Sacha Duverger authored 8 months ago and Jerome Duriez committed 8 months ago

Improvements regarding dependencies versions handling and gmsh geometric tolerance-related possible pipeline issues (now solved)

Solve 1) ResultsReader issue when using A or T motion integration strategies 2) issue when using load_sim_ids 3) various small organisation aspect accross several scripts

Add support for new variables saved in CB-Geo MPM (matrices B, D and gV, with a computation of C from pycbg.postprocessing)

Tweak warnings: now warning formatting is only done when importing the MPMxDEM module, and add a new warning type NewFeatureWarning which is not displayed by default (if the user wants it, he should use warnings.simplefilter)

Correct bash batch script, there was an issue when several different batch ran (they shared the slot file...)

Replace the function  with the class , I still have to document it (and maybe improve it). The behaviour of the old function is still accessible with

Add the possibility to resume a batch from a specific simulation, or to change the executable used when launching the start_batch.sh script

Improve greatly the readability of the start_batch.sh bash script (by creating a template), as well as the setup_batch function, making some new features operational. I still have to do more testing

Add feature to setup_batch, to set the number of thread used for each simulation from the  parameter (I still need to add support for sequential run)

Change the order of mesh dict filling, because _get_boundaries_nodes_set has to be called before checking entity sets (there was an issue for simulations which didn't have any other entity sets than the ones for B_splines)

Improve the way the path of the pycbg script is checked when copying it to the simulation directory (now insensible to redundant slashes)

Remove the forced displaying of all warnings, now PyCBG has the default behaviour and the user facing an issue should reduce warning filtering himself

Tweak the particle filter feature, because the filter function cannot be pickled (had to take it off the attributes)

Add the possibility to use a filter function when creating the material points (I still have to comment it)

Sacha Duverger authored 1 year ago and Jerome Duriez committed 1 year ago

Add an automatic material point generation that uses Gauss-Legendre integration points locations from a Python side