Thesis/paper for SU2 adapter / Adapter Class method

Hi! I am currently exploring some of my first projects using preCICE, and I am particularly interested in developing an adapter for a solver from scratch. My goal is to leave the solver’s source code unchanged.

Looking through the preCICE docs for adapter software engineering, I see that the two methods to develop an adapter are a) by creating a separate class for the the adapter or b) by using callback functionality and developing the adapter as a function object.

While I was able to find @Makis 's thesis for the OpenFOAM adapter (which was very helpful for the Callback Functionality method), I have struggled to find Alexander Rusch’s thesis on the SU2 adapter anywhere online. According to the preCICE docs, this thesis explains the Adapter Class method in detail, so it would be very helpful for my current project.

Hi @juanvalderrama

I just checked the preCICE website and the link to the thesis of Alexander Rusch is indeed outdated and no longer works. In any case, please be aware that the latest state of the SU2 adapter is using the Python wrapper of SU2, and the old design (by Alexander) is not longer applicable. If for some reason you still want to study the old design, I can try to get the thesis PDF for you. But I would encourage you to try and use the latest adapter.

Thanks @IshaanDesai! My goal is also to use the Python API of the solver I am attempting to build an adapter for, so I will definitely check out the latest version. Would you consider that the SU2 and FEniCS adapters follow a similar method of creating an Adapter class? I was able to find the paper for the FEniCS adapter and had shifted my focus to that since 1) it creates an adapter that does not modify the source code and 2) it is entirely Python-based

Hi @juanvalderrama,

nice that you found my thesis useful! :hugs: I am also figuring out how to fix the link to Alexander’s thesis for SU2.

The current SU2 adapter is directly using preCICE and the PySU2 interface to create a solver. See, for example:

So, I would say it is now mostly similar to the approach we follow for Nutils.

Lets discuss this in a separate post :slight_smile:

@juanvalderrama I uploaded the PDF directly into the new system: https://mediatum.ub.tum.de/1461810

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.