Attempting to get fminsearch to work. Likely a silly mistake.

I am writing code to simulate a double pendulum, whose two natural frequencies (wn) should be 2 and 5hz. I would like to optimize L1 and L2 (arm lengths) with fminsearch in order to bring (wn) to 2 and 5 hz.

My difficulty seems to lie in creating the cost function which is being used with fminsearch. What I have written is my best guess, which currently yields an error.

https://pastebin.com/avzJvQys

In short:

I am attempting to determine L1 and L2 such that wn(1) and wn(2) of the system are 2, and 5 hz respectively, and am having trouble setting up the function that interfaces with fminsearch itself.

Any help would be greatly appreciated. Thank you.

[Walter Roberson]

Why are you passing m1 and m2 into double pendulum cost function when that function does not accept them as arguments and does not use them?

Why bother to search at all? You know that for unbound searches the minimum of (x1-a)^2+(x2-b)^2 can only be when x1=a and x2=b so there is no point trying other values.

Answer 1

[Marco Riani]

Dear Gus,

The MATLAB sintax to make fminsearch work is

L1 = 0.0337;  % Length of arm 1 (Rough guess)
L2 = 0.0166;  % Length of arm 2 (Rough guess)
Double_Pendulum_Cost = @(wn) ((wn(1)-12.57)^2+(wn(2)-31.42)^2);

sopt = fminsearch(Double_Pendulum_Cost,[L1 L2]);  % Minimize L1 and L2

Of course, as Walter was suggesting: "why bother to search at all?". in this case sopt is

12.5700 31.4200

irrispective of the values of L1 and L2

Hope it helps

Marco