Mark Wagner wrote:

 > The algorithm used is numerically unstable.  Under certain
 > circumstances, points will start heading off towards infinity.
 > isinfinite() attempts to detect this and currently stops
 > calculations with an error.
 >
	Exactly (no point in spending several hours computing something, just to 
discover that the result is full of NaNs and Infs, especially since as 
soon as they start appearing, the computations slow down). However, I'm 
not sure why the algorithm is unstable like that and why it depends on 
the cloth parameters (intuitively, I would have thought that it should 
depend only on step: being stable if step<1 and unstable if step>=1)

 > If you haven't already implemented it, I'd like interaction with
 > spheres (or even better, blobs ;-).
 >
	Spheres shouldn't be too long in coming. Blobs will probably have to wait 
more...

		Jerome
-- 

* Abandon the search for truth, * mailto:ber### [at] inamecom
* Settle for a good fantasy.    * http://www.enst.fr/~jberger
*********************************

Post a reply to this message