3
$\begingroup$

I am unable to solve this nonlinear differential equation: $$\dot{x}(t)^2+x(t)=\sin(x)^2$$ I tried with Maple without success. Is it possible to solve it without the use of numerical methods? Thanks.

$\endgroup$
  • $\begingroup$ I suspect that a closed formula is out of reach. You can make some phase-plane analysis, however. $\endgroup$ – Siminore Dec 9 '13 at 12:21
5
$\begingroup$

If $x(t)$ is of class $C^2$ you can differentiate the equation to get $$2\dot{x} \ddot{x}+\dot{x}=\sin(2x) \dot{x} $$

$\dot{x}=0$ gives constant solutions $x(t)=x_0$, which by the original ODE must satisfy $x_0=\sin(x_0)^2$, thus reducing to $x_0=0$ alone.

Other solutions satisfy $$2 \ddot{x}=\sin{2x}-1 $$

and this equation seems to be within the capability of Maple (an implicit solution can be obtained).

Another approach could be separating the variables: $$\dot{x}^2=\sin^2 x-x \\\dot{x}= \pm \sqrt{\sin^2{x}-x} \\ \frac{dx}{\sqrt{\sin^2 x-x}}=\pm dt \\ \int \frac{dx}{\sqrt{\sin^2 x-x}}=\pm(t-t_0)$$ Which also gives an implicit solution.

$\endgroup$
1
$\begingroup$

Note that $\sin^2x -x\geqslant0$ if and only if $x\leqslant0$ hence one can rewrite this as $\dot x=\varepsilon u(x)$ on every interval where $x\lt0$, with $u:\xi\mapsto\sqrt{\sin^2\xi-\xi}$ and $\varepsilon=\pm1$. Thus, assuming that $x(0)=x_0$ with $x_0\lt0$, $$ \int_{x_0}^{x(t)}\frac{\mathrm d\xi}{u(\xi)}=\varepsilon t, $$ for every $t$ such that the LHS yields $x(t)\lt0$. This defines implicitely $x(t)$ for every $t$ such that $x(t)$ exists. Two cases arise. If $\varepsilon=+1$, then $x$ is increasing and $x(t)\to0$ when $t\to\tau(x(0))$, where, for every $y\lt0$, $$ \tau(y)=\int_{y}^{0}\frac{\mathrm d\xi}{u(\xi)}, $$ thus, $x$ is defined on $[0,\tau(x_0))$. If $\varepsilon=-1$, then $x$ is decreasing and $x(t)\to-\infty$ when $t\to+\infty$ because the integral $$ \int\frac{\mathrm d\xi}{u(\xi)} $$ diverges at $-\infty$. Both cases are actually the same, since the maximal interval of definition of $x$ is $(-\infty,\tau(x_0))$ when $\varepsilon=+1$ and $(-\tau(x_0),+\infty)$ when $\varepsilon=-1$. Thus, one follows the same and unique solution, starting at some point given by $x_0$, either in the direction where the solution stops to be defined after a finite time or in the direction where it is defined forever.

Said differently, there exists a unique increasing function $\gamma:(-\infty,0)\to(-\infty,0)$ such that $\gamma(0-)=0-$ and $\gamma(-\infty)=-\infty$, and, for every $x_0\lt0$ and every $t$ such that $x(t)$ exists, $$x(t)=\gamma(\varepsilon t+\gamma^{-1}(x_0)), $$ either with $\varepsilon=+1$ or $\varepsilon=-1$. And an explicit formula for $\gamma$ is $$ \int_{\gamma(t)}^0\frac{\mathrm d\xi}{u(\xi)}=-t. $$

$\endgroup$

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.