View Issue Details

IDProjectCategoryView StatusLast Update
0002498OpenFOAMBugpublic2017-03-13 12:53
ReporterShorty Assigned Tohenry  
PrioritynormalSeveritycrashReproducibilityalways
Status closedResolutionsuspended 
PlatformGNU/LinuxOSUbuntuOS Version14.04
Summary0002498: point|edge - CellsLeastSquares crashes for 1D calculation
DescriptionDear everybody,

I am not sure if the pointCellsLeastSquares/edgeCellsLeastSquares gradient scheme can only be used for 2D and 3D calculation. However, doing a 1D calculation, we get a floating point exception during the calculation of the gradient of the displacement vector.

To give an example, I attached a modified version of the solidDisplacmentFoam (plateHole) case. Allrun it and you will get the error during the initialization of sigmaD explicitly here:

 - twoSymm(grad(D))

Steps To ReproduceAllrun script added
TagsNo tags attached.

Activities

Shorty

2017-03-13 10:50

reporter  

plateHole.tar.gz (2,141 bytes)

henry

2017-03-13 11:52

manager   ~0007920

Why would you choose point|edge - CellsLeastSquares for 1D calculations? Surely "Gauss linear" is a better choice in every way?

If you need point|edge - CellsLeastSquares for 1D calculations please provide a patch which avoids the singularity in the scheme.

Shorty

2017-03-13 12:02

reporter   ~0007921

You are right, Guass linear makes much more sense in 1D calculations. However, I just run into that problem and wanted to inform you. You can close that topic. I will reopen it, if I will make a patch.

henry

2017-03-13 12:53

manager   ~0007922

Unimportant fix pending patch.

Issue History

Date Modified Username Field Change
2017-03-13 10:50 Shorty New Issue
2017-03-13 10:50 Shorty File Added: plateHole.tar.gz
2017-03-13 11:52 henry Note Added: 0007920
2017-03-13 12:02 Shorty Note Added: 0007921
2017-03-13 12:53 henry Assigned To => henry
2017-03-13 12:53 henry Status new => closed
2017-03-13 12:53 henry Resolution open => suspended
2017-03-13 12:53 henry Note Added: 0007922