I do alot of stats/numerical stuff but I haven't tried gonum until now. Yesterday I went to port a bunch of C code that uses BLAS/LAPACK into Go, and thought I'd try it (Gonum) out. Now the logic is the original code is very hairy, and does delicate operations like a bunch of QR decompositions... in short, I don't want to mess with the algorithm at all, I just want to port it to Go. The C original code runs fine against OpenBLAS. Of course, being numeric code, it all assumes column-major Fortran style matrices.
But when I tried to run the same logic on Gonum, I was hitting issues left and right. At first I thought they were bugs in Gonum. But then I realized, by default Gonum is assuming that your matrices will be row-major (C-style). There's almost zero documentation for gonum of these kinds of assumptions and even less about how to use Gonum.... Like how do you switch between the Go and C BLAS implementation? Not documented; that I can find. You would think this would be front and center. Ugh. Anyway: somebody here probably knows-- does Gonum also support column-major matrices? In the Go BLAS, or in a cgo binding to OpenBLAS? With Bewildered Thanks. -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/d5586b79-0d3c-4a92-ba12-358a50a080ddn%40googlegroups.com.
