Well, that'll only be as good as the granularity of the points at which you evaluate the function but
x=[x0:dx:x1];
[xmin,idxmin]=min(func(x));
will do what your problem description is given an arbitrary dx. Or you could write as
x=linspace(x0,x1,Npts);
given an arbitrary number of points between the two ranges, Npts
Alternatively, you might consider
doc fminsearch
that will return the minimum of the function itself independent of some arbitrary evaluation vector.
