All of these residential models another feature of complexity which we have as yet neglected throughout: we have so far assumed in both retail and residential models that 'cost' terms like k, q or r are constant. Much more theory can be added to this from urban economics to build in any known spatial variation. This does not change the argument in principle. A term like q above becomes q. and may in part be
1 explicitly a function of terms involving model variables like density (which would include Hi). If this was the case, then new non-linearities would be introduced and hence new bifurcation behaviour.