Nonlinear Equations - UFRJ
Nonlinear Equations - UFRJ
Nonlinear Equations - UFRJ
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
[SEC. 7.3: ESTIMATES FROM DATA AT A POINT 101<br />
and rearrange terms in (1 + α − √ ∆)/(1 + α + √ ∆) = 1/2, we obtain<br />
the second degree polynomial<br />
2α 2 − 13α + 2 = 0.<br />
This has solutions (13 ± √ √ 17)/2. When 0 ≤ α ≤ α 0 = (13 −<br />
17)/2, the polynomial values are positive and hence q ≤ 1/2.<br />
Proof of Theorem 7.15. Let β = β(f, x 0 ) and γ = γ(f, x 0 ). Let h βγ<br />
and the sequence t i be as in Proposition 7.16. By construction, ‖x 1 −<br />
x 0 ‖ = β = t 1 − t 0 . We use the following notations:<br />
Those will be compared to<br />
β i = β(f, x i ) and γ i = γ(f, x i ).<br />
ˆβ i = β(h βγ , t i )) and ˆγ i = γ(h βγ , t i )).<br />
Induction hypothesis: β i ≤ ˆβ i and for all l ≥ 2,<br />
‖Df(x i ) −1 D l f(x i )‖ ≤ − h(l) βγ (t i)<br />
h ′ βγ (t i) .<br />
The initial case when i = 0 holds by construction.<br />
assume that the hypothesis holds for i. We will estimate<br />
So let us<br />
β i+1 ≤ ‖Df(x i+1 ) −1 Df(x i )‖‖Df(x i ) −1 f(x i+1 )‖ (7.12)<br />
and<br />
γ i+1 ≤ ‖Df(x i+1 ) −1 Df(x i )‖ ‖Df(x i) −1 D k f(x i+1 )‖<br />
. (7.13)<br />
k!<br />
By construction, f(x i ) + Df(x i )(x i+1 − x i ) = 0. The Taylor expansion<br />
of f at x i is therefore<br />
Df(x i ) −1 f(x i+1 ) = ∑ k≥2<br />
Df(x i ) −1 D k f(x i )(x i+1 − x i ) k<br />
k!