I am running simulations on a DBR w/ cavity setup. What happens is that at first, the simulation acts like you would expect with energy entering and then leaving. However, at a certain point below autoshutoff level 0.01, the energy starts to increase again and eventually diverges. From the movie, I can see that this increase in energy happens in the silicon regions. I have taken all the usual steps to solve diverging simulations including using steeped angle PMLs. Following Lumerical's article, I switched to metal boundaries and found that the simulation did not diverge so I believe this is not a dt stability issue. I checked the material fit and there is no negative imaginary index in the materials I am using. Is there something I am missing?