demo2.m

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Released under: MIT License
% 2019, Sebastiano Vascon
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

close all
clear all

addpath(genpath('../'));

rng('default');                     %For reproducibility

%% Step1: Create random points
cx = [1 1;10 10 ;20 20];            %center of the clouds of points
npts=100; pts= repmat(cx,npts,1) + randn(npts*size(cx,1),2);
A=pdist(pts);                       %pairwise Euclidean distances

%% Step2: Compute their similarity
s=3*var(A);                         %an euristic to find sigma
A=exp(-A./s); A=squareform(A);      %from distance to similarity
A=A.*not(eye(size(A)));             %the graph should not have the self-loop

%% Step3: Decide the DS parameters (dynamics and other...)
dynType=1; %0=Replicator Dynamics, 1=InfectionImmunization 2=Exponential replicator dynamics
precision=1e-6; %the precision required from the dynamical system
maxIters=1000; %number of maximum iteration of the dynamical system
x=ones(size(A,1),1)./size(A,1); %starting point of the dynamical system
theta=1e-5; %threshold used to extract the support from x.

%% Step4: Clusterize the data
[C]=dominantset(A,x,theta,precision,maxIters,dynType);

%% Step5: Show the results
gscatter(pts(:,1),pts(:,2),C);