RELATE_Q_TO_K relates spiking and synaptic indices
Description:
If Neuron1 and Neuron2 are provided, returns an array that contains q
values for Neuron 2 on the synapse between Neuron 1 and Neuron 2. The
indices of the array are the k values. This array can then be used to
relate the synaptic indices (k) with the spiking indices (q).
For SISO networks, k=q. If Neuron1 and Neuron2 are not provided, a cell
array is returned with K2Q arrays for all connected neuron pairs. This
second method of calling is more efficient than repeatedly calling the
first method since it reduces redundant function calling.
Syntax:
K2Q=RELATE_Q_TO_K(ThNN,ts,[Verbose]);
K2Q=RELATE_Q_TO_K(ThNN,ts,Neuron1,Neuron2,[Verbose]);
K2Q=RELATE_Q_TO_K(ThNN,ts,Neuron1,Neuron2,InputNeurons,Verbose);
Input Parameters:
o ThNN: An object of the theta neuron network class
o ts: A cell array of length NumNeurons containing in each cell an array
of length r_j of neuron j's output spike times.
o Neuron1: Neuron index for the input neuron
o Neuron2: Neuron index for the output neuron
o InputNeurons: Optional input array containing the input neuron indices
for Neuron1, used to reduce redundant computation. Verbose must be
specified if InputNeurons is specified.
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 K2Q: Either an array that contains q values for Neuron 2 on the synapse
between Neuron 1 and Neuron 2 (the indices of the array are the k
values) or a cell array with K2Q arrays for all connected neuron
pairs.
Examples:
>> %SISO Example
>> ThNN=theta_neuron_network('ReferenceTime',-1,'StructureFormat',...
{'LayerArray', [2 1]});
>> ts=run_network(ThNN, [1 5; 2 7]);
>> K2Q=relate_q_to_k(ThNN,ts,2,3,1);
>> K2Q=relate_q_to_k(ThNN,ts,2,3,[1 2],0);
>> K2QAll=relate_q_to_k(ThNN,ts)
>> %MIMO Example
>> ThNN=theta_neuron_network('ReferenceTime',-1,'StructureFormat',...
{'LayerArray', [2 1]});
>> ts=run_network(ThNN, [2 3; 1 5; 2 7]);
>> K2Q=relate_q_to_k(ThNN,ts,2,3,1);
>> K2QAll=relate_q_to_k(ThNN,ts)
See also theta_neuron_network, verbose0001 function K2Q=relate_q_to_k(ThNN,ts,varargin) 0002 %RELATE_Q_TO_K relates spiking and synaptic indices 0003 % 0004 %Description: 0005 %If Neuron1 and Neuron2 are provided, returns an array that contains q 0006 %values for Neuron 2 on the synapse between Neuron 1 and Neuron 2. The 0007 %indices of the array are the k values. This array can then be used to 0008 %relate the synaptic indices (k) with the spiking indices (q). 0009 %For SISO networks, k=q. If Neuron1 and Neuron2 are not provided, a cell 0010 %array is returned with K2Q arrays for all connected neuron pairs. This 0011 %second method of calling is more efficient than repeatedly calling the 0012 %first method since it reduces redundant function calling. 0013 % 0014 %Syntax: 0015 %K2Q=RELATE_Q_TO_K(ThNN,ts,[Verbose]); 0016 %K2Q=RELATE_Q_TO_K(ThNN,ts,Neuron1,Neuron2,[Verbose]); 0017 %K2Q=RELATE_Q_TO_K(ThNN,ts,Neuron1,Neuron2,InputNeurons,Verbose); 0018 % 0019 %Input Parameters: 0020 %o ThNN: An object of the theta neuron network class 0021 %o ts: A cell array of length NumNeurons containing in each cell an array 0022 % of length r_j of neuron j's output spike times. 0023 %o Neuron1: Neuron index for the input neuron 0024 %o Neuron2: Neuron index for the output neuron 0025 %o InputNeurons: Optional input array containing the input neuron indices 0026 % for Neuron1, used to reduce redundant computation. Verbose must be 0027 % specified if InputNeurons is specified. 0028 %o Verbose: Optional flag to indicate if extra information will be 0029 % displayed on the screen. A value of 0 displays no additional 0030 % information (this is the default value), while a value of 1 displays 0031 % all information. Values greater than 1 display partial information. 0032 % See Verbose for more details. 