$\require{amsmath}$ $\DeclareMathOperator{\Tr}{Tr}$ $\DeclareMathOperator{\Det}{Det}$

Investigate the stability of the equilibrium points of the system $\ddot{x}=(x-a)(x^2-a)$ for all real values of the parameter $a$. (Hints: It might help to graph the right-hand side. An alternative is to rewrite the equation as $\ddot{x}=−V′(x)$ for a suitable potential energy function $V$ and then use your intuition about particles moving in potentials.)

I am not really sure on how to approach the problem with the given hints, since it would require plotting the graph for different critical values for $a$, which I am not really sure how to find. Thus, I am wondering if the following is correct.

The system $\ddot{x}=(x-a)(x^2-a)$ can be re-written as $$\begin{cases} \dot{x}=y\\ \dot{y}=(x-a)(x^2-a) \end{cases}$$ with fixed points $P_1(a,0),\,P_2(\sqrt{a},0)$ and $P_3(\sqrt{a},0)$.

The Jacobian is $$J(x,y)=\begin{bmatrix} 0 &1\\ 3x^2-2ax-a &0 \end{bmatrix}$$ and thus $$J(a,0)=\begin{bmatrix} 0 &1\\ a^2-a &0 \end{bmatrix};\quad J(\sqrt{a},0)=\begin{bmatrix} 0 &1\\ 2a-2a^{3/2} &0 \end{bmatrix}; \quad J(-\sqrt{a},0)=\begin{bmatrix} 0 &1\\ 2a+2a^{3/2} &0 \end{bmatrix}.$$ It can be noticed that the $\Tr\left[J(x,y)\right]=0$ and that $$\begin{aligned} &1.\,\Det\left[J(a,0)\right]=a(1-a)\implies \text{Saddle for }a<0 \wedge a>1, \text{Center for }0<a<1.\\ &2.\,\Det\left[J(\sqrt{a},0)\right]=2a(\sqrt{a}-1)\implies \text{Saddle for }0<a<1, \text{Center for }a>1.\\ &3.\,\Det\left[J(-\sqrt{a},0)\right]=-2a(\sqrt{a}+1)\implies \text{Saddle for }a>0. \end{aligned}$$

If $a=0$ the system reduces to $\ddot{x}=x^3$ where the only fixed point is at $(0,0)$ and thus it is unstable. On the other hand, if $a=1$ then the system reduces to $\ddot{x}=x^3-x^2-x+1$ with fixed points at $(1,0),(-1,0)$, both being unstable.

Is my work correct?

  • $\begingroup$ Isn't this question better suited for Physics SE? $\endgroup$ – Bertrand Wittgenstein's Ghost Nov 29 '18 at 2:11
  • 4
    $\begingroup$ @BertrandWittgenstein'sGhost this is part of dynamical systems, so no. I could certainly ask there too, but there is no point really $\endgroup$ – DMH16 Nov 29 '18 at 2:13
  • $\begingroup$ That's funny what you did, It was a genuine question. No need to get defensive. Cheers! $\endgroup$ – Bertrand Wittgenstein's Ghost Nov 29 '18 at 2:23
  • $\begingroup$ I haven't checked all your algebra carefully but it looks like you've go the right idea! $\endgroup$ – Robert Lewis Nov 29 '18 at 2:32
  • 1
    $\begingroup$ @AlexanderJ93 yes you're right the determinant is always negative. Thanks $\endgroup$ – DMH16 Nov 29 '18 at 2:52

Almost everything looks right.

For $a > 0$, $-2a < 0$ and $\sqrt{a}+1 > 0$, so $\Det[J(-\sqrt{a},0)] < 0$. This equilibrium should be a saddle then for positive $a$.

Also, when analyzing the stability of a system with a parameter, it helps to draw the bifurcation diagram, i.e. the graph of the fixed points with respect to the parameter. The fixed point equations $x = a, x = \sqrt{a}, x = -\sqrt{a}$ can be graphed on the $ax$-plane. Intersections of these curves are the bifurcations, which are the only places where the behavior can change.

bifurcation diagram

This is the bifurcation diagram for your system. The lower black curve is the $x=-\sqrt{a}$ equilibrium, which doesn't have any intersections for $a>0$, so there will be no change in behavior.


This isn't specifically asked about in the question but it is an important feature of this equation. For $a>0, a\neq 1$, there are centers at one fixed point and saddles at another nearby fixed point. When this happens, a periodic orbit will collide with the saddle point, creating a closed path from a fixed point back around to itself. This is called a homoclinic orbit, and usually has a "teardrop" shape, as seen below for $a = \frac{1}{2}$.

homoclinic orbit

  • $\begingroup$ Just out of curiosity, what software did you use to sketch the phase plane? $\endgroup$ – DMH16 Nov 29 '18 at 4:49
  • 1
    $\begingroup$ pplane, which can be found here as a java application: math.rice.edu/~dfield/dfpp.html I strongly recommend it to any student in dynamics. It's limited to plane dynamics (obviously) but can be extremely helpful in learning and verifying results. $\endgroup$ – AlexanderJ93 Nov 29 '18 at 5:03
  • $\begingroup$ Thanks you for the suggestion! $\endgroup$ – DMH16 Nov 29 '18 at 5:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.