Hi, I would like to obtain some elements of the inverse matrix, based on a user selection of the indexes. Is it possible to use the selected inversion process of Pardiso?
Acuattly, I tried to play with iparm(36) and iparm(37) with "local (internal) PARDISO version is : 103911000", but I obtain an error.
Current version of MKL doesn't support such functionality. But can i ask you the reason of such request? Probably I can provide advice, how it can be calculated based on current feature set.
Thank you. I have to invert a very sparse matrix representing a sort of base projection. For physical reasons I know the sparsity pattern of the inverse, and I need the elements of the inverse for further elaboration. My question is if is there a method faster than multiply the inverse matrix with N vectors having only one unitary element.