Discussion Forum

Hi

I would like to ask advice regarding my optimisation problem using the optimisation module.

My problem is one where wish to find a set of parameters P1, P2, P3 where a function F(P1,P2,P3,f) (f=frequency) can fit my simulation data M(f) i.e. P1, P2 and P3 such that we have min[sum(F(P1,P2,P3,f)-M(f))^2].

I have solved this initial problem by
i). Solving my simulation to generate M(f) data from a first study
ii). Saving M(f) to a text file
iii). Setting up an optimisation 'physics' which imports this text data, and then defining a least squares objective
iv). Setting up F(P1,P2,P3,f) as a variable
v). Solving my optimisation problem in a new study using Nelder-Mead method.

My questions are thus...

1. I would like to know if it is possible to avoid the 2 stage (study) process, having to compute M(f) separately and saving to a text file. Is there a way to find P1,P2,P3 as additional degrees of freedom in a single stage/study process.

2. Secondly, I would like to do a geometry sweep for X=0,1,2,...,N. How can I set up the problem so I find the array P1(X), P2(X), P3(X) min[sum(F(P1(X),P2(X),P3(X),f)-M(f,X))^2] for X=0:N. I realise this is possible if I define N types of the previous problem, but this is laborious, is there a shortcut?

3. If 2 is possible then I would also be interested to know if its possible to impose such a restraint as P1(X)/P1(0) = P2(X)/P2(0) = P3(X)/P3(0)

Kind regards