looks as if I need to use mkl_?csradd (C++).
However, my csr matrix is zero based; mkl_?csradd does only support one based sparse matrices.
I would appreciate any suggestions for a pragmatic solution.
A simple application example of mkl_?csradd would be nice, too.
I cannot help you as to zero-based indexing, other than to point out something for you to consider: although some programming languages default to zero-based indexing, in mathematics and physics it is more common to use 1-based indices. For example, we use x1, x2 and x3 as position coordinates rather than x0, x1 and x2 . What are the natural indices in your subject domain?
MKL provides the following Fortran examples in examples_core.zip:
As it is, addition of sparse matrices is not a complicated subject. An entry in the result matrix C is non-zero if either or both of the corresponding entries in A and B are non-zero. If only one of the two is non-zero, we copy the value to C; when both are non-zero, we place the sum in C.