Matrix4::rotationZ(35.0_degf));
}
void SvdTest::decomposeRotationShear() {
@ -128,17 +123,14 @@ void SvdTest::decomposeRotationShear() {
/* Like above, but with order flipped, which results in a shear */
Matrix4 a = Matrix4::scaling({1.5f, 2.0f, 1.0f})*Matrix4::rotationZ(35.0_degf);
Matrix3x3 u{Magnum::NoInit};
Vector3 w{Magnum::NoInit};
Matrix3x3 v{Magnum::NoInit};
std::tie(u, w, v) = Algorithms::svd(a.rotationScaling());
Containers::Triple<Matrix3x3, Vector3, Matrix3x3> uwv{*Algorithms::svd(a.rotationScaling())};
CORRADE_COMPARE(u*Matrix3x3::fromDiagonal(w)*v.transposed(), a.rotationScaling());
CORRADE_COMPARE(uwv.first()*Matrix3x3::fromDiagonal(uwv.second())*uwv.third().transposed(), a.rotationScaling());
/* U contains a flipped sign for Z, use it to remove the sign from the
transposed rotation matrix V */
CORRADE_COMPARE(w, (Vector3{1.5f, 2.0f, 1.0f}));
CORRADE_COMPARE(Matrix4::from(u*v.transposed(), {}), Matrix4::rotationZ(35.0_degf));
CORRADE_COMPARE(uwv.second(), (Vector3{1.5f, 2.0f, 1.0f}));
CORRADE_COMPARE(Matrix4::from(uwv.first()*uwv.third().transposed(), {}), Matrix4::rotationZ(35.0_degf));
}
}}}}}

2
src/singles/MagnumMath.hpp
Unescape View File

@ -393,7 +393,7 @@ typedef Math::Frustum<Double> Frustumd;
#include "Magnum/Math/Algorithms/GramSchmidt.h"
#include "Magnum/Math/Algorithms/KahanSum.h"
#include "Magnum/Math/Algorithms/Qr.h"
//#include "Magnum/Math/Algorithms/Svd.h" // TODO: uses <tuple>
//#include "Magnum/Math/Algorithms/Svd.h" // TODO needs Optional and Triple
#ifdef MAGNUM_MATH_GLM_INTEGRATION
// {{includes}}
#include "Magnum/GlmIntegration/Integration.h"

Write Preview
Loading…
Cancel
Save