ThummeTo comments

Results 103 comments of


                                            ThummeTo

Contrary loss update direction of CS-FMU and changeless loss of ME-FMU with FMUParameterRegistrator

There are some tests in FMISensitivity.jl that you can check out: [FMISensitivity tests](https://github.com/ThummeTo/FMISensitivity.jl/blob/main/test/jacobians_gradients.jl) (here, we compare many gradients to ground truth gradients)

Eigen with complex output

Thanks for the info. I would totally agree on switching to an implementation with complex output, but it should work with at least ForwardDiff, ReverseDiff and Zygote. Can you please...

FMU evaluation with dx_refs does not work correctly (incorrect values returned)

Exactly, FMI-standard itself has no feature for this, and FMI.jl will provide a feature for that (in the near future). It's only a little thing to implement it so we...

Wrong calling Sequence of Cosimulation-FMU with Inputs

Correct, will be fixed!

FMU Explorer Pluto example has incompatible syntax with version of FMI package referenced

Should be fixed, please reopen if not.

Use sparse jacobians when part of the FMU

We already have some code on that topic, and try to include this soon :)

Reading modelDescription (during loadFMU) fails when array variables are used.

this throws a warning now in #132 but parsing is not supported yet. However, implementing arrays/vectors/tensors requires additional changes to FMICore. We should do that as part of a dedicated...

Reading modelDescription (during loadFMU) fails when array variables are used.

[see FMI.jl issue](https://github.com/ThummeTo/FMI.jl/issues/257)

Function must be called in mode continuous time

Good, that you already have a hot trace! A good FMU prints a helpful error message, when returning `fmi2StateError`. You can enable displaying logs from the FMU [as explained here](https://thummeto.github.io/FMI.jl/dev/features/#Debugging-/-Logging)...

Use GC Preserve when passing pointers to ccall

Is this actually an issue with FMICore? Or isn't it more like a general thing when using pointers in Julia? One could argue that if the user uses pointers directly...