I'm working on a project and I've come across a seemingly standard ODE, however I have no idea how to solve it. The equation in question is $$ \epsilon'' + \left[g(\tau) + \frac{\beta}{\tau}\right]\epsilon' + \left[\frac{\gamma}{\tau^{4/3}}\right]\epsilon = 0 $$ where $g(\tau) = \alpha \cos(\tau)$ or $\delta\cos(\tau)\sin(\tau)$ (there are two cases). I've tried a series solution but unsurprisingly it's pretty messy and doesn't really solve anything, and I've tried Laplace transforms but unfortunately they're divergent for negative powers of $\tau$.

Does anybody have recommendations for methods I can try, or even how to make a decent ansatz? This is just a step in my project and it's really presenting a roadblock for me, so any help is appreciated.


You have a linear time-variant differential equation. You can cast it into matrix form by introducing $\epsilon = \epsilon_1$, $\epsilon'=\epsilon_1'=\epsilon_2$. You will get

$$\dfrac{d}{d\tau}\begin{bmatrix} \epsilon_1\\\epsilon_2 \end{bmatrix}=\begin{bmatrix} 0 & 1\\-\dfrac{\gamma}{\tau^{4/3}} & -g(\tau)-\dfrac{\beta}{\tau} \end{bmatrix}\begin{bmatrix} \epsilon_1\\\epsilon_2 \end{bmatrix}.$$

In this form, you can invoke the Peano-Baker series to obtain the solution

$$\boldsymbol{\epsilon}(\tau)=\left[\boldsymbol{I}+\int_{\tau_0}^{\tau}\boldsymbol{A}(\tau_1)d\tau_1+\int_{\tau_0}^\tau\int_{\tau_0}^{\tau_1}\boldsymbol{A}(\tau_1)\boldsymbol{A}(\tau_2)d\tau_2d\tau_1+... \right]\boldsymbol{\epsilon}(\tau=\tau_0),$$

in which

$$\boldsymbol{A}(\tau)=\begin{bmatrix} 0 & 1\\-\dfrac{\gamma}{\tau^{4/3}} & -g(\tau)-\dfrac{\beta}{\tau} \end{bmatrix},$$

$\boldsymbol{I}$ is the $2\times 2$ identity matrix, and $\boldsymbol{\epsilon}=[\epsilon_1,\epsilon_2]^T$.

If there is a closed form solution you should be able to retrieve it with this method.


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.