yade-users team mailing list archive

[Janek Kozicki <janek_listy@xxxxx>]

Andreas Plesch said:     (by the date of Mon, 24 Oct 2005 13:14:56 -0700 (PDT))

> I thought it may have been an issue with version .911
> so I checked out the current .924 from svn. Loading
> the LinkedSpheres .xml and starting the simulation I
> get now:
> 
> yade: /usr/include/boost/shared_ptr.hpp:253: T*
> boost::shared_ptr<T>::operator->() const [with T =
> SpheresContactGeometry]: Assertion `px != 0' failed.
> Aborted
> 
> Not sure if I should start filing bugs ? Those issues
> may still be all installation related ?

you can file a bug if you want ;)

Jean-Patrick Plassiard is the author of formulas for DEM with moment
law. A year ago, his formulas were not fully developed and contained
many errors. But we ignored the errors and implemented the formulas
anyway. Current version of moment law in yade is inherently broken.

It works for a very narrow range of stiffenesses, and it was working
just after we implemented it. Many months later after several cleanups
the 'inherent brokeness' showed itself and I was unable to get it back
to work with whatever stiffness combination I tried. The fault is in
fact on yade side, not on the formulas side - we just did it too quick,
without proper consideration of everything that should be done.

But heads up! Jean-Patrick has recently finished his work on the
formulas (after two years), and they are now all correct. He will
publish a scientific paper about them, defend his PhD and all that
stuff. Currently I'm waiting to receive those formulas from him in a
manner that makes them eligible for implementing them. Only after that
SDECLinkedSpheres will work.

But - it all must take some time. If I do it alone - you shouldn't expect
it sooner than in 6 months.

If you want to contact Jean-Patrick and try to implement the formulas
yourself (where I will supervise your work) - you are welcome to. By
such way you should have it much faster (maybe 2 or 3 months).
Just say if you want to, then I will contact him.

-- 
Janek Kozicki                                                         |
_______________________________________________
Yade-users mailing list
Yade-users@xxxxxxxxxxxxxxxx
http://lists.berlios.de/mailman/listinfo/yade-users