Thread: calculus- plz help..

If f(x)=0 has multiple roots,how is N-R method affected?How can N-R formula be modified to avoid this problem

is N-R for Newton's method ? i.e. x_(n+1) = x_n-(f(x_n)/(f'(x_n))
dan

If you mean Newton-Raphson then multiple roots is a common problem of any iterative method. Different starting values may give different solutions. Sketching the graph gives you a good idea of what starting values to try but the shape of the graph will determine how good N-R is and in some situations it doesn't work at all.

if f(x)=0 then f'(x)=0 so N-R method would give you x_n-(0/0) so i dont think you can use it

That isn't true

eg f(x) = 2x - 2

= 0 at x = 1

f'(x) = 2 at all points not 0

you said f(x)= 0 i.e f(x)= x-x or say f(x)=(x/(x^2))-(1/x) in any case f'(x)=0

He means solutions to the equation f(x) = 0 not define f(x) = 0. f(x) has not been defined. He wants a refined method for a general solution.

Originally Posted by bobby77

If f(x)=0 has multiple roots,how is N-R method affected?How can N-R formula be modified to avoid this problem

For simplicity sake, say that is a differenenciable function. Select an interval such as, , i.e. they have opposite signs, and . Then by the intermediate value theorem there is a solution on this interval. Furthermore, since the function is either strictly increasing or strictly deceasing (but not both) implies there cannot be another solution on this interval. Thus, if the function satisfies the condition for the Newton-Raphson algorithm then it will converge to that root.
----
Here is a list of steps.
1)Find an interval such as, . (Gaurentees existence of a root).
2)Confirm that has only one sign. If not try chaning step #1 slightly to adjust step #2. (Gaurentees uniquesness of a root).
3)Select any point on .
4)Algorithm will converge to that solution on that interval..

Originally Posted by Glaysher

He means solutions to the equation f(x) = 0 not define f(x) = 0. f(x) has not been defined. He wants a refined method for a general solution.

a ray of light... thanks i missed his point from the start
dan