Strange behavior in bintprog

I'm running into a strange issue for bintprog for the following example.

A = [1000 0; 0 1000]
[xx] = bintprog(zeros(2,1),-1*A,-100*ones(2,1))
Optimization terminated.

xx =

       1
       1

^ this works but if either entry becomes large, the optimization is infeasible, for example:

A = [1000 0; 0 4000000]
[xx] = bintprog(zeros(2,1),-1*A,-100*ones(2,1))
The problem is infeasible.

xx =

       0
       0

Does anyone know what I'm doing wrong?

Thanks!

[Matt J]

Does it still happen if you do

[xx] = bintprog(zeros(2,1),-1*A,-100*ones(2,1),[],[],[1;1])

[Zhe]

It works when I do that.

This error looks similar to this question: http://www.mathworks.com/matlabcentral/answers/38152-bintprog-returns-incorrect-infeasible-result

I submitted a bug report for the time being.

[Zhe]

After contacting technical support, the problem was that the default tolerance value was not high enough, by setting a smaller tolerance, the optimization will work:

option = optimset('TolRLPFun',1e-9);
[xx] = bintprog(zeros(2,1),-1*A,-100*ones(2,1),[],[],[],option)