Neural Models: FLIF, RF, and IF

Unfortunately PyNN, and more importantly SpiNNaker, does not have our FLIF model.
Both, and HICANN, provide us the ability to put in our new model, but we've not tried that.
Our FLIF model was a 10ms one, and SpiNNaker wants a 1ms model.
So, we used the standard Izhikevich model with 1ms time steps.
There are other standard models, including Hodgkin Huxley models, but we thought Izhekivch would be relatively easy.
Unfortunately, we didn't read the fine print and thought that the Izhikevich model was Integrate and Fire, when it's really reverberate and fire.
That means if you inhibit a neuron enough, it fires.
One thing that we found odd was that the synaptic models are really complex.
In our FLIF model, a neuron fires, and the next (10 ms) step the next neuron gets the activity.
The only thing that is complex is the learning.
The PyNN models all use complex dynamics of transfer so the post-synaptic neuron gets activity over many (1 ms.) time steps.