Current CABot1s
- We decided that the Izhikevich model was the wrong one to follow.
- We just inhibit things too much, and depend on inhibited things not
firing.
- So, we switched to the IF_Curr_Exp model. That's a Leaky Integrate
and Fire model.
- We hacked it pretty quickly, and got a Nest simulation running.
- We then tried to port it to the SpiNNaker emulator.
Unfortunately, that has problems with more than 5000 neurons.
- We're just finishing getting it to the SpiNNaker chips.
- One problem with this is that packing the chips gives each
population a processor. We wrote planning and parsing with
lots of populations.
- Parsing is not in CABot1, but works and is regular so no binding.
- We now need to put it into the closed loop with the new environment
we're writing.
