topic Hi, in Intel® oneAPI Math Kernel Library

What is error code -4 for cluster_sparse_solver routine

Piotr_G_1 — Wed, 28 Mar 2018 15:38:12 GMT

Hi,

I am using CLUSTER_SPARSE_SOLVER with real symmetric matrices. During phase 22 (Numerical factorization) I obtained error code equal to -4. Unfortunately, in the reference manual there is no explanation for -4. Could you help me to interpret it?

Thanks,
Piotr

Hi,

Alexander_K_Intel2 — Wed, 28 Mar 2018 16:18:00 GMT

Hi,

What is the matrix type of your matrix? Such situation can appeared in case of zero pivot on diagonal for positive defined matrix

Thanks,

Alex

Hi Alexander,

Piotr_G_1 — Thu, 29 Mar 2018 11:31:04 GMT

Hi Alexander,

I am defining mtype as 2 (real and symmetric positive definite). I have checked the diagonal and there are no zeros. The sparse matrix I am using was generated by external tool and this matrix is supposed to be of mtype 2 (positive definite). So, -4 means that there is something wrong with the matrix?

Thanks,
Piotr