Preconditioners in 3D FEA
Posted: Sat May 24, 2025 11:12 am
The title says it all - anybody have experience (usage, implementation, or analysis) on preconditioners for 3D magnetics (vector potential formulation, with or without eddies, with or without circuit connections)?
So far, I haven't come across anything that would actually beat incomplete Cholesky + conjugate gradient on realistic problems. Lots has been published on the topic, but I don't want to place my bets (as in implementation time) on anything that's only been tested on a unit cube with a relative permeability of 10.
So far, I haven't come across anything that would actually beat incomplete Cholesky + conjugate gradient on realistic problems. Lots has been published on the topic, but I don't want to place my bets (as in implementation time) on anything that's only been tested on a unit cube with a relative permeability of 10.