Given the following ODE which is supposed to represent an undamped pendulum, with x representing the vertical angle:

$$\frac{d^2x}{dt^2}= -2\sin(2x)$$

Make the substitution $$\frac{dx}{dt}= y $$

And find an implicit solution in the x-y phase plane?

Note: I do know how to find the particular solution after substituting $x = A\sin(2x)+B\cos(2x)$ into the first differential equation, and then computing the values of $A$ and $B$ that would lead to the right hand side of the equation. I am completely unsure how to find the implicit solution in the $x$-$y$ phase planes and would greatly appreciate any help in that regard. Expressing the first differential equation in terms of y should lead to the following $$\frac{dy}{dt}= -2\sin(2x)$$

Beyond that, I am completely lost.

  • $\begingroup$ Use conservation of energy $\endgroup$ – Michael Galuza Jul 20 '15 at 17:12
  • $\begingroup$ mathematicalgarden.wordpress.com/2009/03/29/nonlinear-pendulum $\endgroup$ – Hans Lundmark Jul 20 '15 at 17:14
  • $\begingroup$ By the way, your comment about $A \sin 2x + B \cos 2x$ is not right. You would get a solution of that form if the left-hand side of the equation were $d^2 y/dx^2$, but here it's $d^2 x/dt^2$, so it's not even a linear equation. $\endgroup$ – Hans Lundmark Jul 20 '15 at 17:18

Let $$ E=\frac{\dot x^2}{2} - \cos{2x}. $$ (it's energy, if $m=1$). Hence $$ \frac{dE}{dt} = \dot x \ddot x + 2\dot x\sin{2x}= \dot x(\ddot x + 2\sin2x) = 0 $$ So, if we denote $y=\dot x$, we have $$ y^2 - 2\cos2x = 2E. $$ It's implicit solution. You may draw it on $xy$-plane. Here is plot for various $E$ ($E=0, 1,\ldots,6$): $\hskip.5in$enter image description here

| cite | improve this answer | |
  • $\begingroup$ I'm afraid this has not been taught before. What does E represent and how was it derived? $\endgroup$ – abstract Jul 20 '15 at 18:01
  • 1
    $\begingroup$ @abstract $E$ is the total energy of the system. It consists of the kinetic energy ($mv^2/2$, where $v$ is the speed) and the potential energy ($mgh$, where there is a constant gravitational acceleration $g$ and $h$ is the height of the pendulum above the reference height). $\endgroup$ – Ian Jul 20 '15 at 19:12
  • $\begingroup$ @Ian $mgh$, really? $\endgroup$ – Michael Galuza Jul 21 '15 at 3:39
  • $\begingroup$ @MichaelGaluza I don't understand your comment. Are you objecting to my presentation, asking whether it really is from gravitational potential in the constant acceleration limit, or what? It is in fact $mgh$, there are just some constants hidden in the choice of units. $\endgroup$ – Ian Jul 21 '15 at 13:05
  • $\begingroup$ @Ian I talked about that there is no gravitation or smth like that. There is just a non-linear pendulum, and its potentail energy is $-\cos2x$ $\endgroup$ – Michael Galuza Jul 21 '15 at 13:14


Find $\frac{dy}{dx}$ by dividing the two equations, then solve the ODE by separation of variables.

The system of differential equations is: $$\frac{dy}{dt}=-2\sin(2x)\quad\quad(1)\\ \frac{dx}{dt}=y \quad\quad\quad (2)\\$$

Divide (1) by (2) $$\frac{dy}{dx}=\frac{-2\sin(x)}{y}$$

Then you can solve it using separation of variables.

| cite | improve this answer | |
  • $\begingroup$ what will you solve ? you'll just get an equation $x'(t) = f(x(t))$, but beside $x(t) =0$, there is no closed form solution nor for $x(t)$ nor for $x'(t)$, maybe an analytic series or another type or series ? $\endgroup$ – reuns Jul 20 '15 at 17:50
  • $\begingroup$ @reuns: Solve for $y$ in terms of $x$. I'll explain a little bit more. $\endgroup$ – KittyL Jul 20 '15 at 18:06
  • $\begingroup$ that what I wrote, $y = x' = f^{-1}(x)$ $\endgroup$ – reuns Jul 20 '15 at 18:12
  • 1
    $\begingroup$ @reuns: I am not sure what you meant. The OP wanted an implicit function. This ODE will give him an explicit function $y=f(x)$ as desired. $\endgroup$ – KittyL Jul 20 '15 at 18:14
  • $\begingroup$ no what is really desired is an expression for $x(t)$ knowing starting conditions. it's a differential equation, and would be "solving the equation". because it's impossible, he will study the phase space of $(x', x)$ $\endgroup$ – reuns Jul 20 '15 at 18:14

It is a non-linear pendulum that has to be solved in terms of elliptic integrals. Or else solve it numerically before plotting phase portrait $ ( \dot x - x)$ for given boundary conditions.

$$\frac{d^2\, (2x) }{dt^2}+ 4\sin(2x) = 0 $$

| cite | improve this answer | |

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.