0033 % 0034 %Output Parameters: 0035 %o K2Q: Either an array that contains q values for Neuron 2 on the synapse 0036 % between Neuron 1 and Neuron 2 (the indices of the array are the k 0037 % values) or a cell array with K2Q arrays for all connected neuron 0038 % pairs. 0039 % 0040 %Examples: 0041 %>> %SISO Example 0042 %>> ThNN=theta_neuron_network('ReferenceTime',-1,'StructureFormat',... 0043 % {'LayerArray', [2 1]}); 0044 %>> ts=run_network(ThNN, [1 5; 2 7]); 0045 %>> K2Q=relate_q_to_k(ThNN,ts,2,3,1); 0046 %>> K2Q=relate_q_to_k(ThNN,ts,2,3,[1 2],0); 0047 %>> K2QAll=relate_q_to_k(ThNN,ts) 0048 % 0049 %>> %MIMO Example 0050 %>> ThNN=theta_neuron_network('ReferenceTime',-1,'StructureFormat',... 0051 % {'LayerArray', [2 1]}); 0052 %>> ts=run_network(ThNN, [2 3; 1 5; 2 7]); 0053 %>> K2Q=relate_q_to_k(ThNN,ts,2,3,1); 0054 %>> K2QAll=relate_q_to_k(ThNN,ts) 0055 % 0056 %See also theta_neuron_network, verbose 0057 0058 %Copyright (C) 2008 Sam McKennoch <Samuel.McKennoch@loria.fr> 0059 0060 0061 if nargin<2 0062 disp('Error in relate_q_to_k: Not enough input arguments'); 0063 K2Q=-1; 0064 return; 0065 elseif nargin==2 0066 Verbose=0; 0067 elseif nargin==3 0068 Verbose=varargin{1}; 0069 elseif nargin==4 0070 Neuron1=varargin{1}; 0071 Neuron2=varargin{2}; 0072 Verbose=0; 0073 InputNeurons=ThNN.RelativeInputNeurons{Neuron2};%get_input_neurons(ThNN,Neuron2); 0074 elseif nargin==5 0075 Neuron1=varargin{1}; 0076 Neuron2=varargin{2}; 0077 Verbose=varargin{3}; 0078 InputNeurons=ThNN.RelativeInputNeurons{Neuron2};%get_input_neurons(ThNN,Neuron2); 0079 elseif nargin==6 0080 Neuron1=varargin{1}; 0081 Neuron2=varargin{2}; 0082 InputNeurons=varargin{3}; 0083 Verbose=varargin{4}; 0084 end 0085 0086 %Recursively call this function in single neuron mode 0087 if nargin==2 || nargin==3 0088 Weights=ThNN.Weights; 0089 Conn=(Weights~=0); 0090 K2Q=cell(size(Conn)); 0091 0092 if ThNN.RecursionFlag 0093 %Inline setdiff 0094 %NeuronQueue=setdiff(1:size(Conn,1),ThNN.InputNeurons');%CompletedNeurons); 0095 NeuronQueue=1:size(Conn,1); 0096 for n=1:length(ThNN.InputNeurons) 0097 NeuronQueue=NeuronQueue(~(NeuronQueue==ThNN.InputNeurons(n))); 0098 end 0099 else 0100 NeuronQueue=ThNN.NeuronQueue; 0101 end 0102 0103 Pos=1; 0104 while Pos<=length(NeuronQueue) && ~isempty(NeuronQueue) 0105 %[NeuronQueue,CurrentNeuron]=pop(NeuronQueue); 0106 CurrentNeuron=NeuronQueue(Pos); 0107 if ThNN.RecursionFlag 0108 NeuronQueue=NeuronQueue(2:end); 0109 else 0110 Pos=Pos+1; 0111 end 0112 CurrentInputNeurons=ThNN.RelativeInputNeurons{CurrentNeuron}';%get_input_neurons(ThNN,CurrentNeuron)'; 0113 for j=CurrentInputNeurons 0114 K2Q{j,CurrentNeuron}=relate_q_to_k(ThNN,ts,j,CurrentNeuron,CurrentInputNeurons,Verbose); 0115 end 0116 %Find Similarly connected Neurons 0117 for m=NeuronQueue(Pos:end)' 0118 if min(Conn(m,:)==Conn(CurrentNeuron,:))==1 0119 K2Q(:,m)=K2Q(:,CurrentNeuron); 0120 0121 %Inline setdiff 0122 %NeuronQueue=setdiff(NeuronQueue,m) 0123 NeuronQueue=NeuronQueue(~(NeuronQueue==m)); %Short because m is a scalar 0124 0125 end 0126 end 0127 end 0128 0129 %Single neuron mode 0130 else 0131 0132 Indices=[]; K2Q=[]; 0133 0134 try 0135 for j=1:length(InputNeurons) 0136 Indices=[Indices; [InputNeurons(j)*ones(1,length(ts{InputNeurons(j)})); ts{InputNeurons(j)}]']; 0137 end 0138 0139 %Inline sortrows 0140 %Indices=sortrows(Indices,2); 0141 [Temp1,Temp2]=sort(Indices(:,2)); 0142 Indices=[Indices(Temp2,1) Temp1]; 0143 0144 K2Q=[K2Q; find(Indices(:,1)==Neuron1)]; 0145 0146 if (Verbose==1) 0147 disp(['Calculating the index relationship between neurons ', num2str(Neuron1), ' and ' num2str(Neuron2)]); 0148 disp('Indices contains the input neuron index in row 1 and the input spike time in row 2'); 0149 Indices 0150 disp('K2Q is an array that contains the q values for the second neuron on the synapse between the two neurons'); 0151 K2Q 0152 end 0153 0154 return; 0155 catch 0156 if ~ismember(Neuron1,InputNeurons) 0157 disp('Error in K2Q: Neuron1 is not an input neuron to Neuron2'); 0158 K2Q=-1; 0159 return; 0160 end 0161 end 0162 end