RUN_NEURON calculates the firing behavior of a single theta neuron
Description:
Calculates the firing behavior of a single theta neuron with n input
synapses and q input spikes. q=n for SISO in which there is exactly one
spike per synapse. Likewise, there are r output spikes. There will be at
most one output spike after the last input spike, depending on the neuron
phase and the sign of the baseline current, Io.
Syntax:
RUN_NEURON(Neuron, ti, Weights, Delays, SimulationMethod);
RUN_NEURON(..., [TimeStep], [GradFlag], [Verbose]);
[ts, Theta, Y, Z, thm, thp] = RUN_NEURON(...);
Input Parameters:
o Neuron: An object of the neuron class
o ti: A qx2 array that contains the neuron indices (values between 1 and n inclusive)
for each spike time (column 1) and the input spike times (column 2)
o Weights: An nx1 array indicating the weights of the n input synapses
o Delays: An nx1 array indicating the delays of the n input synapses
o SimulationMethod: A string that determines which method will be used
to calculate the output spike firing times. Possible values are
"Numerical" or "Event-Driven".
o TimeStep: Optional positive scalar for use in numerical integration. The default value
is 0.001.
o GradFlag: Optional boolean flag to indicate whether the gradient matrices Y and Z
should be calculated. If the flag is 0, then -1 is returned for Y and Z.
The default value is 1.
o Verbose: Optional flag to indicate if extra information will be
displayed on the screen. A value of 0 displays no additional information (this is
the default value), while a value of 1 displays all information. Values greater
than 1 display partial information. See Verbose for more details.
Output Parameters:
o ts: An rx1 array of the output firing times, returns an empty array if no
output spikes occur
o Theta: An array of size 2+ts(end)/TimeStep. If numerical integration is used, the
complete trajectory of the phase is returned. For event-driven simulation, -1
is returned (since the trajectory is not calculated).
o Y: An nxr matrix of the effect of each synapse n on output spike r, used
in network training
o Z: A qxr matrix of the effect of each input spike q on output spike r,
used in network training
o thm: A qx1 array of phase values directly before each input spike
o thp: A qx1 array of phase values directly after each input spike
Examples:
>> N=theta_neuron;
>> [ts, Theta] = run_neuron(N, [1 3; 2 4; 1 2], 0.01*rand(1,3), zeros(1,3),'Numerical');
>> figure;
>> plot(0.001*(1:length(Theta)),Theta,ts,zeros(1,length(ts)),'o');
>> xlabel('Time (ms)'); ylabel('Phase'); grid on;
>> N=theta_neuron('Io',0.1);
>> [ts, Theta] = run_neuron(N, [1 3; 2 50], 0.01*rand(1,2), zeros(1,2),'Numerical');
>> figure;
>> plot(0.001*(1:length(Theta)),Theta,ts,zeros(1,length(ts)),'o');
>> xlabel('Time (ms)'); ylabel('Phase'); grid on;
See also theta_neuron, verbose0001 function [ts, Theta, Y, Z, thm, thp] = run_neuron(Neuron, ti, Weights, Delays, SimulationMethod, TimeStep, GradFlag, Verbose) 0002 %RUN_NEURON calculates the firing behavior of a single theta neuron 0003 % 0004 %Description: 0005 %Calculates the firing behavior of a single theta neuron with n input 0006 %synapses and q input spikes. q=n for SISO in which there is exactly one 0007 %spike per synapse. Likewise, there are r output spikes. There will be at 0008 %most one output spike after the last input spike, depending on the neuron 0009 %phase and the sign of the baseline current, Io. 0010 % 0011 %Syntax: 0012 %RUN_NEURON(Neuron, ti, Weights, Delays, SimulationMethod); 0013 %RUN_NEURON(..., [TimeStep], [GradFlag], [Verbose]); 0014 %[ts, Theta, Y, Z, thm, thp] = RUN_NEURON(...); 0015 % 0016 %Input Parameters: 0017 %o Neuron: An object of the neuron class 0018 %o ti: A qx2 array that contains the neuron indices (values between 1 and n inclusive) 0019 % for each spike time (column 1) and the input spike times (column 2) 0020 %o Weights: An nx1 array indicating the weights of the n input synapses 0021 %o Delays: An nx1 array indicating the delays of the n input synapses 0022 %o SimulationMethod: A string that determines which method will be used 0023 % to calculate the output spike firing times. Possible values are 0024 % "Numerical" or "Event-Driven". 