integrated alpha-syn, cleaned up exp-syn

Author: Alexander Ororbia

docs/modeling/synapses.md

@@ -74,6 +74,20 @@ This (chemical) synapse performs a linear transform of its input signals. Note t
     :noindex:
 ```
 
+### Alpha Synapse
+
+This (chemical) synapse performs a linear transform of its input signals. Note that this synapse is "dynamic" in the sense that its efficacies are a function of their pre-synaptic inputs; there is no inherent form of long-term plasticity in this base implementation. Synaptic strength values can be viewed as being filtered/smoothened through a kernel that models more realistic rise and fall times of synaptic conductance..
+
+```{eval-rst}
+.. autoclass:: ngclearn.components.AlphaSynapse
+  :noindex:
+
+  .. automethod:: advance_state
+    :noindex:
+  .. automethod:: reset
+    :noindex:
+```
+
 ### Short-Term Plasticity (Dense) Synapse
 
 This synapse performs a linear transform of its input signals. Note that this synapse is "dynamic" in the sense that it engages in short-term plasticity (STP), meaning that its efficacy values change as a function of its inputs/time (and simulated consumed resources), but it does not provide any long-term form of plasticity/adjustment.

ngclearn/components/synapses/alphaSynapse.py

@@ -90,10 +90,10 @@ def advance_state(
         s = inputs
         ## advance conductance variable
         _out = jnp.matmul(s, weights) ## sum all pre-syn spikes at t going into post-neuron)
-        dhsyn_dt = _out * g_syn_bar - h_syn/tau_syn
+        dhsyn_dt = -h_syn/tau_syn + _out * g_syn_bar
         h_syn = h_syn + dhsyn_dt * dt ## run Euler step to move intermediate conductance h
 
-        dgsyn_dt = -g_syn/tau_syn + h_syn/tau_syn
+        dgsyn_dt = -g_syn/tau_syn + h_syn # or -g_syn/tau_syn + h_syn/tau_syn
         g_syn = g_syn + dgsyn_dt * dt ## run Euler step to move conductance g
 
         i_syn =  -g_syn * (v - syn_rest)

ngclearn/components/synapses/exponentialSynapse.py

@@ -88,7 +88,7 @@ def advance_state(
         s = inputs
         ## advance conductance variable
         _out = jnp.matmul(s, weights) ## sum all pre-syn spikes at t going into post-neuron)
-        dgsyn_dt = _out * g_syn_bar - g_syn/tau_syn
+        dgsyn_dt = -g_syn/tau_syn + _out * g_syn_bar
         g_syn = g_syn + dgsyn_dt * dt ## run Euler step to move conductance
         i_syn =  -g_syn * (v - syn_rest)
         outputs = i_syn #jnp.matmul(inputs, Wdyn * Rscale) + biases