You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I was having a look at the code behind LUFactorized and it appears that there is no LU factorization done at all. For example, the factorize_into source is
where self here is the original matrix. While it is fine to be storing the original matrix A (perhaps the LU decomposition is lazily evaluated), the solve_inplace code indicates that this is actually not the case at all. Instead it just uses the standard solve function without actually doing any LU decomposition at any point.
fnsolve_inplace<'a,Sb>(&self,rhs:&'a mutArrayBase<Sb,Ix1>,) -> Result<&'a mutArrayBase<Sb,Ix1>>whereSb:DataMut<Elem = A>,{assert_eq!(
rhs.len(),self.a.len_of(Axis(1)),"The length of `rhs` must be compatible with the shape of the factored matrix.",);A::solve(self.a.square_layout()?,Transpose::No,self.a.as_allocated()?,&self.ipiv,
rhs.as_slice_mut().unwrap(),)?;Ok(rhs)}
Am I missing something? Or is this a temporary implementation until an LU factorization can be properly implemented?
The text was updated successfully, but these errors were encountered:
I was having a look at the code behind
LUFactorized
and it appears that there is no LU factorization done at all. For example, thefactorize_into
source iswhere
self
here is the original matrix. While it is fine to be storing the original matrixA
(perhaps the LU decomposition is lazily evaluated), thesolve_inplace
code indicates that this is actually not the case at all. Instead it just uses the standardsolve
function without actually doing any LU decomposition at any point.Am I missing something? Or is this a temporary implementation until an LU factorization can be properly implemented?
The text was updated successfully, but these errors were encountered: