function [features1 features2] = sim1overN(dim, s)
% Generate feature matrix based on zipf distribution
% Inputs:
% 1. dim- number of features
% 2. s- constant 

% P(ith feature = 1) = 1/i
const = sum(1./(1:dim));
% Output:
% 1. features1 - feature matrix using class1
% 2. features2 - feature matrix using class2

% case 1
%classOne = (1./(1:dim))/const;
%classTwo = (1 - 1./((1:dim).^2))/const;

% case 2
 classOne = (1./(1:dim))/const;
 classTwo = (1 - 1./(1:dim))/const;

% case 3
%classOne = (1./(1:dim));
%classTwo = 1 - 1./(1:dim);

% case 4
%classOne = 1./(1:dim);
%classTwo = (1 - 1./((1:dim).^2));

% classOne = (1./(1:dim).^2)/const;
%classTwo = (1 - 1./((1:dim).^2))/const;

% populating features1 and features2
features1 = single((rand(1,dim)<classOne));
features2 = single((rand(1,dim)<classTwo));

% Sanity check
% features1 = ones(1,dim);
% features2 = zeros(1,dim);

% both features from same class - stupid
% features1 = single((rand(1,dim)<classOne));
% features2 = single((rand(1,dim)<classOne));