Claw Dev Hackathon, March 2015

Clawpack Developers Meeting and Sprints

Thursday, March 19, 2015

University of Utah -- Leroy Cowels Building (LCB), Room 218

Map of Building

Bus from Salt Palace

The session will run roughly from 9:00am to 5:00pm.

This hackathon will take place following the SIAM CSE Conference. If you're at the conference, don't miss the Clawpack Minisymposterium.

Some of us may also be around on Friday, March 20, continuing to hack in a coffee shop or elsewhere.

Notes:

https://github.com/clawpack/clawpack/wiki/Notes-from-Claw-Dev-March-2015

Schedule:

Set some specific times for discussions that several people might want to participate in...

• 9:00 General discussion and planning the day

• 10:30 Visualization. 3D viz, refactoring VisClaw. Postprocessing, analysis, parameter sweeps, Pandas.

• 11:30 Installation: CEP #6, CEP #7, griddle.

• 1:00 Riemann solvers

  • Pointwise solvers. See possible mini-projects below.
  • Changes to interface? See CEPs below.

• 1:45 Third order methods

• 2:15 Time stepping

• 4:00 Release 5.3.0 and paper

Possible Topics:

• Visualization: visclaw issues, griddle, yt for 3d, etc. Paraview should be considered further.

• Riemann solvers: Further discussion of the enhancement proposals

• IPython notebook tools and Riemann solver notebooks

• Proposal to separate the guts of geometry, solution/state, IO, and plotting as an independent package, possibly outside the Clawpack namespace:

• Installation issues: Library vs. source code, pip install, -e option, path problems, relocatable install,

• CEP discussions

  • CEP #1: Develop pointwise Riemann solvers
  • CEP #2: Change Output File Naming Convention
  • CEP #3: Add rp_data derived type to Riemann solver function interface
  • CEP #4: Place all Riemann solvers in Fortran 90 modules
  • CEP #5: Add geometry to Riemann solver function interface
  • CEP #6: Install all Python code to site-packages
  • CEP #7: Run-Time and Compile-Time Option Handling for Fortran Code
  • CEP #8: Create a Controller to Manage I/O for All Packages
  • CEP #9: Create pure-Python library for geometry, state, I/O, and vis outside of Clawpack

• Identify documentation deficiencies

• Clawpack paper

Mini-projects that people might want to tackle...

• Pointwise Riemann solvers: Two or three Riemann solvers have been converted to pointwise form. Using these as a model, convert others.
• Develop a notebook template for quickly testing a Riemann solver.
• The rjleveque/3d_slices branch of amrclaw contains the first pass for 3d output of 2d slices (rather than the full 3d data set) so it can be viewed with 2d visclaw tools. See $CLAW/amrclaw/dev/advection_3d_inflow/README.txt on that branch for more details and suggestions for improvement.
• Explore state of mplot3d tools for plotting surfaces, e.g. for tsunami surface + topography as in this animation
• Explore vtk output format and write code to output from PyClaw, AMRClaw, etc. (ForestClaw and PetClaw have this capability.)

Please suggest more topics or mini-projects!

Google Earth Visualization

Here is a sample of the current state of our visualization project GeoClaw in GoogleEarth

• I have a list of several short projects (less than 10 lines of code!) for our Google Earth project on visualizing GeoClaw output. For example :

1. Setting up something so the user can specify a time of an event (either in UTC or local time) and pass that to Google Earth so it shows up correctly in the time slider.
2. Fix color map to more closely match the Google Earth colors
3. Add a legend
4. See more issues on the Clawpack page : Issues 125 to Issues 136