v5.3.1 release notes¶
Clawpack 5.3.1 was released on November 10, 2015. See Installing Clawpack.
Changes relative to Clawpack 5.3.0 (May 21, 2015) are shown below.
Changes to clawutil¶
Allow checkpt_style < 0 in setrun.py, see Changes to amrclaw below.
New testing framework.
See clawutil diffs
Changes to visclaw¶
Added amr_data_show attribute to ClawPlotItem to suppress certain amr levels in plotting.
Read gauge locations from setgauge.data file found in plotdata.outdir (the output directory where fort.gauge is located) rather than from datadir. This works better if gauges are changed for new runs but you want to plot old output, since the correct setgauge.data input for a run is copied to the output directory at the start of the run.
Fix colormaps.add_colormap function to work better for combining two different colormaps for different ranges of a variable in certain cases.
See visclaw diffs
Changes to riemann¶
Added src/riemann_interactive.py for making interactive plots of the phase plane in Jupyter notebooks. For an example, see http://nbviewer.ipython.org/github/maojrs/ipynotebooks/blob/master/interactive_test.ipynb.
See riemann diffs
Changes to amrclaw¶
Bug in 3d dimensional splitting fixed — the CFL was computed incorrectly, causing the code to use timesteps that are too large and go unstable in some cases.
Improved the checkpoint and restart capabilities, adding the option to set checkpt_style < 0 in setrun.py if you want the checkpoint to alternate between two files rather than creating a unique new file each time a checkpoint is done. This reduces storage needed if you want to checkpoint frequently but only need the most recent one, e.g. to guard against crashes in a long run. Two files are used in case the code crashes in the middle of doing a checkpoint, the previous one is still intact. You can set checkpt_style = -2, for example, to give the same behavior as checkpt_style = 2 but saving only two latest checkpoint files.
With this feature turned on, instead of files with names like fort.chk00100 and fort.tck00100 being created for a checkpoint after 100 steps, the files are named either fort.chkaaaaa, fort.tckaaaaa or fort.chkbbbbb, fort.tckbbbbb and these names alternate. (The fort.tck files are very small with information about the time of checkpointing, the solution data is all in the fort.chk file.)
Also improved checkpointing so that the output files fort.amr and fort.gauge have buffers flushed whenever checkpointing, to avoid possibly losing some gauge output if the code crashes. Now fort.gauge will have data at least up to the last checkpoint time.
Fixed so that advanc is not called initially if flag_richardson is False, avoiding some unneeded work in this case.
Better reporting of statistics regarding run time and number of cells integrated is now provided to the screen at the end of a run, and to the file _output/fort.amr. In particular, better reports wall time vs. CPU time when OpenMP is used, and breaks this down into several groups to help determine where the code is spending the most time. See Timing Statistics.
New testing framework and several changes in the tests directory.
See amrclaw diffs
Changes to geoclaw¶
NetCDF format can now be used for topography files by specifying topo_type = 4, both in topotools.py and when reading into the Fortran modeling code.
Binary output option added for multi-layer code.
Bug fixes in topotools and dtopotools.
Better reporting of statistics regarding run time and number of cells integrated is now provided to the screen at the end of a run, and to the file _output/fort.amr. In particular, better reports wall time vs. CPU time when OpenMP is used, and breaks this down into several groups to help determine where the code is spending the most time.
Restart option changed so that fort.gauge is appended to rather than clobbering the previous version. This may be modified in future releases, it is currently inconsistent with what’s done in amrclaw. See https://github.com/clawpack/clawutil/issues/93.
New testing framework.
See geoclaw diffs
