I'm studying differential equations and I was trying to solve the following first order differential equation:

$\dot y(t) + \frac{1}{a}y(t)=\sum_i w_i \sum_{t_i} \delta(t-t_i)$

where $w_i$ are constant coefficients. I proceeded solving first the homogeneous equation

$\dot y(t) = -\frac{1}{a}y(t) \implies y(t) = y_0 exp(- \frac{t-t_0}{a})$

To obtain a particular integral I need to solve

$ y_p(t) = e^{-\frac{t}{a}} \int_{t_0}^t exp(\frac{\tau}{a}) \sum_i w_i \sum_{t_i} \delta(\tau-t_i) d\tau = e^{-\frac{t}{a}} \sum_i w_i \sum_{t_i}\int_{t_0}^t\delta(\tau-t_i)exp(\frac{\tau}{a})d\tau $ $ = e^{-\frac{t}{a}}\sum_i w_i \sum_{t_i}exp(\frac{t_i}{a}) = \sum_i w_i \sum_{t_i}exp(-\frac{t-t_i}{a}) $

assuming that $ t_0 <t_i < t ~~~\forall i$.

The overall solution should then be

$y(t) = y_0 exp(- \frac{t-t_0}{a}) + \sum_i w_i \sum_{t_i}exp(-\frac{t-t_i}{a})$

Is this correct? What confuses me is that if I try to substitute the solution I obtained in the original equation I don't get the identity. Sorry if the problem is trivial.


To begin with, $\sum_i w_i \sum_{t_i} \delta(t-t_i)$ does not look like a valid formula (outer summation over $i$, and inside of it, summation over $t_i$?). Perhaps $\sum_i w_i \delta(t-t_i)$ is what was intended.

My preferred interpretation of $\delta(t-t_i)$ in a basic ODE course is that it is the derivative of Heaviside function $H(t-t_i)$. More generally, for any continuous function $f$ we have $$ f(t) \delta_(t-t_i) = f(t_i) \delta_(t-t_i) = \frac{d}{dt}(f(t_i) H(t-t_i)) \tag{1}$$ where the first step is justified by $\delta$ being concentrated at $t_i$, and the second step is linearity of derivative.

Apply (1) to your situation as
$$e^{\tau/a} w_i \delta(\tau-t_i) = \frac{d}{dt}\left(e^{t_i/a} w_i H(t-t_i)\right)\tag{2}$$ Therefore,
$$ \begin{split} e^{-t/a} \int e^{\tau/a} \sum_i w_i \delta(\tau-t_i) d\tau &= e^{-t/a}\left( C + \sum_i e^{t_i/a} w_i H(t-t_i)\right) \\ & = Ce^{-t/a} + \sum_i e^{(t_i-t)/a} w_i H(t-t_i) \end{split}\tag{3}$$ Formula (3) gives the general solution of your ODE. You can satisfy the initial condition by choosing $C$.

You can also use definite integral ($t_0$ to $t$) in (3), but I thought that indefinite integral is easier to write down.


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.