Wolfgang Bangerth
Wolfgang Bangerth
The same is true for the `second_requests` array, by the way.
How does grid 3 look like? And what constraints are missing? It's conceivable that we could build more constraints though I'd be surprised if that was *necessary* -- just because...
On 04/17/2018 10:39 AM, Timo Heister wrote: > Okay. Maybe I should look at that first. It still scares me though, > because the missing entries could chain to locally...
On 04/18/2018 11:08 AM, Timo Heister wrote: > The bad news: The example from the mailing list (mistake: assign > boundary ids manually but only for locally owned cells) is...
On 04/18/2018 04:51 PM, Timo Heister wrote: > One processor got an incorrect boundary id for a face of a ghost cell > (because set_boundary_id was only called for owned...
On 04/19/2018 11:14 AM, Timo Heister wrote: > So this one DoF is then locally active on > both processors, and they will have differing constraints for it > >...
@tjhei -- is this fixed by #6254?
I would love to know where these 1e-16 differences actually originate, but I imagine it to be quite difficult to figure this out. It is true that we could communicate...
@tjhei had a patch that at least detects this kind of problem in parallel.
I don't think this is going to make it for 9.4. Re-tagging for 9.5.