Integral & quadl function gives NaN

Hi everyone,

I am trying to get optimized values by using fmincon function.

In the optimization procedure, a function is integrated with quadl(or integral, alternatively) as following:

quadl(@(e) integProb(e,j,k,Vmat,Vj),a,b,eps1);     

where the 'integProb' function looks like:

function [val]=integProb(e,j,k,Vmat,Vj);

%========|=========|=========|=========|=========|=========|=========|=========|
%        10        20        30        40        50        60        70        80
%========|=========|=========|=========|=========|=========|=========|=========|
%  
% this function returns the value of the integrand at e
% note that e is a row vector 
% matlab function "quad" or "quadl" calls this function for integration
% 
%
%========|=========|=========|=========|=========|=========|=========|=========|
%        10        20        30        40        50        60        70        80
%========|=========|=========|=========|=========|=========|=========|=========|



len = length(e);

x=repmat(e+Vj,k,1)-Vmat(:,1:len);
x2 = normcdf(x); % need the sigma of j (the chosen product)

x2(j,:)=1; % we do not need this. See the integrand in our integration eqn
val = prod(x2).* normpdf(e);

the quadl integration function with integProb returns NaN with warning:'Infinite or Not-a-Number function value encountered.' at some values of a, b, Vmat and Vj. but I don't know exactly what their values make quadl function return NaN because those values are decided in the fmincon optimization process. At least, what I can say is that their values near the ones causing NaN returns some values near zero (eg. 1e-10) from quadl function.

Does anyone have any idea on how to troubleshoot this?

I would appreciate any help,

Thank you.