yade-users team mailing list archive

[Jerome Duriez <duriez@xxxxxxxxxxxxxxx>]

(I write to mailing list because of the two scripts I have to attach)

Hello,

I would like to show you the two attached scripts. It is about two 
spheres (in FrictMat) who do not move. O.engines contains only 
ForceResetter, InsertionSortCollider and Interaction Loop. And that's 
the difference between the two scripts, NewtonIntegrator or not. 
NewtonIntegrator is *not* contained in scriptCrash. Due to its name, 
this script crashes, whereas the second which contains NewtonIntegrator 
does not. The crash occurs more precisely in ForceContainer::Reset 
called by Engine ForceResetter.

I wonder if it is "normal" ? The reason is probably that 
NewtonIntegrator cares about forces of bodies, and then performs 
something indispensable (I do not know what exactly).

If you agree with what I say (?), I'd suggest at least to put a warning 
saying that NewtonIntegrator engine is in fact mandatory, and performs 
not only "integration of newtonian motion equations"

That is the reason of https://bugs.launchpad.net/yade/+bug/739429. My 
script did not contain NewtonIntegrator.

To conclude, the funny thing is that, if you get rid of interactions in 
scriptCrash, either by removing one of the two spheres, or by removing 
the InteractionLoop engine, crash disappears...

Jerome

--
Jérôme Duriez
ATER Polytech' Grenoble - Laboratoire 3S-R
04.56.52.86.49 (ne pas laisser de messages sur le répondeur)

# -*- coding: utf-8 -*-

# Very simple script to test for Bug


#Def of the material which will be used
O.materials.append(FrictMat(young=10e9,poisson=.25,frictionAngle=0.5,density=1e3))

#Def of the bodies of the simulations : 2 spheres, with names which will be useful after
O.bodies.append(utils.sphere([0,0,0], 1, fixed=True, wire=False, color=None, highlight=False))
#O.bodies.append(utils.sphere([0,2,0], 1, fixed=True, wire=False, color=None, highlight=False))

lowerSphere=O.bodies[0]
#upperSphere=O.bodies[1]


#Def of the engines taking part to the simulation loop
O.engines=[
	ForceResetter(),
	InsertionSortCollider([Bo1_Sphere_Aabb()],verletDist=0),# absence of NewtonIntegrator requires verletDist=0
	InteractionLoop(
			      [Ig2_Sphere_Sphere_ScGeom()],
			      [Ip2_FrictMat_FrictMat_FrictPhys()],
			      [Law2_ScGeom_FrictPhys_CundallStrack()]
			      )
	]






O.dt=1e-5
print 'Going for cycling'

O.run(2,True)
print 'Cycles done'


# -*- coding: utf-8 -*-

# Very simple script to test for Bug


#Def of the material which will be used
O.materials.append(FrictMat(young=10e9,poisson=.25,frictionAngle=0.5,density=1e3))

#Def of the bodies of the simulations : 2 spheres, with names which will be useful after
O.bodies.append(utils.sphere([0,0,0], 1, fixed=True, wire=False, color=None, highlight=False))
O.bodies.append(utils.sphere([0,2,0], 1, fixed=True, wire=False, color=None, highlight=False))

lowerSphere=O.bodies[0]
upperSphere=O.bodies[1]


#Def of the engines taking part to the simulation loop
O.engines=[
	ForceResetter(),
	InsertionSortCollider([Bo1_Sphere_Aabb()]),
	InteractionLoop(
			      [Ig2_Sphere_Sphere_ScGeom()],
			      [Ip2_FrictMat_FrictMat_FrictPhys()],
			      [Law2_ScGeom_FrictPhys_CundallStrack()]
			      ),
	NewtonIntegrator()
	]






O.dt=1e-5
print 'Going for cycling'

O.run(2,True)
print 'Cycles done'