Benjamin Uekermann

> Would it be possible to implement preCICE in single precision? Or that preCICE accepted data in single precision? The first suggestion would be quite involved I am afraid of....

I assume there is a motivating application example for this enhancement? Currently, I do not yet understand the user perspective. How would the first participant compute the time window size?...

Clearer now, thanks :+1: > after n= 3 timesteps, because it wants to be able to provide a cubic spline for , (again breaking the black-box to some degree) This...

> Calling before advance() sets the next timewindow size. There would be multiple `advance` calls within one time window. That's the difficulty here. > We can also leave this issue...

> I was close to creating a duplicate issue of this one since this feature was again requested at the preCICE workshop 2024. For what use case exactly? If it...

IMO it is a useful safeguard and helps documentation. Otherwise things like "only read-consistent mappings are supported for parallel participants" become even harder to explain. Remember that in preECO, we...

> Have you also sometimes considered adding connectivity of high-order elements (e.g., a triangle with midpoints)? Independent of this issue: So far, we have looked at midpoint as discretization-specific information....

Sounds interesting. Thanks for reporting. Could you please also upload the convergence and iteration files for the QN runs?

In this step of the configuration, the schema is built up, meaning what is allowed, basically the [configuration reference](https://precice.org/configuration-xml-reference.html). Only later, the actual config is parsed: https://github.com/precice/precice/blob/fc3fddf308652abb59799b5a949b91d7f9ec7c8e/src/cplscheme/config/CouplingSchemeConfiguration.hpp#L75-L79 That's where you...