forked from M-Labs/nalgebra
Merge pull request #1314 from rasmusgo/fix-svd-near-zero
Fix bug in SVD related to values near zero
This commit is contained in:
commit
2e99320d01
@ -8,6 +8,7 @@ use simba::scalar::ComplexField;
|
|||||||
|
|
||||||
use crate::geometry::Reflection;
|
use crate::geometry::Reflection;
|
||||||
use crate::linalg::householder;
|
use crate::linalg::householder;
|
||||||
|
use crate::num::Zero;
|
||||||
use std::mem::MaybeUninit;
|
use std::mem::MaybeUninit;
|
||||||
|
|
||||||
/// The bidiagonalization of a general matrix.
|
/// The bidiagonalization of a general matrix.
|
||||||
@ -227,7 +228,11 @@ where
|
|||||||
|
|
||||||
for i in (0..dim - shift).rev() {
|
for i in (0..dim - shift).rev() {
|
||||||
let axis = self.uv.view_range(i + shift.., i);
|
let axis = self.uv.view_range(i + shift.., i);
|
||||||
// TODO: sometimes, the axis might have a zero magnitude.
|
|
||||||
|
// Sometimes, the axis might have a zero magnitude.
|
||||||
|
if axis.norm_squared().is_zero() {
|
||||||
|
continue;
|
||||||
|
}
|
||||||
let refl = Reflection::new(Unit::new_unchecked(axis), T::zero());
|
let refl = Reflection::new(Unit::new_unchecked(axis), T::zero());
|
||||||
|
|
||||||
let mut res_rows = res.view_range_mut(i + shift.., i..);
|
let mut res_rows = res.view_range_mut(i + shift.., i..);
|
||||||
@ -263,7 +268,11 @@ where
|
|||||||
let axis = self.uv.view_range(i, i + shift..);
|
let axis = self.uv.view_range(i, i + shift..);
|
||||||
let mut axis_packed = axis_packed.rows_range_mut(i + shift..);
|
let mut axis_packed = axis_packed.rows_range_mut(i + shift..);
|
||||||
axis_packed.tr_copy_from(&axis);
|
axis_packed.tr_copy_from(&axis);
|
||||||
// TODO: sometimes, the axis might have a zero magnitude.
|
|
||||||
|
// Sometimes, the axis might have a zero magnitude.
|
||||||
|
if axis_packed.norm_squared().is_zero() {
|
||||||
|
continue;
|
||||||
|
}
|
||||||
let refl = Reflection::new(Unit::new_unchecked(axis_packed), T::zero());
|
let refl = Reflection::new(Unit::new_unchecked(axis_packed), T::zero());
|
||||||
|
|
||||||
let mut res_rows = res.view_range_mut(i.., i + shift..);
|
let mut res_rows = res.view_range_mut(i.., i + shift..);
|
||||||
|
@ -17,7 +17,7 @@ pub struct GivensRotation<T: ComplexField> {
|
|||||||
|
|
||||||
// Matrix = UnitComplex * Matrix
|
// Matrix = UnitComplex * Matrix
|
||||||
impl<T: ComplexField> GivensRotation<T> {
|
impl<T: ComplexField> GivensRotation<T> {
|
||||||
/// The Givents rotation that does nothing.
|
/// The Givens rotation that does nothing.
|
||||||
pub fn identity() -> Self {
|
pub fn identity() -> Self {
|
||||||
Self {
|
Self {
|
||||||
c: T::RealField::one(),
|
c: T::RealField::one(),
|
||||||
@ -88,13 +88,13 @@ impl<T: ComplexField> GivensRotation<T> {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/// The cos part of this roration.
|
/// The cos part of this rotation.
|
||||||
#[must_use]
|
#[must_use]
|
||||||
pub fn c(&self) -> T::RealField {
|
pub fn c(&self) -> T::RealField {
|
||||||
self.c.clone()
|
self.c.clone()
|
||||||
}
|
}
|
||||||
|
|
||||||
/// The sin part of this roration.
|
/// The sin part of this rotation.
