Partition Function for (2+1)-Dimensional Einstein Gravity

TL;DR

This paper explores the relationship between different formulations of the partition function in (2+1)-dimensional Einstein gravity, analyzing gauge-fixing effects, measure factors, and implications for quantum gravity.

Contribution

It clarifies how various reduced phase space formulations of the partition function relate, including measure factors and zero-mode contributions, in the context of (2+1)-dimensional quantum gravity.

Findings

Partition function reduces to a form involving Teichmüller parameters and conjugate momenta.

A measure factor related to Faddeev-Popov determinant appears in the reduced form.

Zero-mode contributions can alter the path-integral measure depending on the domain choice.

Abstract

Taking (2+1)-dimensional pure Einstein gravity for arbitrary genus $g$ as a model, we investigate the relation between the partition function formally defined on the entire phase space and the one written in terms of the reduced phase space. In particular the case of $g=1$ is analyzed in detail. By a suitable gauge-fixing, the partition function $Z$ basically reduces to the partition function defined for the reduced system, whose dynamical variables are $(\tau^A, p_A)$. [The $\tau^A$'s are the Teichm\"uller parameters, and the $p_A$'s are their conjugate momenta.] As for the case of $g=1$, we find out that $Z$ is also related with another reduced form, whose dynamical variables are $(\tau^A, p_A)$ and $(V, \sigma)$. [Here $\sigma$ is a conjugate momentum to 2-volume $V$.] A nontrivial factor appears in the measure in terms of this type of reduced form. The factor turns out to be a Faddeev-Popov determinant coming from the time-reparameterization invariance inherent in this type of formulation. