gnatfinder generates the second-order causal activity graph from a list of spike times and network connectivity.
Weakly connected components of the second-order activity graph correspond to recurrances of causal patterns in the spiking activity.
gnatfinder <N cells> <activity file> <network file>
The activity file is a text file containing spikes sorted in time.
Each line is a spike and has the format:
<type> <timestamp> <neuron id>
type is currently ignored, but could be used to tag different classes of events in the future.
timestamp is the time of the spike in hexadecimal. Usually expressed in units of milliseconds. Integers only!
neuron id is the integer identifying the neuron producing the spike.
The network file is a text file listing connectivity.
Each line is a synapse. The line format is
<src_id> <tgt_id> <rel_w> <delay>
src_id is the id of the presynaptic neuron
tgt_id is the id of the postsynaptic neuron
rel_w is the (float) relative weight of the synaptic connection
delay is the conduction delay along this connection, in the same units as timestamps in the activity file.
The output file is by default "./gnat_output.txt". It is a text file. Each line is an edge int he second order graph.
The line format is:
<presyn_id> <time_1> <time_2> <postsyn_id> <time_1> <time_2>
presyn_id is the id of the presynaptic neuron for this spike pair.
postsyn_id is the id of the postsynaptic neuron for this spike pair.
time_1 is the timestamp of the earlier presynaptic spike of the repeated interaction
time_2 is the timestamp of the later pre/post synaptic spike of the recurring interaction
To compile gnatfinder, use the command:
gcc -o gnatfinder gnatfinder.c quadtree.c raster.c network.c gnats.c -Wall -Wextra -g -lm
To compile the first order gnatfinder:
g++ -std=c++11 -o gnat1 compute_activity_threads.cpp
No other libraries besides the math library are needed for now.
First order GNAT invocation:
<progname> <n_neurons> <connection file> <activity file> <function> <output file>
function = 1 to compute GNATs
function = 2 to compute causal distances (for histogram)