Add show functionality for synapses

[alec-hoyland]

New feature proposal

Add functionality to show so that it works for synapses as well as conductances.

Implementation

Most synapse models have a s_inf function, which should be plotted the same way that m_inf is plotted. Time constants can vary -- some models don't have tau_s functions, but use a sum of exponentials with constant decay rates. Usually, these are called tau_r and tau_d.

At the very least, plotting the s_inf function over where the m_inf function should be doable, and then plotting the tau_s function (if any) over the tau_m plots.

Alternatively, there could be (yet another) method of xolotl that explicitly handles the synapse case.

[sg-s]

• Does every synapse have a s_inf function?
• What happens when a synapse doesn't implement a s_inf function?
• What happens when a synapse implements something that basically is a s_inf function, but it is called something else?

[alec-hoyland]

I think this is doomed because I can't get the function handles when I construct a synapse as a standalone cpplab object.

[alec-hoyland]

>> C = cpplab('prinz/Cholinergic')

C = 

 Cholinergic object (dcc022a) with:

          gmax : 0
             s : 0

>> {C.cpp_child_functions.fun_name}

ans =

  1×10 cell array

    {0×0 double}    {'getFullStateSize'}    {'s_inf'}    {'tau_s'}    {'sdot'}    {'integrate'}    {'integrateMS'}    {'checkSolvers'}    {'getFullState'}    {'connect'}

>> C.cpp_child_functions(3)

ans = 

  struct with fields:

           fun_name: 's_inf'
         fun_handle: []
    fun_return_type: 'double'
     fun_input_type: {'double'}
    fun_input_names: {'V_pre'}

>> C = cpplab('liu/NaV')

C = 

 NaV object (54891d4) with:

             m : 0
             h : 1
          gbar : NaN
             E : 30
>> C.cpp_child_functions(3)

ans = 

  struct with fields:

           fun_name: 'm_inf'
         fun_handle: @(V,Ca)1.0/(1.0+exp((V+25.5)/-5.29))
    fun_return_type: 'double'
     fun_input_type: {'double'  'double'}
    fun_input_names: {'V'  'Ca'}

>> 

[alec-hoyland]

* Does every synapse have a `s_inf` function?

Currently, yes. That's not something that has to happen though.

* What happens when a synapse doesn't implement a `s_inf` function?

Then it doesn't get plotted. Currently show reads from getGatingFunctions, which returns a function mapping to NaN if it can't find a matching function.

* What happens when a synapse implements something that basically is a `s_inf` function, but it is called something else?

Then it doesn't find it, and nothing is plotted. If I named my function s_variable_steady_state then the strcmp() function in getGatingFunctions would return false, and the function s_inf = @(V) NaN is returned instead.

[sg-s]

fair enough. not getting the function handles from synapses is a bug in cpplab. See also: this bug in cpplab

[sg-s]

OK, the function handles are now populated. But it's pretty useless because the parameters are not set. (e.g. Vth)

[sg-s]

this is very complicated to implement.

[github-actions]

Stale issue message

[sg-s]

wontfix