Unexpected Negative Mapping Jacobian

[JaredCrean2]

This mesh

shows a negative mapping jacobian for node 1 of element 1, which seems weird because the elements are not inside out.

The mesh is 0 to 1 in both directions, with perturbation

  double nwaves = 1; //5  // number of waves over the domain (periods of sin)
  double amplitude = 0.25; //0.5 // fraction of hx used as perturbation magnitude

  double pertx = amplitude*hx*sin( M_PI*coords[0]*2*nwaves);
  double perty = amplitude*hy*sin( M_PI*coords[1]*2*nwaves);

[jehicken]

This might be dependent on how the Jacobian is computed in SBP. Does the problem persist if you use different element types?

[JaredCrean2]

Yes, Gamma, Omega, and Diagonal E (all p=1) are afflicted.

[JaredCrean2]

Also, element 1 is a straight-sided triangle (because its vertices sit at the zeros of the sin wave used to prescribe the perturbation).

[jehicken]

Sorry, I did not notice that there was a negative Jacobian on a straight sided element. That should not be an SBP issue, but I could be mistaken. Keep me posted.