Why can cacheAgglomeration in GAMG work
 

Hello everybody,

I'm working with LES using OpenFOAMs GAMG to solve the pressure equation in PISO. After looking at the function of GAMG's parameters I'm wondering why cacheAgglomeration works.

I mean, it seems to me that if I set this to true GAMG uses the agglomeration it found in time step 1 for each further timestep. According to my basic multigrid knowledge this should only work if the structure of the matrix GAMG has to solve does not change or only a little bit.

As it seems to work fine for at least most situations this seems to be true. But I'm quite unsatisfied with using something just because it seems to be true. I'd prefer to know why the matrix's structure does'nt change so that GAMG can exploit it. Does anybody here have an idea about that or might give me a guess where I might find some information about it ?

Thank you very much,

Sebi