> Ok then I will update my law, I missed these introductions. Thanks!
You are very welcome. I added those functions yesterday actually, when I
added the homoDeform==1 (position update) and had tocheck what kind of
homoDeform we had at multiple places (ScGeom, NewtonIntegrator etc).
--
periodic b
On 26 October 2010 20:29, Bruno Chareyre wrote:
>
> You are right, of couse.
>>> But how can I get shift2 in my Law* engine?
>>>
>>>
>> No need here.
>
> Why not? As I understand it, if in Sergei's law you want to use
> getIncidentVelocity function because you need to compute damping, then you
2010/10/26 Václav Šmilauer
> Moving the discussion back to the bug tracker...
>
> Chiara: misunderstanding, sorry, I meant that the prevous
> Ig2_Facet_Sphere_ScGeom did not work for facets because the shift2
> parameter was disregarded. Otherwise it should work with facets just
> fine, as well a
Moving the discussion back to the bug tracker...
Chiara: misunderstanding, sorry, I meant that the prevous
Ig2_Facet_Sphere_ScGeom did not work for facets because the shift2
parameter was disregarded. Otherwise it should work with facets just
fine, as well as any other shapes.
For getting periodi
My reply to mails in this thread is at
https://bugs.launchpad.net/yade/+bug/666246/comments/7, please discuss things
at the bugzilla so that comments related to the issue are kept in one single
place.
___
Mailing list: https://launchpad.net/~yade-
You are right, of couse.
But how can I get shift2 in my Law* engine?
No need here.
Why not? As I understand it, if in Sergei's law you want to use
getIncidentVelocity function because you need to compute damping, then
you need to get shift2 first.
I can't remember if
Ooops, sorry to have this in last commit.
B.
1. Disable debug messages from ScGeom in both debug and non-debug (!!) builds
modified:
pkg/dem/ScGeom.cpp
___
Mailing list: https://launchpad.net/~yade-dev
Post to : yade-dev@lists.launc
On 26 October 2010 19:01, Bruno Chareyre wrote:
>
> You are right, of couse.
>> But how can I get shift2 in my Law* engine?
>>
>>
> No need here.
Why not? As I understand it, if in Sergei's law you want to use
getIncidentVelocity function because you need to compute damping, then you
need to ge
26.10.2010 21:54, chiara modenese:
Btw, according to Vaclav, for Facet-Sphere interaction shift2 would
not work properly.
Only with ScGeom, I meant.
Sorry, I don't understand this.
shift2 is used to put both interacitons bodies to one periodic cell.
So, why is it would not work prop
You are right, of couse.
But how can I get shift2 in my Law* engine?
No need here.
Or to use precomputed shearIncrement from ScGeom?
yes.
But I need incidentVelocity too, but in ScGeom it is not cashed...
Indeed. Use increment/dt?
bad), I recommend turning it false :
Ig2_Fa
Btw, according to Vaclav, for Facet-Sphere interaction shift2 would not work
> properly.
>
Only with ScGeom, I meant.
___
Mailing list: https://launchpad.net/~yade-dev
Post to : yade-dev@lists.launchpad.net
Unsubscribe : https://launchpad.net/~yade-de
You are right, of couse.
> But how can I get shift2 in my Law* engine?
> What is prefered path?
> May be to use HertzMindlin.cpp:181 string:
>
> Vector3r shift2=scene->isPeriodic ?
> Vector3r(scene->cell->Hsize*contact->cellDist.cast()) :
> Vector3r::Zero();
>
> Or to use precomputed shearIncrement
Hi, Bruno.
>
> For the facet, couldn't it be something like :
>
> const Vector3r c2x = (geom.contactPoint - de2.pos + shift2) ?
You are right, of couse.
But how can I get shift2 in my Law* engine?
What is prefered path?
