Determinant of a large sparse matrix

Accelerated Computing GPU-Accelerated Libraries
1

Hi,

I need to calculate the determinant of a large complex sparse matrix.
I start by LU factorization and then perform the sum of all logarithmic entries in order to avoid overflow.
To do the LU factorization I tried cusparseZcsrilu0, cusparseZcsrilu02 and others but all failed
in the manual, it says that using this methods the matrix becomes:
A=LU
However, the results makes non-sense.

The is any better method in cusolver or cupsarse libraries to compute the determimant of a sparse matrix?
Or what is the best method to do the LU factorization?

Also, in some methods we have to create and pass a descriptor for U and L, there is any way to access to this matrices?

Thanks,
Nuno

2

That is expected:

https://devtalk.nvidia.com/default/topic/1046216/gpu-accelerated-libraries/solving-linear-system-with-cusparse/

3

There is any way to get U matrix?

4

It is very disappointing that those NVIDIA people does not respond in a direct manner.
At least they should tell whether it is possible to do so.

It’s really a waste of time digging so much but get nothing except their irresponsible and nonprofessional reply