I'm looking at the following differential equation:

$\frac{dx}{dt} = \frac{\sin^2 x - t^2}{t \cdot \sin(2x)}$

I rewrote it as

$(t \cdot \sin(2x))dx = (\sin^2x - t^2)dt$

$\Leftrightarrow (\underbrace{\sin^2 x - t^2}_{J(x,t)})dt + (\underbrace{-t \cdot \sin(2x)}_{I(x,t)}) dx = 0$

where I put the minus sign inside the function because I think it is important that a plus stands between the two functions in order to check if a differential equation is exact or not. But then:

$\partial_x J(x,t) = 2 \sin x \cos x = \sin(2x)$

$\partial_t I(x,t) = -\sin(2x)$

Those functions don't seem to satisfy the condition for a differential equation to be exact. However my teacher wrote in his answer the following:

$\frac{-\partial_x (\sin^2 x - t^2) - \partial_t (t \cdot \sin (2x))}{t \sin(2x)} = \frac{-2}{t}$

$\implies c' = - \frac{2}{t} c$

$\implies x' = \frac{\frac{\sin^2 x}{t^2} - 1}{\frac{\sin (2x)}{t}}$

and he states that this equation is now exact. What happened there? I've never seen anything like that yet. I understand (more or less) what he is doing after $c' = -\frac{2}{t}c$ (separation of variables and integration), but how he came to the first line I have no idea. Is that a known technic to transform a differential equation that is $\textit{almost}$ exact?

Thanks a lot in advance for your answers.



So if the differential equation of the form $Mdt+Ndx=0$, for $C^1$ functions $M=M(x,t)$ and $N=N(x,t)$ on some rectangular region $R\subseteq\Bbb R^2$, satisfies $M_x=N_t$, then it is an exact equation. If it is not of that form, sometimes one can find an integrating factor $\mu$ (denoted in your problem by $c$) to multiply the equation, so that it becomes exact. There are methods for finding integrating factors in specific case. One specific case is if $\frac{M_x-N_t}{N}$ is a function of only $t$, then one can find an integrating factor $\mu=\mu(t)$ which is only a function of $t$, and satisfies the differential equation $\frac{d\mu}{dt}=\frac{M_x-N_t}{N}\mu$.

In this case, we see that $N(x,t)=-t\sin(2x)$ and $M(x,t)=\sin^2(x)-t^2$, so you can find that $\frac{M_x-N_t}{N}=\frac{-2}{t}$, which is only a function of $t$. So then you can find $\mu(t)$ as a solution of $\frac{d\mu}{dt}=\frac{-2}{t}\mu$ (note that it is easiest to take the solution which has an integration constant equal to zero). Then you simply multiply the equation to get $\mu Mdt+\mu Ndx=0$, and you can verify that this equation is exact by showing $\frac{\partial}{\partial x}(\mu M)=\frac{\partial}{\partial t}(\mu N)$.

  • $\begingroup$ Alright thank you both for your quick and precise answers. I'm starting reading about the topic now. $\endgroup$ – Jxx Dec 3 '16 at 21:18

His $c$ is an integrating factor which he's picking as a function of $t$ only. If you multiply everything through by $c(t)$ and compare your same partial derivatives, you'll get $$2c(t)\sin(2x) = -tc'(t)\sin(2x).$$


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.