Commit Graph

1703 Commits

Author SHA1 Message Date
Martin Diehl 513f1c6726 write out point data
might be possible to use a rectilinear grid for this also in the case of
grid solvers
2019-10-12 12:34:37 +02:00
Martin Diehl 8793050329 fixed handling of subfolders
now inline with post results.
Still a little bit unclear what happens if an absolute directory is
given. Probably best to either specify absolute or relative with respect
to working directory
2019-09-19 16:10:07 -07:00
Martin Diehl 4cedcee0b4 handling of increments follows handling of other 'visible'-items 2019-09-15 17:08:03 -07:00
Martin Diehl f6ac8c995f simplified and selected better names 2019-09-15 16:00:19 -07:00
Martin Diehl 04aefa84bc more appropriate names 2019-09-15 15:02:16 -07:00
Martin Diehl de316f1afe Merge remote-tracking branch 'origin/development' into even-more-HDF5-postprocessing 2019-09-14 12:48:42 -07:00
Martin Diehl b85ac11c49 implemented ideas from discussion with Philip
group matching unified and with wildcard support
time step handling should become more convenient (WIP)
add_norm can not compute abs of scalar, added function for that
general polishing here and there
2019-09-14 09:07:23 -07:00
Martin Diehl c13db4b3ca renaming
on_air/active are not clear. visible seems to be the most appropriate
name
2019-09-13 21:49:22 -07:00
Martin Diehl a6567e0cc6 safer to use functions
don't mess with attributes of the object, they might be renamed
2019-09-13 21:41:35 -07:00
Martin Diehl 002383afc2 solved problem with postprocessing
- to not 'try' with h5py library, it might have another 'try'. Check
explicitly for empty argument

also some polishing
2019-09-13 16:01:30 -07:00
Martin Diehl 52904a81dc taking prospector complaints serious 2019-09-13 06:49:25 -07:00
Martin Diehl a3c6f6682e using eigh should ensure orthogonal eigenvectors 2019-09-13 06:42:01 -07:00
Martin Diehl 890a6cf42f avoid floating point comparison and simply play it safe 2019-09-13 06:40:41 -07:00
Martin Diehl 0497b58629 use existing functionality 2019-09-13 06:38:22 -07:00
Martin Diehl cd6a4d1cfd that's all we need 2019-09-13 06:37:47 -07:00
Martin Diehl 66d8a3e601 stretch is symmetric (play it safe here) 2019-09-13 06:35:48 -07:00
Martin Diehl 79d2432c6c R not needed 2019-09-13 06:33:00 -07:00
Martin Diehl 49b3cd2145 transpose evuivalent for rot but faster 2019-09-13 06:32:05 -07:00
Martin Diehl 462bb07efd Merge branch 'development' into even-more-HDF5-postprocessing 2019-09-11 16:29:44 -07:00
Philip Eisenlohr 926217fcb6 added possibility to reduce rotation to specific fundamental zone, and to output 3-dim Rodrigues vector 2019-09-10 15:09:27 -04:00
Martin Diehl d446248d75 also ouptut materialpoint results if requested 2019-07-15 17:07:04 -07:00
Martin Diehl b6bb8a4650 simple strain calculation for DADF5
further enhancement requires to give optional arguments to
add_genericpontwise
2019-07-06 21:41:29 -07:00
Martin Diehl a8ed982706 [skip ci] better name 2019-06-15 07:31:30 +02:00
Martin Diehl eb8be13203 missing import 2019-06-14 17:27:00 +02:00
Martin Diehl 1cf9c59850 standardizing
import follows PEP style guide, encoding not needed for python3
2019-06-14 13:03:30 +02:00
Martin Diehl 4ff3efb1a5 to 'stack' tables 2019-06-13 23:14:40 +02:00
Philip Eisenlohr bffeaa980d restored Martin's logic, not clear why seemed to be broken with PE's former tests... 2019-06-03 14:43:23 -04:00
Philip Eisenlohr 8d52a3d52a polishing of geom_rescale; adopted modified testing reference 2019-06-03 14:15:23 -04:00
Philip Eisenlohr 2870fdb25b fixed buggy version 2019-06-03 13:39:45 -04:00
Philip Eisenlohr 89e4bc800c removed default size to adjust to changes in geom-class 2019-06-03 13:38:48 -04:00
Philip Eisenlohr 60031cc806 changed periodic option to reflect to better reflect what happens 2019-06-03 13:35:37 -04:00
Philip Eisenlohr 7ea417d585 rotated geometry maintains center of gravity 2019-06-03 13:30:14 -04:00
Martin Diehl 615ff4ed06 [skip ci] vectorized 2019-05-31 23:26:32 +02:00
Martin Diehl bf27505faa Merge branch 'development' into geometry-class 2019-05-31 16:21:50 +02:00
Martin Diehl 02671c5c03 bugfix
shoud now work for any possible combination of new and old grid and
offset
2019-05-31 15:58:47 +02:00
Martin Diehl 979145f023 vectorized 2019-05-31 10:27:26 +02:00
Martin Diehl 46f0ad052e direct support for vtk output
- geom_check can now handle multiple files
- microstructure index is stored as integer in vtk file
2019-05-30 15:35:45 +02:00
Martin Diehl 08052737cf if grid is an option, it is a tuple
tuples do not have the prod() function
2019-05-30 14:28:31 +02:00
Martin Diehl c2a5f37818 round to reasonable precision
without symmetry, there are 36000**3 = 46656000000000 orientations
2019-05-30 14:11:54 +02:00
Martin Diehl bf60cffab5 cleaning 2019-05-30 14:07:49 +02:00
Martin Diehl d0c7f8b934 texture is typically the more interesting information
therefore, but it to the top
2019-05-30 14:05:45 +02:00
Martin Diehl 8380071564 using geom class
- dropped the use of the material class as it does not allow to have
empty phase and crystallite. The material.config class needs a severe
refurbishing (no crystallite, use orderedDict instead of hand written
solution, drop of texture components). Probably best solution is to
postpone this until we have a yaml/json reader for Fortran.
2019-05-30 13:30:38 +02:00
Martin Diehl 7d07a3752f polishing 2019-05-30 13:30:18 +02:00
Martin Diehl 461611cde7 polishing 2019-05-30 11:26:47 +02:00
Martin Diehl 49c3903cc7 explicit loop for simpler code
- no invalid combintations any more (allow to move the original
microstructure out of the new canvas)
- handling of origin needs discussion. When we want to recover the old
behavior, the origin might be located outside of the new geometry
2019-05-30 11:26:07 +02:00
Martin Diehl e974bb9c43 similar structure for building blocks 2019-05-30 10:45:17 +02:00
Martin Diehl 510b55df50 several improvements
- more error checking
- order of points can be arbitrary also for microstructure input
- grouping of same orientation/phase combinations
2019-05-30 10:33:01 +02:00
Martin Diehl 931fa9d36a standardized and cleaned
-write to file if given
-type casting is done in the geom class
2019-05-30 10:20:07 +02:00
Martin Diehl 07276d511c modern python formatting
- 2 digits should be precise enough for euler angles
- no need to pad grain ID with leading zeros
2019-05-30 10:03:55 +02:00
Martin Diehl 73df615ff3 numpy meshgrid can do this now 2019-05-30 09:45:30 +02:00