There has been something that has bothered me about the temperature function in the partition function approach last used here [1]: $f(\ell) = \log \ell < 0$ for $\ell < 1$ (or in the original application [2] in terms of money supply $m$).