Conjugate gradient optimizer

Conjugate gradient optimizer for the unconstrained optimization of functions of n variables.

This function returns the vector x=[x1,...xn] that results in the minimum
objective function value for the given function. Inputs are the function
fcn, the initial x0, the gradient approximation method (1= central diff,
2=fwd. diff.), n for the golden section search, the convergence tolerance
epsilon, and the maximum number of iterations, nmax.
Note that the golden section search is very sensitive to n.