1: \begin{abstract}

2: It is shown how piecewise differentiable functions $F:  \R^n \mapsto \R^m $ that are defined by

3: evaluation programs can be approximated locally by a piecewise linear

4: model based on a pair of sample points $\check x$ and $\hat x$.  

5: We show that the discrepancy between function and model at any point $x$ is of

6: the bilinear order $\mathcal O(\|x-\cx\| \|x-\hx\|)$.

7: %This is a little surprising since $x \in \R^n$ may vary over the whole Euclidean

8: %space, and we utilize only two function samples $\cF=F(\cx)$ and $\hF=F(\hx)$,

9: %as well as the intermediates computed during their evaluation. 

10: As an application of the piecewise linearization procedure we devise a generalized 

11: Newton's method based on successive piecewise linearization and 

12: prove for it sufficient conditions for convergence and convergence rates equaling those of 

13: semismooth Newton. We conclude with the derivation of formulas for the numerically stable implementation 

14: of the aforedeveloped piecewise linearization methods.

15: \end{abstract}

16: