Benjamin Uekermann
Benjamin Uekermann
Yes, we also discussed that. I guess it should be possible. * I agree on your observation about "the need to write initial data in the SolverInterface Ctor" above. *...
closely connected to / kind of a duplicate of #1184
> each data set receives its own Aitken factor This will numerically not work as the data values then become inconsistent with respect to each other. We should stack the...
> I guess stacking data would require some kind of rescaling, right? Yes, could / should use the same rescaling as for the QN methods.
Could this be a building problem? Two conflicting Python versions? I just tried to use Python action in the OpenFOAM-Nutils tutorial on the OpenFOAM side and it works for me....
@mari-shka Let's look a bit closer. If you configure preCICE with cmake what is the screen output of ``` -- Found PythonLibs: XX -- Found PythonInterp: XX -- Found NumPy:...
It's a bit mysterious, yes. For completion: For me OpenFOAM 5 + Python 3.6.9 works.
Maybe a warning would be sufficient here. Let's say you want to do a parameter study on what's the influence of `timesteps-reused` (we have done that before). Then, you might...
Small detail: So far, we mostly used "timestep" in preCICE, not "time step" (in contrast to "time window"), meaning it should be `scale-by-timestep-size`.
Thanks for the heads-up summary :+1: * I agree on the necessary additional synchronization points during a current `cplscheme.advance()`. I can't oversee the compositional scheme yet, but for simple serial...