Question about ABC in self-written FDTD - code

[aalemann]

Hi all,

I've written a simple 2D FDTD code using really simple ABC. In the absorber region the E-Field is damped as follows:
\[E(x) = E(x) \cdot \left[ 1 - \alpha \cdot { \left( \frac{x-d}{d} \right)}^{ 2} \right]\],
where α is a damping constant defining the strength of the absorber.

Does anybody of you know if this type of absorber (which does not work well for angles smaller than 40 degrees) has a specific name?

Thanks in advance,
aalemann

 

[aalemann]

*push*

maybe this absorber is too simple to have a specific name...?

 

[CG0210794]

Hello ,

If you look at my group's website there are some notes on FDTD. Within those notes you will find a section on Perfectly Matched Layers (PML). They are quite effective .
**broken link removed**
Go under the resources tab.

Cesar

 

[aalemann]

thanks for posting the link to your notes, but my question was not about how to improve my simple "non"-radiating boundary, I just wanted to know if the method as I posted it above has a separate name.

 

[rodgerwxh]

aalemann, you misunderstand the concept of ABC and PML, ABC has no thickness, it is only a boundary condition. predicting the outside field using 1st order, 2nd order ......... approximations.

PML has a thickness, its formulation theoretically claim that all angle can be perfectly absorbed. And it has a thickness (Layer). Which will make the E and H field decay as it propagate deeper into the layer. Which is what you wrote in your first post.