Improved strategies for accurate capacitance matrix compuation
Posted: 29 Aug 2023, 18:08
Hi,
There has recently been some work to improve the capacitance matrix computation of Elmer. This is a features that has been in Elmer code since many years but rather recently the features we made compatible with adaptive mesh refinements (AMR), p-elements, curved boundaries etc. The single one most important feature for accurate results is use of suitable p-elements (maybe 2-4 order). This is made efficient by the use of p-multigrid. The adaptivity requires MMG3D library which may not be easily compiled so this may be more difficult to take into use. The p-elements should scale pretty well in parallel whereas AMR may have some issues in parallel.
For more details see the recent commit here:
https://github.com/ElmerCSC/elmer-elmag ... OfTwoBalls
We would be interested hearing whether this provides improvements for your computations.
-Peter
There has recently been some work to improve the capacitance matrix computation of Elmer. This is a features that has been in Elmer code since many years but rather recently the features we made compatible with adaptive mesh refinements (AMR), p-elements, curved boundaries etc. The single one most important feature for accurate results is use of suitable p-elements (maybe 2-4 order). This is made efficient by the use of p-multigrid. The adaptivity requires MMG3D library which may not be easily compiled so this may be more difficult to take into use. The p-elements should scale pretty well in parallel whereas AMR may have some issues in parallel.
For more details see the recent commit here:
https://github.com/ElmerCSC/elmer-elmag ... OfTwoBalls
We would be interested hearing whether this provides improvements for your computations.
-Peter