Hi,
It is easy using the csort(M,n) command but it only operates on one column.
Is there an easy way to sort on first two columns.
Also need to add if first two columns are identical ?
Thanks
Solved! Go to Solution.
I think there have been a couple of threads on this issue. Tom Gutman, one of the old cadre, wrote one using rank sort. Here's a thread in which @RichardJ and @LucMeekes give solutions (Richard's was an improvement on Tom's, and Luc's is different - AFAICT, just scanning the message. I haven't looked at the worksheets, but if they're by Tom, Richard, and Luc, the routines should be more than good enough for most purposes (Luc's handles larger matrices more reliably)
https://community.ptc.com/t5/Mathcad/Csort-is-it-true/m-p/706968/highlight/true#M193865
By an interesting coincidence, I'm in the process of updating an old complex-number mergesort worksheet of mine to MP10. I'm also fiddling around with it to see if I can get it to sort on functions ... no idea what the sorting criterion should be.
Stuart
I think there have been a couple of threads on this issue. Tom Gutman, one of the old cadre, wrote one using rank sort. Here's a thread in which @RichardJ and @LucMeekes give solutions (Richard's was an improvement on Tom's, and Luc's is different - AFAICT, just scanning the message. I haven't looked at the worksheets, but if they're by Tom, Richard, and Luc, the routines should be more than good enough for most purposes (Luc's handles larger matrices more reliably)
https://community.ptc.com/t5/Mathcad/Csort-is-it-true/m-p/706968/highlight/true#M193865
By an interesting coincidence, I'm in the process of updating an old complex-number mergesort worksheet of mine to MP10. I'm also fiddling around with it to see if I can get it to sort on functions ... no idea what the sorting criterion should be.
Stuart
x=0 and/or y=0 can treat. Bug fixed. Now P with 10,000 rows can treat by IJS(P) function.
Hi
It is being used for finite element analysis of large sparse matrices. The first two columns are the matrix indices. Last column is the value in the matrix.
Currently working on 46328x46328 matrix with 16,000,000 non zero elements. Each element is typically visited four times meaning the start matrix is nearly 48,000,000 x 3 with the largest value in first and second columns of 46327. The solver I have written in C++ using CUDA programming library cuDSS that is called by a Prime custom function.
Your file is very useful thank you.
Cheers
Terry