May be to use HertzMindlin.cpp:181 string:
Vector3r shift2=scene->isPeriodic
BTW on another machine (same maverick amd64, g++ build):
GenericSpheresContact Dem3DofGeom FrictPhys GlIPhysFunctor
PeriodicEngine GlIGeomFunctor BoundaryController BodyCallback
KinemSimpleShearBox TriaxialStressController Recorder GlShapeFunctor
--
unregistered class when built at maverick
https
Václav Šmilauer said: (by the date of Tue, 26 Oct 2010 18:58:32 +0200)
> > so, for example a function that calculates axis angle, could do a
> > check if that argument to acos is higher that 1.0. And if yes - then
> > perform a normalization, to bring back down that quaternion to an
> > identi
> so, for example a function that calculates axis angle, coule do a
> check if that argument to acos is higher that 1.0. And if yes - then
> perform a normalization, to bring back down that quaternion to an
> identity quaternion. This method will prevent quaternions from growin
> bigger than indent
** Attachment added: "script testing io of all yade classes"
https://bugs.launchpad.net/bugs/666876/+attachment/1712032/+files/save-class.py
** Changed in: yade
Importance: Undecided => High
** Changed in: yade
Status: New => Confirmed
** Changed in: yade
Milestone: None => 0.
Public bug reported:
The problems with "unregistered class" seem to re-surface at maverick.
The packaged version (bzr2460) runs the attached script just fine, but
manually-compiled versions fail:
This is what classes fail to save, a g++ build: ['FrictMat',
'PeriodicEngine', 'BodyCallback', 'GlIGe
> 3SR lab plans to grow henriette, its cluster, and we try to evaluate the
> needs of GPU computing solutions. Is there a plan for yade to use such
> techno (CUDA, OpenCL) ?
Not that I know of. I discussed it with some people recently, and it
does not seem to be fit for yade as it currently is. We
Janek Kozicki said: (by the date of Tue, 26 Oct 2010 18:40:50 +0200)
> Bruno Chareyre said: (by the date of Tue, 26 Oct 2010 17:09:12 +0200)
>
> > Below is an example (from revno 2514). If somebody see an obvious reason
> > for getting a NaN, please let me know; if not, I'll send this to
Please go on.
I am attaching a full example. The toAxisAngle routine implemented by
hand is only to check the calculation results. It is not how it
_should_ be implemented. In fact I didn't look at wm3 or at any other
source code , I just looked at the quaternion formulas and coded them down.
Th
revno: 2516
committer: Václav Šmilauer
branch nick: trunk
timestamp: Tue 2010-10-26 18:31:33 +0200
message:
1. Disable debug messages from ScGeom in both debug and non-debug (!!) builds
modified:
pkg/dem/ScGeom.cpp
--
lp:yade
https
On 26 October 2010 16:40, Bruno Chareyre wrote:
> Now the question is : why does it work?!!
>
Hi Bruno,
the problem arose because I was including the moment law in Hertz-Mindlin
but somehow I was getting segmentation fault and in debug mode I realized it
was not creating an ScGeom instance rather
Now the question is : why does it work?!!
Because I've been testing only with
Ig2_ChainedCylinder_ChainedCylinder_ScGeom...
B.
___
Mailing list: https://launchpad.net/~yade-dev
Post to : yade-dev@lists.launchpad.net
Unsubscribe : https://launc
Not sure it needs a 3rd common function actually. It should be enough
to call new in ScGeom6D::go when needed, then apply
ScGeom::go on the interaction, safely casting ScGeom6D to ScGeom3D .
So new<> would never be called from inside the 3D version.
Bad idea, the distance would be checked twi
Not sure it needs a 3rd common function actually. It should be enough to
call new in ScGeom6D::go when needed, then apply ScGeom::go on
the interaction, safely casting ScGeom6D to ScGeom3D . So new<> would
never be called from inside the 3D version.
I'll think about it.
B.
On 26/10/10 17:26,
revno: 2515
committer: Václav Šmilauer
branch nick: trunk
timestamp: Tue 2010-10-26 17:39:49 +0200
message:
1. Fix laoding of unitialized LawTester
modified:
pkg/dem/DomainLimiter.cpp
--
lp:yade
https://code.launchpad.net/~yade-dev
Oh, yes, you are right. Initially, I put all code in a third function
that was called from both go(), but it was more compact with the current
style - and I didn't realize the obvious allocation problem. Feel free
to fix that, if you don't I'll do it a.s.a.p (i.e. tomorrow maybe).
