Difference between revisions of "Adding Llama to the ET"
(→Examples) |
(→Examples) |
||
Line 35: | Line 35: | ||
These run in about 20 minutes on a laptop, and evolve a wave on a given background. These have been tested and work. | These run in about 20 minutes on a laptop, and evolve a wave on a given background. These have been tested and work. | ||
− | There is a new [ | + | There is a new [https://einsteintoolkit.org/about/gallery/mpwave/ Multipatch Wave Equation gallery page] based on the Kerr-Schild_Multipole example (not yet linked to the main gallery page). |
− | There is also the [ | + | There is also the [https://einsteintoolkit.org/about/gallery/gw150914/ GW150914 gallery page]. |
Revision as of 09:04, 21 November 2016
In order to add Llama to the ET, we require:
- Tests
- Documentation
- Example(s)
The Llama infrastructure has been used in many published papers, so is clearly considered robust enough to produce correct scientific results.
Tests
We have the following test cases:
- Llama/Coordinates/test/test7patch: Sets up the 7-patch system. Checks coordinates. Makes sure that this doesn't crash. No evolution.
- Llama/LlamaWaveToy/test/llamawavetoy_6patch.par: Sets up 6-patch (spherical) system. Evolves l=2,m=2 x Gaussian for 4 iterations. Checks 1D ASCII output. Runs on any number of processes (only 1 and 2 have been tested).
- Llama/LlamaWaveToy/test/llamawavetoy_7patch.par: Same as 6-patch but uses 7-patch system. Works on 1 process. Currently has a problem on 2 processes; looks like a bug. Assertion failure from Carpet. Restricted to run on 1 process as a result.
- McLachlan/ML_BSSN_Test/test/ML_BSSN_MP_O8_bh.par: Runs a few iterations of BBH initial data using 8th order McLachlan on 7-patch (spherical+Cartesian) system. Currently fails on my laptop with tolerance errors.
Documentation
The following thorns have documentation:
- Coordinates: Provides an introduction to the Llama multipatch infrastructure and the different patch systems
- GlobalDerivative: Describes the multipatch-aware finite differencing implementation
- Interpolate2: Interpatch interpolation
- LlamaWaveToy: A scalar wave equation on various backgrounds.
- WaveExtractL: Llama-optimised wave extraction using Zerilli-Moncrief and Regge-Wheeler gauge invariant perturbations of Schwarzschild
The documentation is probably not all up-to-date; there are comments in some of the files that the description does not match what is implemented.
Examples
There are three examples in LlamaWaveToy/par:
- Kerr-Schild_Gaussian.par: Radial Gaussian evolved with wave equation on Kerr-Schild background. Uses 6-patch (spherical) system.
- Kerr-Schild_Multipole.par: l=2, m=2 mode with radial Gaussian evolved on a Kerr-Schild background. Uses 6-patch (spherical) system.
- Minkowski_Multipole.par: l=2, m=2 mode with radial Gaussian evolved on a flat background. Uses 7-patch (spherical+Cartesian) system.
These run in about 20 minutes on a laptop, and evolve a wave on a given background. These have been tested and work.
There is a new Multipatch Wave Equation gallery page based on the Kerr-Schild_Multipole example (not yet linked to the main gallery page).
There is also the GW150914 gallery page.