Parameter changes before resimulation are ignored if kinsol is used as nonlinear solver

[RSchiemann]

I am building a faily large model and found simulation speed to be much better when I use kinsol as nonlinear solver.

However, when using kinsol, parameter changes before a resimulation of a previously completed simulation run are ignored. With the large model I am working with, I can reproduce this 100%.

The issue prevents productive usage of the kinsol solver, despite being much more performant. Could you please fix this for the upcoming version?

My setup: Win7 64bit
OMEdit v1.14.0-dev-4225-gacb265361 (64-bit)
Connected to v1.14.0-dev-212-ga2024ab5c (64-bit)
Connected to OMSimulator v2.1.0-dev-67-gc11f942-mingw

[RSchiemann]

The above happens although "Evaluate all parameters at compile time" is not activated, by the way.

[Francesco Casella]

Does this happen also with other models that you can share?

[RSchiemann]

Thank you for your prompt answer.

I will try to reproduce the problem next week with sharable models and then get back to you.

[RSchiemann]

Unfortunately, I am not able to reproduce the behavior with simpler models. I tried the following (both using kinsol as nonlinear solver):

1. a very simple stand alone model with one parameter and one variable. I was able to adjust the parameter value before resimulation and it took effect.
2. a slightly more complex model where a parameter of one component was defined by a parameter of another component (component.parameter1 = system.parameter2, but please note that in my case, the system parameters are not global/inner/outer variables, but just normal parameters without additional modifiers. I did this to represent model structure of the original model where adjusting parameters fails).

After these unsuccessful trials, I tried to reproduce the error again with the original, complex model, but was not able to either. I am a bit puzzled about this... My hardware and software setup (nor the model) have not changed since my initial report. Maybe there was some hiccup in communication between OMEdit and OMC, who knows.

So I guess, we can close this for now. If I encounter the issue again and if I can submit more useful information, we can re-open the issue.

[RSchiemann]

To clarify: Both scenarios (1 and 2) outlined above worked as expected.

[Francesco Casella]

Replying to RSchiemann:

So I guess, we can close this for now. If I encounter the issue again and if I can submit more useful information, we can re-open the issue.

OK.