0025 %o TimeStep: Optional positive scalar for use in numerical integration. The default value 0026 % is 0.001. 0027 %o GradFlag: Optional boolean flag to indicate whether the gradient matrices Y and Z 0028 % should be calculated. If the flag is 0, then -1 is returned for Y and Z. 0029 % The default value is 1. 0030 %o Verbose: Optional flag to indicate if extra information will be 0031 % displayed on the screen. A value of 0 displays no additional information (this is 0032 % the default value), while a value of 1 displays all information. Values greater 0033 % than 1 display partial information. See Verbose for more details. 0034 % 0035 %Output Parameters: 0036 %o ts: An rx1 array of the output firing times, returns an empty array if no 0037 % output spikes occur 0038 %o Theta: An array of size 2+ts(end)/TimeStep. If numerical integration is used, the 0039 % complete trajectory of the phase is returned. For event-driven simulation, -1 0040 % is returned (since the trajectory is not calculated). 0041 %o Y: An nxr matrix of the effect of each synapse n on output spike r, used 0042 % in network training 0043 %o Z: A qxr matrix of the effect of each input spike q on output spike r, 0044 % used in network training 0045 %o thm: A qx1 array of phase values directly before each input spike 0046 %o thp: A qx1 array of phase values directly after each input spike 0047 % 0048 %Examples: 0049 %>> N=theta_neuron; 0050 %>> [ts, Theta] = run_neuron(N, [1 3; 2 4; 1 2], 0.01*rand(1,3), zeros(1,3),'Numerical'); 0051 %>> figure; 0052 %>> plot(0.001*(1:length(Theta)),Theta,ts,zeros(1,length(ts)),'o'); 0053 %>> xlabel('Time (ms)'); ylabel('Phase'); grid on; 0054 % 0055 %>> N=theta_neuron('Io',0.1); 0056 %>> [ts, Theta] = run_neuron(N, [1 3; 2 50], 0.01*rand(1,2), zeros(1,2),'Numerical'); 0057 %>> figure; 0058 %>> plot(0.001*(1:length(Theta)),Theta,ts,zeros(1,length(ts)),'o'); 0059 %>> xlabel('Time (ms)'); ylabel('Phase'); grid on; 0060 % 0061 %See also theta_neuron, verbose 0062 0063 %Copyright (C) 2008 Sam McKennoch <Samuel.McKennoch@loria.fr> 0064 0065 0066 %Error handling for input arguments 0067 if nargin<5 0068 disp('Error in run_neuron!!! Not enough input arguements'); 0069 disp(['Needed at least 5 inputs but only got ' num2str(nargin)]); 0070 return; 0071 end 0072 if (nargin < 6) 0073 TimeStep=0.001; 0074 end 0075 if (nargin < 7) 0076 GradFlag=1; 0077 end 0078 if (nargin < 8) 0079 Verbose=0; 0080 end 0081 0082 %Initialize return parameters to error return values 0083 Theta=-1; ts=[]; Y=-1; Z=-1; thm=-1; thp=-1; 0084 0085 %Create the Network 0086 NumInputs=max(ti(:,1)); 0087 NeuronIndex=NumInputs+1; 0088 ThNN=theta_neuron_network('Io',Neuron.Io,'Alpha',Neuron.Alpha,'TimeStep',TimeStep,'SimulationMethod',SimulationMethod,... 0089 'StructureFormat',{'LayerArray',[NumInputs 1]}); 0090 ThNN.Weights(1:NumInputs,NeuronIndex)=Weights(1:NumInputs); %Ignores extra weights 0091 ThNN.Delays(1:NumInputs,NeuronIndex)=Delays(1:NumInputs); %Ignores extra delays 0092 0093 %Run the Network 0094 [Theta, ts, Y, Z, thm, thp] = run_networked_neuron(ThNN, ti, NeuronIndex, GradFlag, Verbose); 0095 0096 return;