A few minor bugfixes (Thanks Wes)
iSALE3D: Force all processes to dump at the same time. Previously, processes could get out of sync, now we use the master process to determine if a dump is needed, then broadcast that to all other processes.
If dump-interval is defined as a wall-time in iSALE2D, number the dump files sequentially from 1 instead of always using XXXXX
When using walltime to choose the interval to write dumpfiles (TDUMP<=0), force iSALE to dump at the next savetime, rather than at the time exactly. This makes restarting models nicer, since the model will be restarted at a time saved in the datafile.
pySALEPlot: bugfix, solves a problem when opening a model in interactive mode with scale='um'
iSALE3D bugfix: When restarting a 3D model with tracers run in parallel, a segfault often happened. This was caused when the number of tracers in a domain changed, as the array tracerini was not re-allocated with the correct number of tracers (instead, was allocated based on how many tracers were in the domain at the beginning of the calculation). iSALE also then didn't know the tracers IDs stored in each domain after the restart. Now, tracerini is written to the dumpfile, and then also read in (and allocated properly) upon restart. ...
pySALEPlot: simplified 3D_slice_example using the new tracerGrid function from r1986
pySALEPlot: Added model.tracerGrid function, to create a list of tracer indices that make up a grid from (x,y,z) lines, so you don't need to loop over the lines and plot each one individually (if you don't connect the tracers). Can also define offsets in case tracer lines are not aligned between tracer units. See documentation string for how to use.
removing my fix from r1982 for now, since it causes problems with gfortran <= 4.8.x. To get rid of the IEEE messages in the future, we need a way to add the fortran flag '-ffpe-summary=zero,invalid', but only for gfortran >= 4.9.
pySALEPlot: Added step.findTracerCell, which given the tracer ID number, returns the i,j(,k) indices of the cell containing the tracer at the given timestep
View all revisions
Also available in: Atom