- a bug of mesh creation in ECHOz1 and ECHOz2 is fixed;
- small changes in GUIs;
- the manual is updated.
The example with inconsistent results was found by Jean-Yves Raguin (PSI). It was a bug in the mesh creation. Following my debugging expierence with this example I have placed the following text in the manual.
“The codes use a conformal finite-difference method. In the time-domain wake field calculations the mesh resolution with 5 mesh points on the rms bunch length should be enough to obtain accurate results. To check the accuracy I would advice to do 2 tests:
- for the coarsest mesh resolution to change the bunch offset (with indirect integration algorithm switched on);
- double the mesh resolution.
If the first test fails it means that there is a meshing error. In this case change a little the mesh steps to try to avoid it and contact me for fixing a possible bug in the code. If the first test is OK then check the convergence and the accuracy with the second test.”