Ask your own question, for FREE!
Mathematics 16 Online
OpenStudy (dls):

A small doubt regarding Newton Raphson method.

OpenStudy (dls):

I have to find a real root to the equation x = e^(-x) , using Newton Raphson method correct to 3 decimal places. I know that the relation is : \[\Large x_{n+1} = x_{n} - \frac{f(x_n)}{f'(x_n)}\] One thing I notice here is that \[\Large \frac{f(x_n)}{f'(x_n)} = -1\] always. So our relation boils down to \[\Large x_{n+1} = x_n + 1\].

OpenStudy (dls):

Now my doubt lies in the fact that, I need to make an initial approximation here. Say I take x0 = 0. I evaluate f(0) and it turns out to be -1. Then my x1 would be 1 ..then 2..3 and so on. This is a never ending sequence..because answer is nearly 0.60 (by manual calculations) but I don't ever come across it.

ganeshie8 (ganeshie8):

that's clever! very useful observation !

OpenStudy (dls):

hehe thanks :P

OpenStudy (dls):

So that relation never allows me to visit 0.60 because minimum difference is always 1. So how do I do this via NR?

ganeshie8 (ganeshie8):

it seems your relation may not work

ganeshie8 (ganeshie8):

what expression are you taking as f(x) ?

OpenStudy (dls):

e^(-x)

ganeshie8 (ganeshie8):

why ?

OpenStudy (jango_in_dtown):

f(x)=x-e^-x

OpenStudy (dls):

Don't we represent the given function as x = (something) ?

OpenStudy (jango_in_dtown):

f'(x)=1+e^-x

OpenStudy (jango_in_dtown):

x= something in fixed point iteration

ganeshie8 (ganeshie8):

you're given an "equation" to solve not function @DLS

OpenStudy (dls):

ohh..I see.. I confused both the methods XD thanks for clarification^^ so it should be : f(x) = e^-x - x right?

OpenStudy (jango_in_dtown):

yeah or x-e^-x

OpenStudy (jango_in_dtown):

any of them is correct

OpenStudy (dls):

I see, thanks for clearing :) @ganeshie8 @jango_IN_DTOWN

OpenStudy (jango_in_dtown):

Yeah N-R method will fail only if f'(x) is zero in the nbd of the point you took

OpenStudy (dls):

So this question can't be done with fixed point iteration right ?

OpenStudy (jango_in_dtown):

https://www.quora.com/When-does-Newton-Raphson-fail

OpenStudy (jango_in_dtown):

why not?

OpenStudy (jango_in_dtown):

you check if it satisfies the condition

OpenStudy (dls):

okay yes, the condition of convergence holds true :D

OpenStudy (jango_in_dtown):

then it will work

OpenStudy (dls):

Any tips on how to take the initial approximation btw?

OpenStudy (dls):

Currently I try to graph the given function if possible and find an approximate root..then take a value nearby.

OpenStudy (jango_in_dtown):

use graphical method or tabulation metod

OpenStudy (jango_in_dtown):

if you know where the root will be, say the root lies in between 0 and 1

OpenStudy (amistre64):

f= e^(-x) f'=-e^(-x) y= c y= f'(c) (x-c) + f(c) 0= f'(c) (x-c) + f(c) - c 0= xf'(c) -c f'(c) + f(c) - c xf'(c) = c f'(c) - f(c) + c xf'(c) = c (f'(c)+1) - f(c) x = c (1 + 1/f'(c)) - f(c)/f'(c)

OpenStudy (jango_in_dtown):

then you divide the inteval [0,1] into 10 equal subintervals and check in which interval the sign changes

OpenStudy (amistre64):

|dw:1447596637420:dw|

Can't find your answer? Make a FREE account and ask your own questions, OR help others and earn volunteer hours!

Join our real-time social learning platform and learn together with your friends!
Can't find your answer? Make a FREE account and ask your own questions, OR help others and earn volunteer hours!

Join our real-time social learning platform and learn together with your friends!