Bug #78

VTK saving does not make sense for high order and discontinuous elements

Added by Murtazo Nazarov over 4 years ago. Updated over 3 years ago.

Status:In ProgressStart date:11/05/2013
Priority:NormalDue date:
Assignee:Kaspar Müller% Done:

0%

Category:-
Target version:0.9.0-hpc

Description

The current implementation of the VTK files either considers that the function is P1 or interpolates it into P1, which does not make sense for Pk elements.

One (ugly) suggestion could be if F_P2 \in Mesh_1 a P2 function do the following:
  • create Mesh_2 = refine(Mesh_1), define F_P1 \in Mesh_2
  • interpolate F_P2 into F_P1
  • save F_P1.

SHOEVTK.pdf (1.05 MB) Aurélien Larcher, 04/26/2013 12:28 am


Subtasks

Bug #114: VTK saving does not make sense for high order elementsNew


Related issues

Related to DOLFIN HPC - Bug #122: Review request [dolfin-hpc|commit:00c4566da00c5d501e8a93d... Closed 12/02/2013 02/07/2014

History

#1 Updated by Niclas Jansson over 4 years ago

Murtazo Nazarov wrote:

The current implementation of the VTK files either considers that the function is P1 or interpolates it into P1, which does not make sense for Pk elements.

One (ugly) suggestion could be if F_P2 \in Mesh_1 a P2 function do the following:
  • create Mesh_2 = refine(Mesh_1), define F_P1 \in Mesh_2
  • interpolate F_P2 into F_P1
  • save F_P1.

True, P1 elements has been one of VTK's limitation for a while. However, they have recently added support for non-linear scalar and vector-fields. It is not a general implementation, and I think it only supports up to 4th or 6th order elements at the moment.

Implementing this should be easy, but we need to dig out some documentation. As we all know, VTK is notoriously poor at documenting its fileformat.

#2 Updated by Aurélien Larcher over 4 years ago

Apparently the job was done by Sandia.
Attached is a techreport from 2004.

#3 Updated by Aurélien Larcher almost 4 years ago

  • Assignee set to Kaspar Müller
  • Target version set to 0.9.0-hpc

#4 Updated by Aurélien Larcher almost 4 years ago

  • Subject changed from VTK saving does not make sence for Pk elements to VTK saving does not make sense for Pk and discontinuous elements

#5 Updated by Aurélien Larcher almost 4 years ago

  • Subject changed from VTK saving does not make sense for Pk and discontinuous elements to VTK saving does not make sense for high order and discontinuous elements

#6 Updated by Kaspar Müller almost 4 years ago

  • Status changed from New to Closed

#7 Updated by Aurélien Larcher almost 4 years ago

  • Status changed from Closed to In Progress

The proposed patch does not allow compatibility with UFC 1.x and has slight issues with code style compliance.
I propose the fix to be pushed to your own branch, open a code review and then I will merge it.
Best regards,

Aurelien

#8 Updated by Kaspar Müller over 3 years ago

Aurélien Larcher wrote:

The proposed patch does not allow compatibility with UFC 1.x and has slight issues with code style compliance.
I propose the fix to be pushed to your own branch, open a code review and then I will merge it.
Best regards,

Aurelien

Are you refering for code style to:
http://dryad.csc.kth.se/projects/dolfin-hpc/repository/revisions/master/entry/doc/misc/policy
?

Also available in: Atom PDF