Bruno, This temporary fix works. I was unable to create a small example to illustrate the problem with the detail I desired, so unfortunately cannot share this. The workflow that resolved this issue is as follows: 1. generate coarse mesh 2. load checkpoint 3. reinitialize and prepare associated data structures now that the triangulation has been set. 4. load data.
Since the mesh I am loading the data into is coarse, this re-generation of the mesh is reasonable. For me, the key observation was that I need to also reinitialize some associated data structures once the triangulation has been loaded. Thanks all. On Thursday, February 20, 2025 at 6:43:17 PM UTC-7 blais...@gmail.com wrote: > I think a temporary fix could be to > generate_coarse_mesh in the triangulation > load checkpoint > load data > > And then for each step you want to load, to > clear > generate_coarse_mesh > load checkpoint > load data > > Which when thinking about it looks a bit absurd. If you have a small > example i'd be happy to look at it! > > On Wednesday, February 19, 2025 at 1:02:26 p.m. UTC-5 jere...@gmail.com > wrote: > >> triangulation.clear() removes all data. Apparently, load() requires that >> triangulation already contain a coarse mesh. Therefore clear() cannot work. >> I'll write a small program that demonstrates the issue today. >> >> Would it be to add a line in the implementation of load() which clears >> the buffer? I would argue that if a user calls load(), they intend to load >> different data into that buffer. I still don't understand the structure of >> the triangulation class entirely, so maybe there is some reason that the >> authors want to prevent a data buffer from being used before it is >> 'released'. >> >> But even that word is confusing, since there is no indication that the >> triangulation class has any way to tell the data serializer to 'release' >> the buffer's data. Note that CellAttachedDataSerializer<dim, spacedim> has >> a clear() method which would do what I want. >> >> -Jerett >> >> On Monday, February 17, 2025 at 7:09:44 PM UTC-7 Wolfgang Bangerth wrote: >> >>> >>> Nice to see you here on this forum, Jerett! :-) >>> >>> > I think that the checkpointing in deal.II probably assumes this >>> process only >>> > happens once, so I may be exceeding the design of the Triangulation >>> class. >>> >>> Quite possibly. What happens if you call triangulation.clear() between >>> calls? >>> >>> If that doesn't help, can you create a small program that is (i) >>> self-contained, (ii) demonstrates the problem? >>> >>> Best >>> W. >>> -- >>> ------------------------------------------------------------------------ >>> Wolfgang Bangerth email: bang...@colostate.edu >>> www: http://www.math.colostate.edu/~bangerth/ >>> >>> >>> -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/dealii/ef2d2ec0-f0c0-4ae1-95c8-81427fc7622dn%40googlegroups.com.
