Learning Forum

[Yeolheong Seong]

Hi.

I'm using the latest Lumerical solver and have a problem with using optimization with EME.

In optimization, I can select 'user s parameter' as a figure of merit, but it does not optimize the value that I want.

So I used FOM script like below:

(T_fom2 is user s matrix, 3-port configuration)

T1 = abs(T_fom2(1, 2))^2;

T2 = abs(T_fom2(1, 3))^2;

a1 = T1 + T2;

a2 = abs(T1 - T2);

fom = a1 - a2;

it looked like it worked, but after the simulation ended, 

I found that the 'best FOM during the optimization' and 'FOM from manual calculation(after EME propagate) in the best solution'

are different from each other.

I also found that in FOM script, I could not use 'emepropagate' neither 'getresult("eme", "user s matrix")'.

Also, I made the same result in model analysis and used it for FOM, but it had the same problem.

I think EME with optimization has some bugs especially for user s parameter, like no updating or whatever.

Is there any solution I can do? 

[Amrita Pati]

Hi Yeolheong,

I apologize for the delay in response. We seemed to have miss some posts in December. If I understand you correctly, when you look at the FOM from the optimized geometry and then do the EME calculation for the same geometry, they seem to be different. And you suspect that the user s-matrix is not being updated properly during the optimization? 
From the code you have shared, I don't think anything is wrong and that it should work properly. But let me run some tests and get back to you on this.

Regards,
Amrita