SpaldingTucker

Spalding-Tucker normalization for a level set function.

Defined in Rodin/Distance/SpaldingTucker.h

class SpaldingTucker;

This class implements the Spalding-Tucker normalization method, which evaluates an algebraic approximate-distance construction and interpolates it into the input finite element space. The pointwise transformation is

\[ d(x) = \frac{2u(x)}{|\nabla u(x)| + \sqrt{|\nabla u(x)|^2 + 2|u(x)|}} \]

where $ u $ is the input level set function and $ d $ is the normalized approximate signed distance.

Pointwise properties and discrete output

A positive denominator preserves the input sign and zero set before interpolation. The expression is undefined where both the field and its gradient vanish. Finite element interpolation and one-sided gradient evaluation affect the discrete result. The input scale and geometry determine the accuracy of the distance approximation; an analytic reference or refinement study supplies the relevant numerical check.

References

[3] [11] [13]

Usage Example

GridFunction u(fes);  // Some level set function
SpaldingTucker st;
auto dist = st(u);  // Normalized distance approximation

Public functions

template<class FES, class Data>
auto operator()(const Variational::GridFunction<FES, Data>& gf) -> auto
Applies Spalding-Tucker normalization to a grid function.

Function documentation

template<class FES, class Data>
auto operator()(const Variational::GridFunction<FES, Data>& gf)

Applies Spalding-Tucker normalization to a grid function.

Template parameters
FES Finite element space type
Data Data storage type
Parameters
gf in Input level set grid function
Returns Normalized grid function approximating signed distance