Skip to content

Microstructure v2 niac changes#1

Open
mkuehbach wants to merge 50 commits into
developfrom
microstructure_v2_niac_changes
Open

Microstructure v2 niac changes#1
mkuehbach wants to merge 50 commits into
developfrom
microstructure_v2_niac_changes

Conversation

@mkuehbach
Copy link
Copy Markdown
Collaborator

No description provided.

atomprobe-tc and others added 30 commits May 16, 2024 12:51
…ttings for datasets, for attributes the code changes apply in almost the same way but have not been implemented here as currently not needed, default formatting is variable, nullterminated, utf8 to match the default of h5py
…ion and check if runs performantly, 3, implicit gridifying working again, next-up: i) ipf, ii) microstructure, iii) odf, iv) pf
…ion and check if runs performantly, 3, implicit gridifying working again, ipf color keys and plotting working again, strcat removal remains in this function
…se using vector3d.X projection and using precomputed IPF legends to not compute them for every of the approx. 4.5k EBSD maps with 4-10 phases each
…, oftentimes the colormaps get chopped off by the matlab renderer, this bug is next up, using calcGrains on Ubuntu 24.04 uses compiled jcvoronoi which demands glibcxx 3.4.32 but this is not available on 24.04 :(
…y, is it possible that on Linux the graphics rendering works different because of which occasionally IPF legends are chopped?, fixed NXdata bug
…es that Mtex reports pre and suffixed with 1
mkuehbach and others added 20 commits April 10, 2025 00:55
…ng calcGrains also on Linux and Mac by using fallback to quickhull
…used to substantiate arguments on future work for the scidat_nomad_em as the EBSD datasets from for the database are virtually all grain2d and ebsd2d
…ed that will be used in the batch queue via the ipf_lgds.mat look-up table
…mensionality was passed when computing grains.area, removed debug output in hdf5 wrapper, added y-flipping of precomputed ipf maps to such that for visualization with h5web we do not need to demand that OIM maps are correctly oriented but ipf legends are consistently flipped vertically
…document which EBSD maps were collected using square or hexagon grid
…est clearing of variables, ii) reduce resolution for ODF (1deg, float, if not even uint8), iii) check that everything is available for the microstructure, iv) run-through
…ence between pixel and polygon area, polishing before the batch queue run
… cleaning of the microstructure reporting part of the code for the run throughs
…h off reporting of the microstructure for many cases as processing this takes too long, i) replace containers.Map with dictionary, ii) evaluate whether misori loop nest can be rewritten to run faster, it is currently THE bottleneck
… at least one if not two orders of magnitude too slow, specifically i) compiled jcvoronoi for the target system replacing the sequential and slower qhull, ii) rewrote code to get per interface patch misorientations
…ly the computation of volume fractions from ODFs was oftentimes facing a bug, other than this, the queue run through fine, the processing of a few datasets was postponed to bundle them together, next steps, i) evaluate results and run remaining files that will not be bundled together but for which problems occurred specifically with the volume fraction computations, rerun them with having the code adjusted such that already computed partial results are kept to save computing time
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants