I am way more familiar with energy transfer across boundaries than coding, ultimately I am an end user. The acceptable value of error (difference-delta-what have you) will greatly impact the time taken for convergence. The HP FEM that I was referring to in my post before the last, is a much faster (linear) solve that methods used by CFD. Maybe in the future our machines and techniques will allow these CFD models to be solved faster. But I am in a differnt realm of application, I'm sure your right.Urs wrote: To a surprisingly large degree the laws of electricity and mass spring systems are interchangeable. If you model a fluid as an array of nodes that have a mass each and that are connected by springs, the equations work the same way (the energy that leaves a node equals the energy that enters it, just like Kirchhoff's current law, that kind of stuff).
However, a more or less accurate model of a water glass will calculate a night or two for a few seconds of audio. This stuff works well for video (60 or so frames a second), but it'll be difficult to run it with 44100 frames per second
I really need to check out U-He software after after all my yakety. I've heard nothing but praise for Diva.
