[dune-pdelab] Enquiry: testelasticity.cc

[Christian Engwer]

Dear Nagesh,

the function `g` was never actually used. It was jsut copied and never
properly implemented, as nobody took a decision whether to implement
traction or stress.

> param_.g(eg.entity(),qp.position(),y);  (eg stands for 'Entity geometry' I
> presume)
> 
> However, the function 'lambda_boundary' has arguments:
> const IG& ig, const LFSV_HAT& lfsv_hat, R& r (ig stands for "Intersection
> geometry" I presume)

the intersection yields access to global coordinates and to the
neighboring cells. So you could evaluate g using the inside entity of
the intersection. On the other hand, the intersection represents the
boundary part of the cell, so it seems more appropriate to use that as
the parameter to `g`.

If you want to evaluate some global position you can use the geometry
of the intersection to map from intersection local coordinates to
global coordinates. This would be part of your particular
implementation of the parameter class.

Ciao
Christian