Learning Forum

[akoral19]

Hi,

I'm using the simulation which is "Inverse design of a Y splitter using topology optimization". I made some changes on the given .lsf and .py files in topology optimization. When I ran the forward and adjoint simulations that were created in the working directory after the entire simulation was over, I saw that these simulations ended in a short time. But when I examine the python code with the timer function, I see that the time spent in the forward simulation is more. You can see the table below where I made this comparison. When I run the forward FDTD simulation file (forward_0.fsp), it takes 10 seconds. On the other hand, as you can see in the table below, approximately 17 seconds are spent in the fdtd.runjobs part of the forward simulation. What could be the reason for this difference?

[Guilin Sun]

I think it is reasonable for this due to two main reasons:n1: the two simulations did not run in the same environment, eg, the system may have some background software running differently. To verify this, you may run the same pyton at different times and see the difference.n2: it is the device physics. To verify this, you may test a straight waeguide and see if the simulation time is significantly different. n