Clawpack v5.10.0 released!

[Randall J LeVeque]

Dear Clawpack Users,

We are happy to announce that Clawpack 5.10.0 has been released.

See the release notes for a summary of all changes:
  http://www.clawpack.org/release_5_10_0.html
 
One important change of behavior in GeoClaw:  when solving equations on the sphere, a spherical source term in the mass equation is now included by default. This term was missing previously and so the computed results may change. See https://www.clawpack.org/sphere_source.html for more discussion and instructions for omitting this source term.

And more exciting news for GeoClaw users:  a major new enhancement, GeoClaw can now be used to solve dispersive Boussinesq-type equations using adaptive mesh refinement (AMR). The recommended equations are a modified version of the Serre-Green-Naghdi (SGN) equations. Like the shallow water equations (SWE) normally solved in GeoClaw, these are depth-averaged equations, but higher-order derivative terms are included to better match the dispersion relation of the linearized Airy wave theory.  These equations better model wave propagation in situations where the wavelength is not sufficiently long relative to the fluid depth for the SWE approximation to be accurate.  Applications include the study of tsunamis generated by landslides, asteroid impacts, or other localized phenomena.  Because of the higher-order derivatives, explicit numerical methods are no longer efficient.  In the approach we use, an elliptic system of equations must be solved at each AMR level on every timestep.  We use PETSc to solve the sparse linear systems using MPI (combined with continuing to use OpenMP to parallelize over grid patches for the explicit SWE portion of the code).  This adds new dependencies to the code.  For more discussion of the new capabilities and how to use them, see https://www.clawpack.org/bouss2d.html.

GeoClaw also now includes a new one-dimensional version, with solvers for both shallow water equations and Boussinesq-type equations.  AMR is not supported, but general mapped grids can be used and several examples are included.  See https://www.clawpack.org/geoclaw1d.html and https://www.clawpack.org/bouss1d.html.

How to get the latest version:
------------------------------

Please see
    http://www.clawpack.org/installing.html
for several installation options and the list of prerequisites.

A new tar file is available at:
    https://github.com/clawpack/clawpack/releases/
 
The tarfile is also permanently archived on Zenodo with DOI
10.5281/zenodo.10896214,
    https://doi.org/10.5281/zenodo.10896214
So if you publish results (and also, we hope, the code used to generate them) based on this version, you can rest assured that it will always be available to your readers.  For earlier versions, see
    http://www.clawpack.org/releases.html
   
Tar files for all recent releases can also now be referenced with the single DOI 10.17605/osf.io/kmw6h, convenient if you have used multiple versions in your work, or you want to cite a generic Clawpack DOI.   For more information on how best to cite Clawpack, please see
    http://www.clawpack.org/about.html#citing
 
Many thanks to everyone who contributed to this release, particularly Marsha Berger, who has been working with @rjleveque for the past several years on the Boussinesq capabilities, Jihwan Kim, who paved the way for this work with his BoussClaw software, and Barry Smith, who provided much help with PETSc and created the new version (now included in PETSc v3.20) that allows OpenMP code in combination with MPI.

You can view all the commits, diffs, and contributors starting from the release notes page
    http://www.clawpack.org/release_5_10_0.html.

Happy Computing,
  The Clawpack Development Team