search

flexcompute / tidy3d

Fast electromagnetic solver (FDTD) at scale.
https://docs.flexcompute.com/projects/tidy3d/en/latest/
GNU Lesser General Public License v2.1
186 stars 41 forks source link

Automatic setup of path integrals #1735

Open dmarek-flex opened 5 months ago

dmarek-flex commented 5 months ago

As simulations become more complex, setting up the path integrals for each port will become tedious! We should implement a method that automatically creates path integrals from very simple user input. For example, defining a port should only require the user to specify a port plane along with the structures representing the positive terminal and the negative terminal (ground reference). Then, using the geometry of the chosen structures we should be able to automatically define both integrals needed for computing voltage and current.

dmarek-flex commented 5 months ago

The contour integrals needed by current computations can also be improved. Ideally we want the path around the conductor to be as close as possible to the surface. This will require access to the Yee grid and possibly depend on the conformal mesh strategy used. @weiliangjin2021 I am curious if the magnetic field of a Yee cell that has been modified by conformal meshing could still be used for these computations, or would a buffer cell be needed?

weiliangjin2021 commented 5 months ago

I am curious if the magnetic field of a Yee cell that has been modified by conformal meshing could still be used for these computations, or would a buffer cell be needed?

For those cells, its permeability is also affected (no longer 1). In that regard, best to avoid crossing material interfaces?

dmarek-flex commented 5 months ago

Ok, so we need the closest cells that do not intersect with the surfaces of material interfaces.