Skip to content

[WIP] Provide test and fix for interpolation of general vector valued functions

Jakub Both requested to merge feature/interpolation-for-general-vector-valued-functions into master

This merge request aims at fixing the issue mentioned in issue #59 (closed) .

Two changes are provided:

  1. A very simple test interpolating constant (scalar and vector valued) functions, ultimately showing that the interpolation of the Raviart-Thomas basis does not work as intended.
  2. A fix in interpolate.hh which resolves 1).

The main idea in the fix is one has to distinguish whether the basis to be used for the interpolation employs scalar or vector valued weights/coefficients. For scalar valued function spaces, the distinction is redundant, but it is not e.g. for Raviart-Thomas elements opposed to linear Lagrange elements powered to some degree larger 1.

The fix seems quite heavy for the intention to simply return either a component or the entire function which is to be interpolated. Expecting some propositions for changes, the names of the helper classes have been chosen merely non-explanatory and should not be understood as final. Therefore this merge request is still marked as WIP.

Merge request reports