v5.4.0 release notes

Clawpack 5.4.0 was released on January 17, 2017. See Installing Clawpack.

Changes relative to Clawpack 5.3.1 (November 9, 2015) are shown below.

Global changes

Python 3 compatibility. Python code in all repositories has been updated so that it should work with either Python 2 or Python 3. In particular the statements:

from __future__ import absolute_import
from __future__ import print_function

have been added and print statements have been replaced by print functions. Various other minor changes were also required.

Makefile structure for Fortran codes. The Makefile in all Fortran examples and tests has been modified to rely on a common list of library source code files, rather than listing all of these in every Makefile. Capabilites include the ability to specify whether a library file should be replaced by one from the local directory. This is cleaner and will make it easier to update code in the future – if a new library routine is required only one master list needs updating rather than the Makefile in every example and users’ application directories. See Library routines in Makefiles for more details on how to specify local files in place of default library files.

It is also no longer necessary to set the Makefile variable RESTART to True or False. Instead you can set it to None (or omit setting it at all, since this is the default), in which case the setrun.py file will be used to determine if this is a restart run (in which case the previous output directory should be added to, rather than replaced).

Improved Gauge Output Options Gauges in amrclaw and geoclaw now support a number of additional output options including:

  • specification of output fields, i.e. you can now specify the q and aux fields that are output;

  • specification of output field format, i.e. you can now specify the number of digits to output;

  • a minimum length of time at which a gauge is allowed to output, i.e. if this was set to 10 time units then the gauge would only output every 10 time units or longer;

  • support for future file format specifications (only ASCII is supported now);

Other improvements to gauge handling include:

  • a refactor of how the code stores gauge data has been done in the Fortran gauges_module.f90 source file in each library.

  • Gauge output is accumulated in a buffer internally and written out intermitently, instead of writing to disk every time step. (The parameter MAX_BUFFER in the amrclaw library routines gauges_module.f90 controls the size of this buffer.)

  • The gauge output for the gauges is written to distinct files in the output directory, e.g. gauge00001.txt for gauge number 1. In previous versions of Clawpack all gauges were written to a single file fort.gauge. The new approach allows gauges to be written in parallel and also facilitates reading in a single gauge more quickly.

  • Some header info appears in each of these files to describe the gauge output.

  • When doing a restart (see Checkpointing and restarting), gauge output from the original run is no longer overwritten by the second run. Instead gauge output from the restart run will be appended to the end of each gaugeXXXXX.txt file.

Updated regression testing framework for Fortran. The Fortran code uses an updated framework and so the regression data has changed.

Changes to classic

Makefile structure. See discussion above, under Global changes.

See classic diffs

Changes to clawutil

Makefile structure. The Makefile.common was updated to support the changes described in the discussion above, under Global changes.

Better support for gauges. New supporting code added.

Updated regression testing framework for Fortran. New supporting code added.

See clawutil diffs

Changes to visclaw

Parallel Plotting in setplot.py. A new capability has been added to plot multiple frames at once on a multicore machine when doing make plots (i.e. not interactive). The png files for different frames can be simultaneously generated. To use this feature you need to:

  • Add the line plotdata.parallel = True (usually at the bottom) to setplot.py.

and then either:

  • Add the line plotdata.num_procs = 4 (or however many processes you wish to use), or

  • Alternatively you can set the shell environment variable OMP_NUM_THREADS to the number of processes desired.

The value specified by OMP_NUM_THREADS is used only if plotdata.num_procs is not set. If neither is set, the default is to use only one process.

Gauge plots. Updates to go with improvements to how gauges are handled.

KML files for GeoClaw output. Some improvements have been made to the capabilities for creating KML and KMZ files for plotting on Google Earth or with other GIS tools.

See visclaw diffs

Changes to riemann

GeoClaw Riemann solver. The Riemann solver generally used in GeoClaw has been updated to fix a couple issues:

  • The transverse velocity jump is now put into the 1-wave or 3-wave rather than the 2-wave. This avoids some cases where transverse velocity does not propagate past jump in bathymetry, may improve some instability issues. See https://github.com/clawpack/riemann/pull/111 for details.

  • The tolerance used in the transonic test has been modified to be better scaled.

These changes cause some changes to results computed with GeoClaw. They have been fairly extensively tested by now and give results that are generally believed to be at least as good or better than the previous version.

Some other solvers were added or updated.

See riemann diffs

Changes to amrclaw

Makefile structure. See discussion above, under Global changes.

Gauge output See discussion above, under Global changes.

Ghost Cell (filpatch) Filling. A list of the neighboring grids at same the level of refinement that are used for filling ghost cells for each grid patch is saved between regridding steps. This improves the speed of filpatch operations. (Not yet implemented for neighboring grids at coarser level, still have to search for neighbors.)

Proper Nesting. Insidious but rare bug fixed, where occasionally a fine level grid had cells with no underlying coarse grid cell from which to interpolate the new values. The fix can make regridding more expensive when more than 3 levels of refinement are used. (This will be addressed in future revisions). Also, there were several different ways of projecting a cell to a coarser level. This was made consistent across all routines. The refined grids that are generated are now somewhat different and may cover a slightly larger area than in previous releases.

3D filpatch bug fix. Fixed a bug in calculating indices used when interpolating from coarse to fine grid ghost cells. (Fixed in 2D in previous release.)

Output Formats. Enlarged formats in many format statements used for ascii output throughout.

See amrclaw diffs

Changes to geoclaw

Changes to Riemann solver. The default Riemann solver used for single-layer shallow water equations was modified, causing potential changes to computed results. See the discussion above, under Changes to riemann.

Makefile structure. See discussion above, under Global changes.

Gauge output See discussion above, under Global changes.

The changes in amrclaw titled Ghost Cell (filpatch) Filling, Proper Nesting and Output Formats also affect geoclaw. See notes above.

fgmax Checkpoint/Restart Capability. If checkpoints have been requested, fgmax variables are added to the end of the checkpoint file. This enables a calculation to restart for a longer simulation time and still compute valid fgmax amplitudes and arrival times, instead of reinitializing the fgmax arrays. See Fixed grid monitoring.

See geoclaw diffs

Changes to PyClaw

Python 3 compatibility. See discussion above, under Global changes.

For changes in PyClaw, see the PyClaw changelog.

See pyclaw diffs