Replies: 2 comments
-
I worked on this a bit more, and made a bit of progress. I was misinterpreting the documentation: Thanks. |
Beta Was this translation helpful? Give feedback.
-
Sorry to keep replying to my own question. The problem (overly sparse SuiteSparse/CHOLMOD/Demo/cholmod_dl_demo.c Line 457 in 8ac3f51 B2 vector that contains exactly one non-zero value. And we then solve with and without Bset . We compare the results here: SuiteSparse/CHOLMOD/Demo/cholmod_dl_demo.c Line 501 in 8ac3f51 Xset . Outside Xset , the values of X1x should be 0, but they're not, and this is not checked.
I'm observing this with suitesparse 7.8.3 on Debian:
|
Beta Was this translation helpful? Give feedback.
-
Hi. I'm using
cholmod_solve2()
to solvePLx = b
(notAx = b
) many times with different vectorsb
. Eachb
is mostly 0, with a small number of non-zero entries. I have working code that specifiesb
as full, dense matrices, but this is very slow (I'm calingcholmod_solve2()
many times). Can I somehow specify a sparseb
and have this become fast? There'scholmod_spsolve()
, but that reconstructs the full denseb
. Andcholmod_solve2()
takes aBset
argument, but I think this applies to the outputx
, not the inputb
; right?I want to say the permutation
P
makes this difficult; is that right? Would this be any easier if I was asking aboutAx = b
?Thank you
Beta Was this translation helpful? Give feedback.
All reactions