Fisher's renormalization relations (1968) between the 'ideal' critical indices and the 'true' ones are proved to be symmetric. This property is shown to be independent of the sign of the ideal specific heat index.
The critical properties of systems under constraint differ from their ideal counterparts through Fisher renormalization. The mathematical properties of Fisher renormalization applied to critical exponents are well known: the renormalized indices obey the same scaling relations as the ideal ones and the transformations are involutions in the sense that re-renormalizing the critical exponents of the constrained system delivers their original, ideal counterparts. Here we examine Fisher renormalization of critical amplitudes and show that, unlike for critical exponents, the associated transformations are not involutions. However, for ratios and combinations of amplitudes which are universal, Fisher renormalization is involutory.
The extent to which the renormalization of critical-point behavior should be visible experimentally is investigated on the basis of detailed numerical calculations for a three-dimensional soluble model (a mobile-electron Ising ferromagnet). If a dilution parameter x is defined such that the change in critical temperature from the "pure" or unrenormalized system is $|{T}_{c}(x)\ensuremath{-}{T}_{c}^{0}|=\frac{x{T}_{c}}{f}$, where $f=0.6\ensuremath{-}0.9$, then we conclude that the effective exponents ${\ensuremath{\beta}}_{\mathrm{fit}}(x)$ and ${\ensuremath{\gamma}}_{\mathrm{fit}}(x)$ which will be observed experimentally, vary roughly as ${\ensuremath{\beta}}_{\mathrm{fit}}\ensuremath{\approx}\ensuremath{\beta}+x\ensuremath{\Delta}{\ensuremath{\beta}}_{X}$ and ${\ensuremath{\gamma}}_{\mathrm{fit}}\ensuremath{\simeq}\ensuremath{\gamma}+\frac{1}{5}x\ifmmode\times\else\texttimes\fi{}(1+2{x}^{2})\ensuremath{\Delta}{\ensuremath{\gamma}}_{X}$. Here $\ensuremath{\beta}$ and $\ensuremath{\gamma}$ are the ideal exponents for the order parameter and total fluctuation or susceptibility of the pure system, while $\ensuremath{\Delta}{\ensuremath{\beta}}_{X}={\ensuremath{\beta}}_{X}\ensuremath{-}\ensuremath{\beta}$ and $\ensuremath{\Delta}{\ensuremath{\gamma}}_{X}={\ensuremath{\gamma}}_{X}\ensuremath{-}\ensuremath{\gamma}$, in which ${\ensuremath{\beta}}_{X}=\frac{\ensuremath{\beta}}{(1\ensuremath{-}{\ensuremath{\alpha}}^{\ensuremath{'}})}$ and ${\ensuremath{\gamma}}_{X}=\frac{\ensuremath{\gamma}}{(1\ensuremath{-}\ensuremath{\alpha})}$ are the fully renormalized exponents, while $\ensuremath{\alpha}$ and ${\ensuremath{\alpha}}^{\ensuremath{'}}$ (assumed positive) describe the divergence of the specific heats of the pure system. [Theoretically the true limiting asymptotic behavior at the transition is described by $\ensuremath{\beta}(x)={\ensuremath{\beta}}_{X}$ and $\ensuremath{\gamma}(x)={\ensuremath{\gamma}}_{X}$ for all x>0.] The renormalized specific heats are found to be sensitive to x but their true renormalized behavior is not evident until $x\ensuremath{\gtrsim}0.3$. Various techniques of data analysis such as logarithmic, semilogarithmic, Heller-Benedek, and Kouvel-Fisher plots have been tested.