yade-dev team mailing list archive

[Christian Jakob <1474956@xxxxxxxxxxxxxxxxxx>]

Hi,

I agree, that in some special cases aabb should be used for
discretization, but...

... it should be downward compatible (user can decide if he/she want the discretization this way or this way ...  so a new bool is needed in Clump::updateProperties: discretionAabb = false [default])
... it needs a lot of extra code modification (each function, that calls updateProperties directly or indirectly)
... it needs an update of clump section in the DEM background manual [1]

Regards,

Christian


[1] https://yade-dem.org/doc/formulation.html#clumps-rigid-aggregates

-- 
You received this bug notification because you are a member of Yade
developers, which is subscribed to Yade.
https://bugs.launchpad.net/bugs/1474956

Title:
  Clump::updateProperties discretization not defined robustly

Status in Yade:
  Confirmed

Bug description:
  Most likely the reason for performance drop reported in [1]

  dx in the voxel decomposition is minRad/discretization.
  The problem is when a sphere of radius 1 is member of a clump which aabb has the size 1000 (for instance).
  Using the minimal discretization=1 gives a number of voxel = 1e9 in such case and it is no possible to go below this number.
  I think discretization should refer to aabb size, not member size. I have this fixed locally, let me know if you agree with the fix.

  [1] https://answers.launchpad.net/yade/+question/268954

To manage notifications about this bug go to:
https://bugs.launchpad.net/yade/+bug/1474956/+subscriptions