yade-users team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> Well, looking at the sweep and prune algorithm actually I see that it 
> sorts the min and max coordinates of the bb. My question is: Why do we
> need to 
> extend the sorting to all the bb and not just for instance to bb close
> to each 
> others?

You would have to put the bb's to some groups to get notion about being
"close", which would entail another sorting step. Insertion sort is
optimal for substantially pre-ordered series, which is our case
(particles don't move too much between collider runs).

(BTW you're probably the first person besides me to read that code.)

>  Is this contact detection method more efficient in terms of search
> time 
> than other methods based on spatial subdivisions? 

No.