Different result for Autotuning example

[ZYY0844]

Hi,

When I try autotuning.ipynb, I got the result like

which is different from the result in the example at Out[41].

I tried to change the initial state to a non-zero vector to achieve a similar response

May I ask why the example doesn't output the same result?

[baggepinnen]

I guess there can be any number of things that have changed in the years since that example was run. My guess would be that the diff eq solver behaves slightly differently or somthing, causing convergence to a different point.

Note that to make the example more practical, it would require some robustness constraint and control signal penalty as well. Some of this is done in the notebook
https://github.com/JuliaControl/ControlExamples.jl/blob/master/gain_scheduling.ipynb

[amburosesekar]

To solve similar problem in MATLAB, we use "rng default"
its similar to seed based random number generator. My guess if you use similar command in Julia and apply optimization gives always same result. thank you

[baggepinnen]

Setting the random seed is good for reproducibility in general, but I don't think that's the problem here, there is nothing random in this example. The difference is probably because some library has changed slightly over the years.