Problem with implementing large synaptic delays

[Jian Carlo Nocon]

Hi all, I'm trying to implement a large synaptic delay between two populations of neurons in Dynasim, with the pre-synaptic cells having an effective refractory period of 0.5 ms. Here's how I implemented the synaptic current:

tau_rise = tauD*tauR/(tauD-tauR)
b = ((tauR/tauD)^(tau_rise/tauD) - (tauR/tauD)^(tau_rise/tauR))^-1
f(t) = b * ( exp(-t/tauD) - exp(-t/tauR) ).*(t>=0)
syn(X,t) = gSYN .* ( f(t - tspike_pre - delay) * netcon ) .* (X - ESYN)

When I used a large synaptic delay (10ms), I found that the post-synaptic current  shunted to 0 for the whole 10ms, even if there was current from a previous pre-synaptic spike. I also found that if multiple spikes occur within the delay, only the last spike gives a post-synaptic current. I'm certain this is because t < tspike_pre - delay makes f(t) = 0.

One solution that I tried was creating a dummy variable called tnew, which stores the time at which a pre-synaptic spike should reach the post-synaptic cell:

syn(X,t) = gSYN .* ( f(t - tnew) * netcon ) .* (X - ESYN)

if(any(t == tspike_pre + delay,1))(tnew = t)

I ran into the same problem with this approach, as post-synaptic current is only created if there are no other spikes within that delay period.

Is there a way for me to create multiple EPSCs using a large delay? I'm aware I can adjust the buffer size of the spike matrices, but I don't know how I would use that to solve my problem.

- Jio