Programming expectation algorithm for negative binomial distribution

For the negative binomial distribution, there is a dispersion parameter to be estimated along with beta parameters. There is an error in defining the dispersion parameter.

How to define the dispersion parameter correctly.

Following is the code:

function logL = em_nbl(theta,ex,j) 
global y x k n ;
beta= theta(k);
al=theta(k+1);

for i= 1:n,
L(i) = ex(i,j)*(log(gamma(y(i)+1/(al)))+y(i)*log(al)+(y(i)-1)*(x(i,:)*beta)-(y(i)+1/al)*log(1+al*(x(i,:)*beta))-log(gamma(1/a1)));
end;

      logL = -sum(L);
  return;

Following is the error:

Error using em_nbl (line 6)
Not enough input arguments.

[Walter Roberson]

How are you invoking the code?

[Walter Roberson]

You should avoid using global.

http://www.mathworks.com/help/matlab/math/parameterizing-functions.html