https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/const-correctness/m-p/884605#M9939 Hi,<BR /><BR />I wrote a modular PDE-solver in C++ which makes heavy use of PARDISO and CRS-matrix-vector-multiplication. Everything works fine so far and PARDISO is really a great solver for a huge class of PDEs. The declaration for the multiplication-subroutine in the MKL is as follow <PRE>[cpp]void mkl_dcsrgemv(char *transa, int *m, double *a, int *ia, int *ja, double *x, double *y);<BR />[/cpp]</PRE> Although nothing is affected by calling mkl_dsrgemv except *y, there are no const modifiers. A better declaration should be like <PRE>[cpp]void mkl_dcsrgemv(char *transa, int *m, const double *a, const int *ia, const int *ja, const double *x, double *y);<BR />[/cpp]</PRE> This would allow better and more correct coding, since there is no need to const-cast-away modifiers or declare friend classes. Debugging or proving correctness of code could be more easier. Same for the PARDISO-Interface, of course.<BR /><BR />Anyway, thanks for the nice tools!<BR />Fabian Mon, 09 Mar 2009 15:29:05 GMT Fabian_K 2009-03-09T15:29:05Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/const-correctness/m-p/884605#M9939 Hi,<BR /><BR />I wrote a modular PDE-solver in C++ which makes heavy use of PARDISO and CRS-matrix-vector-multiplication. Everything works fine so far and PARDISO is really a great solver for a huge class of PDEs. The declaration for the multiplication-subroutine in the MKL is as follow <PRE>[cpp]void mkl_dcsrgemv(char *transa, int *m, double *a, int *ia, int *ja, double *x, double *y);<BR />[/cpp]</PRE> Although nothing is affected by calling mkl_dsrgemv except *y, there are no const modifiers. A better declaration should be like <PRE>[cpp]void mkl_dcsrgemv(char *transa, int *m, const double *a, const int *ia, const int *ja, const double *x, double *y);<BR />[/cpp]</PRE> This would allow better and more correct coding, since there is no need to const-cast-away modifiers or declare friend classes. Debugging or proving correctness of code could be more easier. Same for the PARDISO-Interface, of course.<BR /><BR />Anyway, thanks for the nice tools!<BR />Fabian Mon, 09 Mar 2009 15:29:05 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/const-correctness/m-p/884605#M9939 Fabian_K 2009-03-09T15:29:05Z