Answer:
Part A.
Let f(x) = 0;
suppose x= a+h
such that f(x) =f(a+h) = 0
By second order Taylor approximation, we get
f(a) + hf'±(a) +
f''(a) = 0
± ![\frac{\sqrt[]{(f'(a))^{2}-2f(a)f''(a) } }{f''(a)}](https://tex.z-dn.net/?f=%5Cfrac%7B%5Csqrt%5B%5D%7B%28f%27%28a%29%29%5E%7B2%7D-2f%28a%29f%27%27%28a%29%20%7D%20%7D%7Bf%27%27%28a%29%7D)
So, we get the succeeding equation for Newton's method as
± ![\sqrt{f(x_{i})^{2}-2fx_{i}f''x_{i} } ]](https://tex.z-dn.net/?f=%5Csqrt%7Bf%28x_%7Bi%7D%29%5E%7B2%7D-2fx_%7Bi%7Df%27%27x_%7Bi%7D%20%7D%20%5D)
Part B.
It is evident that Newton's method fails in two cases, as:
1. if f''(x) = 0
2. if f'(x)² is less than 2f(x)f''(x)
Part C.
In case
is close to
, the choice that shouldbe made instead of ± in part A is:
f'(x) =
⇔ ![x_{i+1} = x_{i}](https://tex.z-dn.net/?f=x_%7Bi%2B1%7D%20%3D%20x_%7Bi%7D)
Part D.
As given
=
= h
or h =
- ![x_{i}](https://tex.z-dn.net/?f=x_%7Bi%7D)
We get,
f(a) + hf'(a) +(h²/2)f''(a) = 0
or h² = -hf(a)/f'(a)
Also, (
-
)² = -(
-
)(f(
)/f'(
))
So, f(a) + hf'(a) - (f''(a)/2)(hf(a)/f'(a)) = 0
It becomes h = -f(a)/f'(a) + (h/2)[f''(a)f(a)/(f(a))²]
Also,
=
-f(
)/f'(
) + [(
-
)f''(
)f(
)]/[2(f'(
))²]