yade-users team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> Q1. I noticed that in SCG, CohesiveFrictionalLaw etc, we have switched
> from incremental shearForce to incremental shear displacement.   One
> reason is to prepare for the possibility of having a periodic
> boundary?  What are the drawbacks of using shearForce increments?
> There was a problem with CPM, as mentioned by Vaclav.  But I haven't
> fully understood the problem.  
shearForce depends on rigidity, which might change during contact
lifetime (think plasticity); it is more natural to compute strain (as
geometric characteristic) from the geometrical configuration, since you
don't have to pass rigidity as parameter. I think the reason was
something like this.

(I have a clean implementation of incremental geometry in mind, but it
will take some time)

> Q4.   I would need 4Matrix/ 5Matrices, and do multiplications Inverse,
> NewtonRaphson etc...  I can code Newton-Raphson myself if there's no
> available code, but would like to have libraries for Matrices. The
> easiest library to use is wildmagiclibrary?  This is for my
> superquadratic.  Glad that Sergei has solved aspherical rotations. :)
> But I'm still learning to code in YADE.  Will take maybe 2-3 months.
> I'll look at VTKMarchingCube (Thx Vaclav).  
> http://yade.wikia.com/wiki/Installation_details#Installing_the_wildmagic_3_library_on_debian
Don't use wm3, it is outdated and (I think) not the best one. I would go
with eigen (http://eigen.tuxfamily.org), it is very well documented,
well designed, actively used and developed and very fast. If we drop wm3
at some (distant) point in the future, I think it is the bast candidate
to replace it for numerics anyway. You might have a look at boost::ublas
as well, but I have no experience with that one.
> 
Cheers, Vaclav