The Dirac function is commonly used to model the action potential of neurons, and in the text I am reading the "neural response function" is defined as:
$$\rho(t) = \sum_{i=1}^n \delta(t - t_i)$$
The author then follows to say that any well-behaved function h(t) can be expressed as such:
$$\sum_{i=1}^n h(t - t_i) = \int_0^T h(\tau)\rho(t-\tau)d\tau$$
I am confused on two points
Wouldn't the first function simply be 1 for all $t$, since the integral of the delta function is 1 by definition? If so, I do not understand the purpose of the function $\rho(t)$.
Related to my confusion on the first point I am simply not following this identity. I understand the left half of the equation to essentially be an integral of the shifted function $h(t)$, but then the right hand side is the integral of $h(\tau)$ multiplied by $\rho(t-\tau)$, the latter of which is one for every $\tau$. So as I follow it the right hand side is simply the integral of $h(\tau)$ and the $\rho(t-\tau)$ contributes nothing.
Does anyone understand this enough to explain why this identity is true or useful?
Thank you!
You could convince yourself the identity works as follows:
$$\int_0^T h(\tau) \rho (t - \tau) \mathrm{d} \tau = \int_0^T h(\tau) \sum_{i=1}^n \delta (t - t_i - \tau) \mathrm{d} \tau = \\ \int_0^T h(\tau) \delta (t -t_1 - \tau) \mathrm{d} \tau + \int_0^T h(\tau) \delta (t -t_2- \tau) \mathrm{d} \tau +\cdots +\int_0^T h(\tau) \delta (t -t_n- \tau) \mathrm{d} \tau = \\ h(t -t_1) + h(t-t_2) + \cdots h(t-t_n) = \sum_{i_1}^n h(t-t_i)$$
hopefully this hints at how the "Dirac comb", the sum of Dirac deltas, is sampling out values in the integral and yielding the sum with the function $h$