modification to grid quadrature so that the quadrature points are the mi... #212

Merged
merged 1 commit into from Apr 15, 2014

Conversation

Projects
None yet
5 participants
@manavbhatia
Contributor

manavbhatia commented Feb 27, 2014

modification to grid quadrature so that the quadrature points are the mid-points of each interval, as opposed to the ends of each interval.

modification to grid quadrature so that the quadrature points are the…
… mid-points of each interval, as opposed to the ends of each interval.
@moosebuild

This comment has been minimized.

Show comment Hide comment
@moosebuild

moosebuild Feb 27, 2014

A build job for 2dca766 from recipe libmesh is waiting for a developer to activate it here: https://www.moosebuild.com/view_job/65

A build job for 2dca766 from recipe libmesh is waiting for a developer to activate it here: https://www.moosebuild.com/view_job/65

@jwpeterson

This comment has been minimized.

Show comment Hide comment
@jwpeterson

jwpeterson Feb 27, 2014

Owner

@roystgnr Any thoughts on this? I think you wrote the QGrid class to implement Newton-Cotes type rules? If anything I think this feature should go in a new class rather than modifying the existing QGrid.

Owner

jwpeterson commented Feb 27, 2014

@roystgnr Any thoughts on this? I think you wrote the QGrid class to implement Newton-Cotes type rules? If anything I think this feature should go in a new class rather than modifying the existing QGrid.

@friedmud

This comment has been minimized.

Show comment Hide comment
@friedmud

friedmud Feb 27, 2014

Owner

I just activated the tests - let's see if that turns anything up.

But I agree with @jwpeterson this seems like a different qrule to me... (or maybe an option on the same object...)

Owner

friedmud commented Feb 27, 2014

I just activated the tests - let's see if that turns anything up.

But I agree with @jwpeterson this seems like a different qrule to me... (or maybe an option on the same object...)

@roystgnr

This comment has been minimized.

Show comment Hide comment
@roystgnr

roystgnr Feb 27, 2014

Owner

My QGrid was a dumb-as-dirt rule, written solely for debugging the creation of new macroelements (by making it possible to test the element without having a proper subdivision-respecting quadrature rule for it). Not only is it not Newton-Cotes it's not even an accurate trapezoidal rule; we're talking O(1/N) convergence here thanks to the overweighted endpoints.

So while I'm usually reluctant to fundamentally change the behavior of existing code, I think it might be reasonable to make an exception for the case of really awful existing code.

But if we do, could we change all the geometric element cases to be consistent? The tensor product cases pick up the 1D rule but the tri and tet cases are each their own code.

Owner

roystgnr commented Feb 27, 2014

My QGrid was a dumb-as-dirt rule, written solely for debugging the creation of new macroelements (by making it possible to test the element without having a proper subdivision-respecting quadrature rule for it). Not only is it not Newton-Cotes it's not even an accurate trapezoidal rule; we're talking O(1/N) convergence here thanks to the overweighted endpoints.

So while I'm usually reluctant to fundamentally change the behavior of existing code, I think it might be reasonable to make an exception for the case of really awful existing code.

But if we do, could we change all the geometric element cases to be consistent? The tensor product cases pick up the 1D rule but the tri and tet cases are each their own code.

@roystgnr

This comment has been minimized.

Show comment Hide comment
@roystgnr

roystgnr Feb 27, 2014

Owner

s/overweighted endpoints/one overweighted, one dropped endpoint/

Owner

roystgnr commented Feb 27, 2014

s/overweighted endpoints/one overweighted, one dropped endpoint/

@moosebuild

This comment has been minimized.

Show comment Hide comment
@moosebuild

moosebuild Feb 28, 2014

Results of testing 2dca766 using libmesh recipe:

Passed on: linux

View the results here: https://www.moosebuild.com/view_job/65

Results of testing 2dca766 using libmesh recipe:

Passed on: linux

View the results here: https://www.moosebuild.com/view_job/65

@roystgnr

This comment has been minimized.

Show comment Hide comment
@roystgnr

roystgnr Apr 15, 2014

Owner

I'll merge this as-is and make up a subsequent PR for the tri/tet cases.

Owner

roystgnr commented Apr 15, 2014

I'll merge this as-is and make up a subsequent PR for the tri/tet cases.

roystgnr added a commit that referenced this pull request Apr 15, 2014

Merge pull request #212 from manavbhatia/grid_quadrature_modification
modification to grid quadrature so that the quadrature points are the mi...

@roystgnr roystgnr merged commit 73a1793 into libMesh:master Apr 15, 2014

1 check passed

default Successfully passed all tests
Details
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment