Possibility of using C/Cpp

[tobiasfalk]

If I saw it correctly, this is built using C/Cpp, and I wanted to ask if it is possible to use it with C/Cpp?
And if no, if this is something that could be added.

[thliebig]

Can you define "use"? The python interface is a wrapper around the API of CSXCAD and openEMS and thus uses the C++ interface.
Of course other tools/software can use the C++ API too to be enabled to do FDTD simulations.

If with "use" you mean setup simulations like in python but using the C++ interface. Sure it would be possible. But I would really questioning your sanity. Because if you are capable of coding C++, learning python and be at least an order of magnitude more efficient would still be the better approach.

[tobiasfalk]

Yes, I understand that setting up a simulation in C++ is kind of insane.

Why I ask is because I wanted to find out the possibility to create a graphical UI. One with KiCAD integration in mind.

[thliebig]

Well my preferred way of creating a GUI these days is using python too, e.g. using PySide or PyQt. Because it is so much quicker to do... But in any case, yes the CSXCAD and openEMS C++ interface is intended to be used by third-party applications.

[montanaviking]

I find this post very interesting! I too am interested in directly using the C++ API. I use both Python and C++ for my software projects. So why would I be interested in the C++ API? Because I'd like to integrate OpenEMS in a C++ optimization loop using Pagmo2 C++ optimization library. I plan to throw a four-node EPYC box at electromagnetic design problems. Perhaps I'm missing some things here, but I believe that using the C++ API can be useful for integrating OpenEMS into a C++ project. Any ideas? Thanks

[dsa-t]

I've started some work at https://github.com/dsa-t/openEMS-Calc

Ported RCS_Sphere.py and MSL_Losses.m so far with results matching the references.

[biergaizi]

It's technically possible, but I believe it's misguided and pointless.

The openEMS design has separated frontend and backend. The frontend is CSXCAD, which is responsible for 3D modeling, the backend is the openEMS solver. To do anything in openEMS, you generate a .xml file that describes the problem. Basically everything relevant to your engineering problem is encoded in this .xml. The .xml file can be produced using any XML editor according to the CSXCAD specifications. This includes both the CSXCAD library itself, or independently (like how it's done in the Matlab/Octave binding).

Once you have a .xml, you can do two things:

1. Embed the simulation settings into the .xml as well, and just run the ./openEMS simulation.xml to execute it. This is how it's done in the Matlab/Octave binding.
2. Alternatively, openEMS itself exposes functions as shared library, so you can pass parameters directly to openEMS as a C++ library. This is how it's done in the Python binding. But this library is currently only meant for internal use, may change at any time, and there's no public API/ABI for third-party code. The only official API is the CSXCAD .xml file (and the Matlab/Octave and Python bindings themselves).

So what you probably want is using the CSXCAD library from C/C++, not using the openEMS library from C/C++.

So why would I be interested in the C++ API? Because I'd like to integrate OpenEMS in a C++ optimization loop using Pagmo2 C++ optimization library.

You can use C++ to change the CSXCAD .xml file. It's easier than creating an openEMS binding from scratch.

[montanaviking]

Thanks! I will likely look into this. I would like to be able to parameterize openems simulations within C++ if possible for a C++ optimization loop. Of course, C++ could just dynamically modify the Python script and call the Python interface and that might be easier than generating the .xml file. So it seems that as was stated earlier, the pure C++ approach would not add much value and having the intermediate C++ - generated Python scripts from the proposed optimization loop would also make the proposed system easier to debug.
Thanks!