Avoid divide by epsilon in macroscopic algorithms
#4448
Conversation
… macroscopic-properties-avoid-divide-by-epsilon
… macroscopic-properties-avoid-divide-by-epsilon
ax3l
added
enhancement
|
@eebasso just curious: for which (meta?) material is epsilon=0? |
I have not yet tested it. I'll try it soon.
According to "N. Garcia, E. V. Ponizovskaya, John Q. Xiao; Zero permittivity materials: Band gaps at the visible. Appl. Phys. Lett. 18 February 2002; 80 (7): 1120–1122", one possible realization would be a dielectric medium with embedded metallic nanoparticles and nanowires. They state
Furthermore, there are materials such as superconductors that can have negative permittivity at radio frequencies. |
… macroscopic-properties-avoid-divide-by-epsilon
|
@eebasso would you like to add your test result in the PR description just to double check ? Thank you! |
Summary
This PR makes the calculation of
alphaandbetain the algorithms of MacroscopicProperties.H more efficient and avoids dividing by the relative permittivityepsilon. The check onepsilonupon initialization is relaxed to require non-negativity instead of strict positivity.Additional Information
The changes allow for the possibility of
epsilonbeing zero while reducing the numbers of operations to calculate the coefficients. In particular, the number of division operations gets cut from two to one. In the case ofbeta, an additional multiplication operation is cut. This could be significant since these calculations occur inside the for loops to evolve the E-field.