yalmip + ipopt: check robustness to initialization

[Heman Shamachurn]

Hello Professor,

In yalmip, I am using optimizer + ipopt. Everything is working fine. I now want to check the robustness to initialization.

Before the optimizer is run, I see that my decision variables are all initially set to 'NaN'. Yet, after the first optimization, the decision variables get their desired values. I am not too sure about the overall software architecture, but I want to  set my decision variables to some initial values, and then see the results. Is that possible please?

I have tried sending initial values for the decision variables when calling the optimizer. However, the optimizer fails because of NaN values. 

Grateful for your advice/guidance.

Thank you.

Heman

[Johan Löfberg]

No, at the moment this is not possible as the model is severely reformulated inside optimizer (variables are eliminated etc) and the map from initials to new variables is not done.

How complicated is the model, i.e. what is the parameterization? It might be possible to hack around

[Heman Shamachurn]

My problem is NLP. I have 9 decision variables and 13 constraints.

If I do not use optimizer, and I just optimize for a 2-3 steps, will it be possible? If yes, how do I the set the initial values for the decision variables please?

Thank you.

Heman

[Johan Löfberg]

The important thing is how the parameters enter the model.

The second question I don't understand. Of course you can simply use optimize, that is the standard way. Initials are set using assign, and then activated with the 'warmstart' option.