|
||||||
#[must_use]
|
#[must_use]
|
||||||
pub fn s(&self) -> T {
|
pub fn s(&self) -> T {
|
||||||
self.s.clone()
|
self.s.clone()
|
||||||
|
@ -1,6 +1,6 @@
|
|||||||
#![cfg(feature = "proptest-support")]
|
#[cfg(feature = "proptest-support")]
|
||||||
|
mod proptest_tests {
|
||||||
macro_rules! gen_tests(
|
macro_rules! gen_tests(
|
||||||
($module: ident, $scalar: expr) => {
|
($module: ident, $scalar: expr) => {
|
||||||
mod $module {
|
mod $module {
|
||||||
#[allow(unused_imports)]
|
#[allow(unused_imports)]
|
||||||
@ -54,8 +54,9 @@ macro_rules! gen_tests(
|
|||||||
}
|
}
|
||||||
);
|
);
|
||||||
|
|
||||||
gen_tests!(complex, complex_f64());
|
gen_tests!(complex, complex_f64());
|
||||||
gen_tests!(f64, PROPTEST_F64);
|
gen_tests!(f64, PROPTEST_F64);
|
||||||
|
}
|
||||||
|
|
||||||
#[test]
|
#[test]
|
||||||
fn bidiagonal_identity() {
|
fn bidiagonal_identity() {
|
||||||
@ -74,3 +75,31 @@ fn bidiagonal_identity() {
|
|||||||
let (u, d, v_t) = bidiagonal.unpack();
|
let (u, d, v_t) = bidiagonal.unpack();
|
||||||
assert_eq!(m, &u * d * &v_t);
|
assert_eq!(m, &u * d * &v_t);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn bidiagonal_regression_issue_1313() {
|
||||||
|
let s = 6.123234e-16_f32;
|
||||||
|
let mut m = nalgebra::dmatrix![
|
||||||
|
10.0, 0.0, 0.0, 0.0, -10.0, 0.0, 0.0, 0.0;
|
||||||
|
s, 10.0, 0.0, 10.0, s, 0.0, 0.0, 0.0;
|
||||||
|
20.0, -20.0, 0.0, 20.0, 20.0, 0.0, 0.0, 0.0;
|
||||||
|
];
|
||||||
|
m.unscale_mut(m.camax());
|
||||||
|
let bidiagonal = m.clone().bidiagonalize();
|
||||||
|
let (u, d, v_t) = bidiagonal.unpack();
|
||||||
|
let m2 = &u * d * &v_t;
|
||||||
|
assert_relative_eq!(m, m2, epsilon = 1e-6);
|
||||||
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn bidiagonal_regression_issue_1313_minimal() {
|
||||||
|
let s = 6.123234e-17_f32;
|
||||||
|
let m = nalgebra::dmatrix![
|
||||||
|
1.0, 0.0, -1.0;
|
||||||
|
s, 1.0, s;
|
||||||
|
];
|
||||||
|
let bidiagonal = m.clone().bidiagonalize();
|
||||||
|
let (u, d, v_t) = bidiagonal.unpack();
|
||||||
|
let m2 = &u * &d * &v_t;
|
||||||
|
assert_relative_eq!(m, m2, epsilon = 1e-6);
|
||||||
|
}
|
||||||
|
@ -499,3 +499,17 @@ fn svd_regression_issue_1072() {
|
|||||||
epsilon = 1e-9
|
epsilon = 1e-9
|
||||||
);
|
);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
// Exercises bug reported in issue #1313 of nalgebra (https://github.com/dimforge/nalgebra/issues/1313)
|
||||||
|
fn svd_regression_issue_1313() {
|
||||||
|
let s = 6.123234e-16_f32;
|
||||||
|
let m = nalgebra::dmatrix![
|
||||||
|
10.0, 0.0, 0.0, 0.0, -10.0, 0.0, 0.0, 0.0;
|
||||||
|
s, 10.0, 0.0, 10.0, s, 0.0, 0.0, 0.0;
|
||||||
|
20.0, -20.0, 0.0, 20.0, 20.0, 0.0, 0.0, 0.0;
|
||||||
|
];
|
||||||
|
let svd = m.clone().svd(true, true);
|
||||||
|
let m2 = svd.recompose().unwrap();
|
||||||
|
assert_relative_eq!(&m, &m2, epsilon = 1e-5);
|
||||||
|
}
|
||||||
|
Loading…
Reference in New Issue
Block a user