Two methods for identifying the inflection point of a curve have been established:
1)they work similar to the bisection method with controlled accuracy
2)they work for errored data as consistent estimators of inflection point
The contribution is twofold:
i)a pure numerical analysis method to find inflection point
ii)a statistical estimator of inflection point
So, which is the 'main' contribution: Numerical Analysis Or Computational Statistics?
http://arxiv.org/abs/1206.5478