Can MAD of TOMLAB work alone to provide automatic differentiation to Fmincon of MATLAB?

I am using fmincon of MATLAB with sqp algorithm to deal with a constraint nonlinear optimization problem. Since in my constraint function, there is a 2-D lookup table which I used for searching an intermediate parameter based on current variables which I'm going to optimize (and it's the only possible way for this parameter), thus providing analytic gradient and/or Jaccobian is not possible.

MAD (Matlab automatic differentiation) is a package of TOMLAB optimization toolbox which can generate automatic differentiation for Matlab code, and it is said to be very good. But I have no idea how to use it as an additional source to provide my fmincon with automatic differentiation in Matlab. I will be very grateful if anyone knows how to implement it. Thanks.