← Back to team overview

yade-users team mailing list archive

Re: R: Re: Contact detection

 

> so you wrote from scratch the InsertionSortCollider... Why didn't 
> you choose to implement a grid collider? Do you have that algorithm written in 
> another language (e.g. Fortran)? Could it be possible to translate it in C++?

Back at that point, I needed to have results quickly and writing
InsertionSortCollider was the fastest way to alleviate my frustration
from crappy code I was stumbling upon all the time. In last 2 years, I
had to rewrite (almost) all parts of yade I use, since the code was so
horrible...

Besides that, grid-based collider will (probably) need to adjust way of
storing bodies/interactions. I wanted to get other things sorted out
first (like removing bodies, which is OK now).

> Honestly could you let me know me know if, within a 
> reasonable period of time, we will have the grid collider implemented in YADE?

I might do it before June 2010 when I finish PhD, but it is not the most
important for me now. The algorithm itself is not problem, it's been
described in articles plenty of time.

(I am still thinking why Olivier Galizzi and Janek didn't use
state-of-the-art algorithm in 2004 when yade was started??)

Unfortunately, it seems guys in Grenoble are not able to get developers
to work on yade seriously.

> Sorry if I guess but I am at the beginning of my D.Phil. and I am looking for 
> the "best" DEM code according to my needs.

Have a look at esys-particle. (If I do the grid collider, I will have a
look at how they do things first.)

Cheers, Vaclav





Follow ups

References