Now the quest
Hi Bruno,
we were just reading the ScGeom6D code and there a serious bug. Suppose
a potential interaction of 2 spheres; Ig2_Sphere_Sphere_ScGeom6D::go
calls Ig2_Sphere_Sphere_ScGeom::go, which, if it succeeds, creates a new
ScGeom object and assigns it to I->geom. Then the execution goes back to
I
Below is an example (from revno 2514). If somebody see an obvious reason
for getting a NaN, please let me know; if not, I'll send this to Eigen
(the first line should be enough).
Bruno
NaN angle found in angleAxisr(q), for quaternion 1 -1.56132e-11
7.47311e-11 -1.7066e-10, after quaternion p
revno: 2514
committer: bchareyre
branch nick: yade
timestamp: Tue 2010-10-26 16:59:45 +0200
message:
- more debug info for quaternions with nan angle.
modified:
pkg/dem/ScGeom.cpp
--
lp:yade
https://code.launchpad.net/~yade-dev/yad
revno: 2513
committer: Václav Šmilauer
branch nick: trunk
timestamp: Tue 2010-10-26 16:29:01 +0200
message:
1. Update the dispatch-torture script.
modified:
scripts/test/dispatcher-torture.py
--
lp:yade
https://code.launchpad.net/~
Luc, if you want to investigate nan's with quaternions, you can try
rev2511 with #define Q_DEBUG.
A good example to show the problem is
scripts/test/chained-cylinder-spring.py
B.
___
Mailing list: https://launchpad.net/~yade-dev
Post to : yade-de
revno: 2512
committer: bchareyre
branch nick: yade
timestamp: Tue 2010-10-26 16:09:37 +0200
message:
- Fix cylinder reg. test with respect to r2509(2510).
- Add output for nan quaternion products #ifdef Q_DEBUG.
- Document some fla
Hi Vaclav
>I was thinking about this many times, and discussed it with Bruno and Chiara.
>The reason to not do that is that we don't want to do things behind user's
>back, since he/she is supposed to understand what is going on. (Currently,
>Cell::integrateAndUpdate is called in such a hidden
BTW Bruno, can you run --test and fix chained cylinders attributes you
changed?
Done, sorry.
I also included in the same commit tests in order to spot nan results of
quaternion products, and send report to Eigen.
If Q_DEBUG is defined and if you run --test, you'll see your screen ful
Hi Sergei,
Are you sure these lines apply for facet-sphere?
> + const Vector3r c1x = geom.radius1*geom.normal;
> + const Vector3r c2x = -geom.radius2*geom.normal;
>
For the facet, couldn't it be something like :
const Vector3r c2x = (geom.contactPoint - de2.pos + shift2) ?
This is
Hi Jan,
> just one (maybe very stupid, I don't know :-) suggestion. Does
> ForceResetter() has any input parameters? and if not, is there any
> possibility to move it out from the python scripting and perform it
> automatically in the moving to the next time step? or oppositely, is
> there any nec
Hi Sergei,
Are you sure these lines apply for facet-sphere?
+ const Vector3r c1x = geom.radius1*geom.normal;
+ const Vector3r c2x = -geom.radius2*geom.normal;
For the facet, couldn't it be something like :
const Vector3r c2x = (geom.contactPoint - de2.pos + shift2) ?
This i
Hello,
just one (maybe very stupid, I don't know :-) suggestion. Does ForceResetter()
has any input parameters? and if not, is there any possibility to move it out
from the python scripting and perform it automatically in the moving to the
next time step? or oppositely, is there any necessity
revno: 2511
committer: Václav Šmilauer
branch nick: trunk
timestamp: Tue 2010-10-26 15:41:30 +0200
message:
1. Deprecate NewtonIntegrator::homotheticCellResize in favor of
Cell::homoDeform (compatibility interface with warning)
2. A
> If I'm the author of this comment in the sources (I am, probably),
> "close to identity" means close to null rotation. In other words the
> corresponding rotation matrix is very close to identity matrix.
> So, let M be a rotation matrix defining a rotation of 10^-8 degrees.
> Theoretically, M*
2/ Then I rediscovered that aa.angle() can return nan, as commented in
source file, it happens when the quaternion is close to identity. Please
apologise for my lack of culture, but what does represent physicaly a
quaternion close to identity?
hmm. All quaternions that we use are id
43 matches
Mail list logo