This paper completes the argument that lattice simulations of partially quenched QCD can provide quantitative information about QCD itself, with the aid of partially quenched chiral perturbation theory. A barrier to doing this has been the inclusion of ${\Phi }_0,$ the partially quenched generalization of the ${\eta }^{\prime },$ in previous calculations in the partially quenched effective theory. This invalidates the low energy perturbative expansion, gives rise to many new unknown parameters, and makes it impossible to reliably calculate the relation between the partially quenched theory and low energy QCD. We show that it is straightforward and natural to formulate partially quenched chiral perturbation theory without ${\Phi }_0,$ and that the resulting theory contains the effective theory for QCD without the ${\eta }^{\prime }.$ We also show that previous results, obtained including ${\Phi }_0,$ can be reinterpreted as applying to the theory without ${\Phi }_0.$ We contrast the situation with that in the quenched effective theory, where we explain why it is necessary to include ${\Phi }_0.$ We also compare the derivation of chiral perturbation theory in partially quenched QCD with the standard derivation in unquenched QCD. We find that the former cannot be justified as rigorously as the latter, because of the absence of a physical Hilbert space. Finally, we present an encouraging result: unphysical double poles in certain correlation functions in partially quenched chiral perturbation theory can be shown to be a property of the underlying theory, given only the symmetries and some plausible assumptions.

• Received 8 August 2001

DOI:https://doi.org/10.1103/PhysRevD.64.114510