Menu

Discussion Forum

New Discussion
Filter

Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Effective index imaginary part doesn't appear.

Posted Jul 20, 2015, 5:56 a.m. EDT Version 4.4 5 Replies

Nafiz
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Hi


I am trying to measure the confinement loss in a photonic crystal fiber using the fundamental mode effective index. for a wavelength 1.55um and upwards, the fundamental mode has an inedx of 1.270578-1.825e-10i , which means I could calculate the confinement loss from the imaginary part. But for any smaller wavelength, the effective index of fundamental mode becomes real. It seems like COMSOL isn't calculating imaginary part below e-10i range.

Any solution to this problem ?

Attachments:

5 Replies Last Post Mar 28, 2016, 2:52 p.m. EDT
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 20, 2015, 7:00 a.m. EDT
Hi

you should always suspect numerical issues if your ratios are > 1E7 or thereabout, as the way numbers are presented by binary coding in the computer limits the resolution to several times the internal variable "eps=2.2E-16" (is FPU dependent).

And as most physics equations are of second order the maximum ratio one should expect is rather several times sqrt(eps)=1.5E-8

Therefore I'm not that astonished that you have issues resolving index ratios in the order of 1E-10 with any numerical binary computing algorithm

--
Good luck
Ivar
Hi you should always suspect numerical issues if your ratios are > 1E7 or thereabout, as the way numbers are presented by binary coding in the computer limits the resolution to several times the internal variable "eps=2.2E-16" (is FPU dependent). And as most physics equations are of second order the maximum ratio one should expect is rather several times sqrt(eps)=1.5E-8 Therefore I'm not that astonished that you have issues resolving index ratios in the order of 1E-10 with any numerical binary computing algorithm -- Good luck Ivar
Nafiz
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 20, 2015, 7:16 a.m. EDT
Hi

Thank you. So you are saying this is not an issue with the PML, or giving any imaginary refractive index to the cladding material ?

but confinement loss comes from the equation- Loss= 40*pi*Im[neff]/(loge(10)*L) dB/km. I could only measure losses of order 1e-3 dB/km only if the imag(neff) is of order ie-11. How do you advise to calculate it then ?
Hi Thank you. So you are saying this is not an issue with the PML, or giving any imaginary refractive index to the cladding material ? but confinement loss comes from the equation- Loss= 40*pi*Im[neff]/(loge(10)*L) dB/km. I could only measure losses of order 1e-3 dB/km only if the imag(neff) is of order ie-11. How do you advise to calculate it then ?
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 20, 2015, 7:38 a.m. EDT
Hi

I'm "just" saying take care and look very closely to the algorithms when you are having such large differences on derived variables.
Therefore meshing resolutions (discretization of your equations) is even more important when you have such large relative differences.

Unfortunately I'm not experienced enough with the "tricks" to ensure stable numerical values for that particular physics you are using there.

--
Good luck
Ivar
Hi I'm "just" saying take care and look very closely to the algorithms when you are having such large differences on derived variables. Therefore meshing resolutions (discretization of your equations) is even more important when you have such large relative differences. Unfortunately I'm not experienced enough with the "tricks" to ensure stable numerical values for that particular physics you are using there. -- Good luck Ivar
Jiaqi Ye
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jan 1, 2016, 8:56 a.m. EST
Hi, have you solved this problem, I also have this issues, it would be nice to get your advice.

Thanks.
Hi, have you solved this problem, I also have this issues, it would be nice to get your advice. Thanks.
Nafiz
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Mar 28, 2016, 2:52 p.m. EDT
Yes, later i found that the imaginary part in fact existed, but it was too small for COMSOL to calculate. anything below 10^-9 doesn't seem to appear in comsol. Change the deciding parameters like wavelength and you'll see the imaginary part starts to appear after a certain increase or decrease of the parameter.
Yes, later i found that the imaginary part in fact existed, but it was too small for COMSOL to calculate. anything below 10^-9 doesn't seem to appear in comsol. Change the deciding parameters like wavelength and you'll see the imaginary part starts to appear after a certain increase or decrease of the parameter.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.

Suggested Content