#1099 evaluateJacobian seems to be wrong for Q1LocalBasis
Metadata
| Property | Value |
|---|---|
| Reported by | Andreas Lauser (andreas.lauser@iws.uni-stuttgart.de) |
| Reported at | May 8, 2012 12:09 |
| Type | Bug Report |
| Version | 2.1 |
| Operating System | Unspecified / All |
| Last edited by | Christian Engwer (christi@conan.iwr.uni-heidelberg.de) |
| Last edited at | May 8, 2012 12:33 |
| Closed by | Christian Engwer (christi@conan.iwr.uni-heidelberg.de) |
| Closed at | May 8, 2012 12:33 |
| Closed in version | Unknown |
| Resolution | Not a bug |
| Comment | multilinear is not linear |
Description
I've just noticed that evaluateJacobian() for Q1 elements seem to produce jacobian vectors which depend on the position. if I'm not completely wrong (which is quite possible), this finite element is supposed to be linear in respect to all coordinates and the local gradients should be feature either -1 or 1 for all coordinates. Taking the partial derivatives on in[j] in q1localbasis.hh is also consistent with this view. I've attached a patch for dune-localfunctions (against the current dune 2.1 SVN release branch) to fix this issue. I suppose that since this issue is not present at the corners of the element, the gradients of this finite element have probably never been tested in the interior. On the other hand I might just miss something; in this case I would like to apologize for the noise.