# Chaos Theory Problem

• Apr 19th 2010, 05:58 PM
Khonics89
Chaos Theory Problem
The one-dimensional mapping x(n+1) =ax^2(n)(1 − x(n))
has been used as a toy model for the sunspot cycle.
0 < x(n) < 1 and 0 < a< 6.75; here x(n) is the phase space variable, and a
is a
parameter.

(a) Show algebraically that there are one, two or three fixed points, giving
the values of
corresponding to each case. State the type of bifurcation
that occurs at
a = 4.

(b) Calculate the derivative of the RHS function at the fixed point(s). Hence
show that the fixed point at the origin is always stable. Show further
that the fixed point obtained by taking a negative square root is always
unstable, but that the one that features a positive square root is stable
for 4 < a < 16/3. By noting which stability inequality is violated at

a = 16/3, state what kind of bifurcation occurs at this value, and what
behavior you think will ensue just above it.

This is one of my past exam papers that i'm currently studying but I don't know how to start it off. I've done part a except "state the type of bifurcation at a=4".

If you could give me the steps necessary it would be great :)
• Apr 20th 2010, 05:06 AM
Can you put up what you've done next time...

skipping forward a bit...

Let $F_a(x) = ax^2(1-x)$

Have,

If $a<4$, 1 fixed point $x=0$.

If $a>4$, 3 fixed points at $x=0$ and $\frac{1}{2} \pm \sqrt{1/4- 1/a}$.

If $a=0$, 2 fixed points at $x=0$, 1/2.

We consider what happens when a=4.

Now without doing any calculations I can tell you it's a Saddle node bifurcation. Why?

Definition...

For $\mu \in (a_c - \epsilon, a_c)$, $F_a$ has no fixed points in the interval $I$.
(ii) At $a = a_c$, $F_a$ has one fixed point in $I$ at $x_c$ and $F'_{a_c}(x_c) = 1$.
(iii) For $a \in (a_c, a_c+\epsilon)$, $F_a$ has two fixed points in $I$, one attracting and one repelling.

Where the c denoted 'critical' value. (a=4 and x=1/2 in this case)

So looking at that.

Condition i) is satisfied since if a<4 we get no real fixed points near x=1/2.

Condition ii) is satisfied since we get a single fixed point at x=1/2 when a=4.

Condition iii) is satisfied since when a>4 two fixed points appear. I haven't calculated stability yet but you can do that to check.

So lets us find the derivative at x=1/2 to see what we get just to be sure...

$F'_a(x) = 2ax(1-x)-ax^2$.

Evaluate this at $x=1/2$ and $a=4$ and we get...

$|F'_a(x)| = |8x - 12x^2|$

=> $|F'_a(1/2)| = |4-3| = 1$.

Use the absolute value of the derivative to prove stability/unstability at the fixed points when $a>4$ if you wish to double check that it is in fact, a saddle-node bifurcation.

$|F'_a(x)|<1$ => Stable

$|F'_a(x)|>1$ => Unstable
• Apr 20th 2010, 04:39 PM
Khonics89
Thank alot, and my apology for not putting in my working out.
• Apr 20th 2010, 04:53 PM
Quote:

Originally Posted by Khonics89
Thank alot, and my apology for not putting in my working out.

It's cool. Just speeds things up if we have things like fixed points. Can get straight onto the stuff you're stuck on.

For the second part...

I haven't done the second part but I would think (based purely on a guess) that it would be flip bifurcation. I would imagine the stable point will become unstable and a period 2 cycle will appear when a>16/3. You'll have to test that though...
• Apr 20th 2010, 05:26 PM
Khonics89
Quote:

It's cool. Just speeds things up if we have things like fixed points. Can get straight onto the stuff you're stuck on.

For the second part...

I haven't done the second part but I would think (based purely on a guess) that it would be flip bifurcation. I would imagine the stable point will become unstable and a period 2 cycle will appear when a>16/3. You'll have to test that though...

I think your right, period doubling occurs at 16/3 :)

By any chance do you have any website that will help me get exposed to many problems similar to this?
• Apr 20th 2010, 05:43 PM
Quote:

Originally Posted by Khonics89
I think your right, period doubling occurs at 16/3 :)

By any chance do you have any website that will help me get exposed to many problems similar to this?

Afraid not. I just do them from my uni stuff. Perhaps hunt around for some e-books or something.
• Apr 20th 2010, 05:45 PM
Khonics89
Quote:

It's cool. Just speeds things up if we have things like fixed points. Can get straight onto the stuff you're stuck on.

For the second part...

I haven't done the second part but I would think (based purely on a guess) that it would be flip bifurcation. I would imagine the stable point will become unstable and a period 2 cycle will appear when a>16/3. You'll have to test that though...

Quote:

Afraid not. I just do them from my uni stuff. Perhaps hunt around for some e-books or something.

Are you studying chaos theory at uni ?
• Apr 20th 2010, 05:51 PM
Quote:

Originally Posted by Khonics89
Are you studying chaos theory at uni ?

Not exclusively.

I do a dynamical systems class, a mathematical biology one (which is all about fixed points and stuff, no chaos though) and I did a project on the saddle-node bifurcation last year.

Also did a project on fractals this year and hoping to write my first (own authored) paper on them this summer...

Will be hoping to do a masters next year and will do a lot of chaos/bifurcation/dynamical systems stuff I think.