The first thing I notice is that \$$\chi(m,n)\$$ can be replaced by a function of a single variable \$$\psi(m-n)\$$. This also means that flipping the arguments to \$$\chi\$$ just flips the sign of the single argument to \$$\psi\$$. Is there a reason you are using the two-argument version here instead?