% generate a sparse random matrix of given density
m=10; n=100; density=0.2; A = sprandn(m,n,density);

% choose feasible x, y, s at random, with x and s each about half-full
xfeas = [rand(n/2,1); zeros(n-(n/2),1)];
sfeas = [zeros(n/2,1); rand(n-(n/2),1)];
xfeas = xfeas(randperm(n)); sfeas = sfeas(randperm(n)); 
yfeas = (rand(m,1)-0.5)*4;

% choose b and p to make this (x,y,s) feasible
b = A*xfeas; p=A'*yfeas+sfeas;

% call the solver
[x,y,s,f] = pathfollow(A,b,p);

fprintf(1,' final primal value: %12.6e,  dual value: %12.6e\n', ...
   p'*x, b'*y);
fprintf(1,' primal infeasibility: %12.6e, dual infeasibility: %12.6e\n', ...
  norm(A*x-b), norm(A'*y+s-p));