function n = logerr(r,k,n0,noise,ifrel,nostep)
%simulates logistic growth with stochastic component

%r - intrinsic growth rate
%k - carrying capacity
%n0 - initial population size
%ifrel: 0 - absolute noise (no of ind)
%       1 - relative noise (proportion of number of individuals)
%       2 - demographic stochasticity
%nostep - number of steps for simulation

  n = [n0];
  for t=1:nostep
     if ifrel == 1
        nprime = n(t) + r*n(t)*(k-n(t))/k + randn*noise*n(t);
     elseif ifrel == 2
        if n(t) == 0
           nprime = n(t);
        else
           nprime = n(t) + poisr(r*n(t)*(k-n(t))/k,1);
           %nprime = poisr(n(t) + r*n(t)*(k-n(t))/k,1);
        end
     else
        nprime = n(t) + r*n(t)*(k-n(t))/k + randn*noise;
     end
     if nprime < 0
        nprime = 0;
     end
     n =[n nprime];
  end
  
  
function y = poisr(lambda,n)
%POISR Poisson random number generator.
%	POISR(LAMBDA,N) generates N random deviates from the
%	Poisson distribution with mean LAMBDA.
%
%	POISR(LAMBDA) generates a single random deviate.
%
%	See POISP, RAND.

% GKS 31 July 1993, 28 July 1999
% Uses naive inversion method.
if lambda == 0
   for i = 1:n,
      y(i) = 0;
   end;
   return;
end

if nargin==1, n=1; end;
low = max( 0 , floor( lambda - 8*sqrt(lambda) ));
hi = ceil( lambda + 8*sqrt(lambda) + 4/lambda );
x = (low:hi)';
p = cumsum( exp( -lambda + x.*log(lambda) - gammaln(x+1) ));
u = rand(n,1);
y = zeros(n,1);
for i = 1:n,
   k = find( u(i) < p );
   y(i) = x( k(1) );
end;