compute determinant using Cholesky decomposition

I need to compute determinant of a positive definite, hermitian matrix in fastest way for my code. So the best way is to compute by cholesky decomposition, but on writing code for it there is no improvement over MATLAB built-in function "det" which is based on LU decomposition (more complex than cholskey). Can anyone help, can we modify matlab buit-in function "chol" to determine determinant from it directly.

[Gaurav Gupta]

Can MATLAB people help me with this

[youtha]

Try using

:)

L=chol(A)
p=1;
for i=1:n
p=p*L(i,i)^2
end

Answer 1

[Walter Roberson]

Keep in mind that for sufficiently large matrices, MATLAB is going to invoke multi-threaded library code that has been heavily optimized for the target architectures. (It doesn't do that for smaller matrices because there is notable overhead in re-arranging the arrays into the form required by those libraries.)

Answer 2

[Teja Muppirala]

You could try

prod(diag(chol(A)))^2

But I have no idea if/when this would be faster than simply det(A).

[Gaurav Gupta]

I have tried this before posting question, but there is no improvement over time.