What, exactly, do you mean by 'fixed point' in this context ?
What is it about the two values that you give that makes them fixed points ?
Ok so I have a function
the function has two real fixed points in the interval [-1,1]. One is -1 and the other is approx 0.8504
My question is can we expect the fixed point iteration converge to either of these fixed points? One, Both, None?
In case that , that is for all then you can generate a sequence,
starting with some point , and define
If the fixpoint p is unique in and for then as . (wich is a theorem )
However in this case has 2 fixpoints in [-1,1]...and the problem is slightly different. So we need to find out for which , the sequence converges to 1 or to 0.8504...
But we may not even expect it to converge...since only for
And I'm still trying to find out myself
My point was that the first of these values cannot possibly be correct and that the second (approximation) is not too good either.One is -1 and the other is approx 0.8504
The 'function' is even, so any zeros will appear in pairs equidistant from the origin.
For a given 'function' there will be an infinite number of fixed point iterations, some of which converge to a particular fixed point and some that do not. To refer to the fixed point iteration doesn't make sense unless we are told what the iteration is.
??? Unless the original post has been edited, -1 certainly is a fixed point: . And 0.8504 is a fixed point to 4 decimal places, which is all that is claimed.
Yes, the function is even, but we are not talking about zeroes, we are talking about fixed points: f(x)= x. is the same as which is neither even nor odd.
Yes, there are many different iterations that can be called "fixed point iterations". But the one that is used, for example, in the "Banach fixed point theorem" can probably be called the "standard" and is the simplest- to find a fixed point for f(x), that is, to solve f(x)= x, start with some and define . If that converges, it will converge to a fixed point of f.
ductiletoaster, you can always find some region around a fixed point such that that iteration converges to the given fixed point for all points in the region. Of course, it may be very small and there may also be points far from the given fixed point for which that iteration converges. In general, the set of all points a which this iteration converges to a fixed point is fractal. I don't know of any simple way of characterizing it.
However, is it true to say that 'you can always find some region around a fixed point such that the iteration converges to the given fixed point for all points within the region' ? I thought that it was necessary that was a contraction mapping, and that doesn't seem to be the case for this example ?
Let be a fixpoint, then
Can we find small enough such that the above inequality holds for both fixpoints?? And if so...i doubt this is true for all f..
I think f must be a contraction mapping...
edit: for ... this is not true
This means that will definitely not converge to for any starting point in
SO to confirm. U guys are saying that the function will not converge on the interval (-1,1) to either of the values -1 or .8504?
When i did the fixed point iteration my x value kept alternating between almost 1 and almost 0. and it seemed it would continue forever.
I appreciate all of your guys help and discussions
The way you have set up your iteration scheme, it will not converge, because, as has been mentioned, the function is not a contraction. However, you're not confined to one iteration scheme. Try this one:
So set up your function
and look for fixed points of that. The fixed points of this new function will be the fixed points of the old, and this one appears to be more stable. I think you could even prove that it's a contraction mapping.
Thank You all for your help on this proble! Really Appreciate it.
Dinkydoe you gave a great explanation to why it wont converge toward .8504. Can that same reasoning be applied to why it wont converge to -1. I guess im a little confused on the reasons for it. Im trying to get as simple as an explanation as possible. Im pretty new to this stuff. Thanks agian
Also how would i show / prove that this particular function in the form will not converge?
Ok so now im really confused. Everyone here says it wont converge to either -1 or .8504. But when using maple 13 I got it to converge to -1 using several random points. I think people are forgetting about the Interval [-1,1].... I belive it does converge on that interval but not on some value outside that interval
If a function is a contraction mapping on a complete nonempty metric space, then the Banach Fixed-Point Theorem says that the function has one unique fixed point. Furthermore, the iterative sequence converges to the fixed point.
Here's the upshot: contraction mappings will converge to the fixed point using the iteration scheme I just mentioned. With non-contraction mappings, you're not guaranteed anything. It might converge, it might not.
What commands did you use, exactly?But when using maple 13 I got it to converge to -1 using several random points.
Might be alluding to round-off error. But that would also depend on what is meant by "by hand". For me, in this situation, I'd probably use a calculator, in which case round-off error would presumably be similar to what it would be on a computer. Your professor's statement is certainly non-intuitive for me. I'd be curious to know more about what he meant.Also my proffesor said that if you were to code this up in matlab u could expect it to have the oppsiite behaviour as if u did it by hand?
Tried some values for , for the function and I got for every small value I tried, that
This is not a proof ofcourse, but it makes it very likely to diverge....For a=0.8504 divergence can pretty easily be proved. (as I partially did).
But for It's hard to estimate the term with a good lower-bound.
For a=0.8504 also must be shown. (wich is true..but a little harder to show)
So my conclusion is..., there's no convergence at all! Except for the values