Open dimpase opened 6 years ago
Hi dimpase! thanks for using SDPA-GMP 1) Unfortunately, not. 2) I've never tried. I tried over 100 decimal digits, and it was okay
Oh well. In SPDs I have (very badly conditioned) I end up with errors in Cholesky factorisation saying something like "matrix ix not positive semidefinite".
Could it be that some precision is checked using the doubles rather than arbitrary precision?
Hi dimpase, Yes, you can control using 'precision' parameter in param.sdpa .
No matter how big I make it, it seems not being enough. Perhaps I should really try huge values...
Hi dimpase, could you please send me your logs to maho@riken.jp? thanks Nakata Maho
I will try to reproduce this on the hardware I have with me (I am on vacation, and do not have access to the machine I used to compute these). It might take a bit of time.
Is it possible to accept exact rational input?
If not, is there any limit on the size of integers one can have in
.dat
files?