Learning Forum

[Jason Kawaja]

In 2021, we ran the simulations using Lumerical version 2021 on Windows and stored the results in the 'Version 2021' folder. This folder includes a Matlab code for data processing (please update the file path before running).

 

Recently, we conducted the simulations using Lumerical version 2024 R1.2 and saved the results in the 'Version 2024 R1.2' folder. We observed an unexpected discrepancy in the results, specifically a sudden shift in the range L=1.6~1.8um in the Slot_Phase (derived from the angle of S in the Slot_Sweep.fsp sweep) and notable differences in Slot_dPhase1_cor (Lines 7 and 8) in RefractiveIndex_Slot.m.

 

We also ran the simulations using version 2023 on Linux and encountered the same issue with this jump.

 

All affected simulation points were completed 100% shown in the attached screenshot. Could this issue be related to changes in the initial E field during the sweep or other modifications in hidden parameters between the versions?

[Guilin Sun]

Since the forum can only show limited information, it is hard for us to diagnoze the cause. 

You mentioned "Could this issue be related to changes in the initial E field during the sweep or other modifications in hidden parameters between the versions?" may I know what does "the initial E field" mean here? did you chage the source amplitude? or any other change to the simulation files? the phase change is a 2pi jump, or slight change in magnitude? please elaborate more on this. Thank you!

 

 

[hangbo.yang]

Hi Guilin,

Thank you so much for your reply!

It's better for us to share the original fps files to you via email. Otherwise, what we described above is meaningless. 

may I know what does "the initial E field" mean here? "The initial E field" means the initial Input of the source in FDTD model.

did you chage the source amplitude? No.

or any other change to the simulation files? No.

the phase change is a 2pi jump, or slight change in magnitude? Slightly (sub-rad) phase change.

[Guilin Sun]

For source amplitude, please use the default 1v/m, except you are doing some non-linear effects;

Small change of phase usually is ok, in particular when the real or imaginary part of the field or S parameters is too small. you can use wunrap to avoid 2pi jump.

Please check the real and imaginary parts.

 

 

[hangbo.yang]

Hi Guilin,

Thank you for your suggestion!

For the source, we use ports directly, shown below.

For the strange phase change, we have double checked both the real and imaginary parts and found it is not a 2pi jump, but slightly (sub-rad) phase change in the range L=1.6~1.8um.

[Guilin Sun]

" in the range L=1.6~1.8um" do you mean the location of the monitor? or wavelength?

Since one can install two versions, I would suggest to reinstall the older version (uninstall the newest version first) and then reinstall the newest version, then compare the results and see what has been changed. From the release notes that you can search online, I did not find any mention for FDTD meshing, port/monitor. Maybe material fitting changed? it is hard to diagnose with only one version. 

[hangbo.yang]

Thanks, Guilin!

L means the length of our simulated structure.

I have just rebuilt our structure from scratch and am currently running it. Let's see what the results show.

I will try your suggestion later if rebuilding does not work.