Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Inverting a 4x4 matrix with try_inverse_mut doesn't leave self unchanged if the inversion fails. #1384

Merged
merged 8 commits into from May 5, 2024
Merged

Inverting a 4x4 matrix with try_inverse_mut doesn't leave self unchanged if the inversion fails. #1384

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
d331bbd
inversion
jnyfah Apr 19, 2024
a0da49f
early return
jnyfah Apr 22, 2024
c21df4e
remove else
jnyfah Apr 22, 2024
5e87799
missed one cofactor
jnyfah Apr 28, 2024
825d078
add tests
jnyfah Apr 30, 2024
e3a08c9
test with a fixed input
jnyfah May 2, 2024
914a7cf
add assert
jnyfah May 2, 2024
343eb21
format
jnyfah May 5, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
73 changes: 40 additions & 33 deletions src/linalg/inverse.rs
View file
Open in desktop
Expand Up @@ -145,13 +145,44 @@ where
{
let m = m.as_slice();

out[(0, 0)] = m[5].clone() * m[10].clone() * m[15].clone()
let cofactor00 = m[5].clone() * m[10].clone() * m[15].clone()
- m[5].clone() * m[11].clone() * m[14].clone()
- m[9].clone() * m[6].clone() * m[15].clone()
+ m[9].clone() * m[7].clone() * m[14].clone()
+ m[13].clone() * m[6].clone() * m[11].clone()
- m[13].clone() * m[7].clone() * m[10].clone();

let cofactor01 = -m[4].clone() * m[10].clone() * m[15].clone()
+ m[4].clone() * m[11].clone() * m[14].clone()
+ m[8].clone() * m[6].clone() * m[15].clone()
- m[8].clone() * m[7].clone() * m[14].clone()
- m[12].clone() * m[6].clone() * m[11].clone()
+ m[12].clone() * m[7].clone() * m[10].clone();

let cofactor02 = m[4].clone() * m[9].clone() * m[15].clone()
- m[4].clone() * m[11].clone() * m[13].clone()
- m[8].clone() * m[5].clone() * m[15].clone()
+ m[8].clone() * m[7].clone() * m[13].clone()
+ m[12].clone() * m[5].clone() * m[11].clone()
- m[12].clone() * m[7].clone() * m[9].clone();

let cofactor03 = -m[4].clone() * m[9].clone() * m[14].clone()
+ m[4].clone() * m[10].clone() * m[13].clone()
+ m[8].clone() * m[5].clone() * m[14].clone()
- m[8].clone() * m[6].clone() * m[13].clone()
- m[12].clone() * m[5].clone() * m[10].clone()
+ m[12].clone() * m[6].clone() * m[9].clone();

let det = m[0].clone() * cofactor00.clone()
+ m[1].clone() * cofactor01.clone()
+ m[2].clone() * cofactor02.clone()
+ m[3].clone() * cofactor03.clone();

if det.is_zero() {
return false;
}
out[(0, 0)] = cofactor00;

out[(1, 0)] = -m[1].clone() * m[10].clone() * m[15].clone()
+ m[1].clone() * m[11].clone() * m[14].clone()
+ m[9].clone() * m[2].clone() * m[15].clone()
Expand All @@ -173,12 +204,7 @@ where
- m[9].clone() * m[2].clone() * m[7].clone()
+ m[9].clone() * m[3].clone() * m[6].clone();

out[(0, 1)] = -m[4].clone() * m[10].clone() * m[15].clone()
+ m[4].clone() * m[11].clone() * m[14].clone()
+ m[8].clone() * m[6].clone() * m[15].clone()
- m[8].clone() * m[7].clone() * m[14].clone()
- m[12].clone() * m[6].clone() * m[11].clone()
+ m[12].clone() * m[7].clone() * m[10].clone();
out[(0, 1)] = cofactor01;

out[(1, 1)] = m[0].clone() * m[10].clone() * m[15].clone()
- m[0].clone() * m[11].clone() * m[14].clone()
Expand All @@ -201,12 +227,7 @@ where
+ m[8].clone() * m[2].clone() * m[7].clone()
- m[8].clone() * m[3].clone() * m[6].clone();

out[(0, 2)] = m[4].clone() * m[9].clone() * m[15].clone()
- m[4].clone() * m[11].clone() * m[13].clone()
- m[8].clone() * m[5].clone() * m[15].clone()
+ m[8].clone() * m[7].clone() * m[13].clone()
+ m[12].clone() * m[5].clone() * m[11].clone()
- m[12].clone() * m[7].clone() * m[9].clone();
out[(0, 2)] = cofactor02;

out[(1, 2)] = -m[0].clone() * m[9].clone() * m[15].clone()
+ m[0].clone() * m[11].clone() * m[13].clone()
Expand All @@ -222,12 +243,7 @@ where
+ m[12].clone() * m[1].clone() * m[7].clone()
- m[12].clone() * m[3].clone() * m[5].clone();

out[(0, 3)] = -m[4].clone() * m[9].clone() * m[14].clone()
+ m[4].clone() * m[10].clone() * m[13].clone()
+ m[8].clone() * m[5].clone() * m[14].clone()
- m[8].clone() * m[6].clone() * m[13].clone()
- m[12].clone() * m[5].clone() * m[10].clone()
+ m[12].clone() * m[6].clone() * m[9].clone();
out[(0, 3)] = cofactor03;

out[(3, 2)] = -m[0].clone() * m[5].clone() * m[11].clone()
+ m[0].clone() * m[7].clone() * m[9].clone()
Expand Down Expand Up @@ -257,21 +273,12 @@ where
+ m[8].clone() * m[1].clone() * m[6].clone()
- m[8].clone() * m[2].clone() * m[5].clone();

let det = m[0].clone() * out[(0, 0)].clone()
+ m[1].clone() * out[(0, 1)].clone()
+ m[2].clone() * out[(0, 2)].clone()
+ m[3].clone() * out[(0, 3)].clone();
let inv_det = T::one() / det;

if !det.is_zero() {
let inv_det = T::one() / det;

for j in 0..4 {
for i in 0..4 {
out[(i, j)] *= inv_det.clone();
}
for j in 0..4 {
for i in 0..4 {
out[(i, j)] *= inv_det.clone();
}
true
} else {
false
}
true
}
10 changes: 9 additions & 1 deletion tests/core/matrix.rs
View file
Open in desktop
Expand Up @@ -931,7 +931,7 @@ mod inversion_tests {
use super::*;
use crate::proptest::*;
use na::Matrix1;
use proptest::{prop_assert, proptest};
use proptest::{prop_assert, proptest, prop_assert_eq};

proptest! {
#[test]
Expand Down Expand Up @@ -970,6 +970,14 @@ mod inversion_tests {
}
}

#[test]
fn test_inversion_failure_leaves_matrix_unchanged(m in matrix4()) {
let original_matrix = m.clone();
if m.try_inverse().is_none() {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this branch actually get hit? I'm not sure how often proptest generates non-invertible matrices. Would a hardcoded case work better?

Copy link
Sponsor Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Agree with @Ralith, I think the likelihood of matrix4() producing singular matrices is likely (very) small (though I haven't checked), so a test with a fixed input is better in this case.

prop_assert_eq!(m, original_matrix);
}
}

#[test]
fn self_mul_inv_is_id_dim6(m in matrix6()) {
if let Some(im) = m.try_inverse() {
Expand Down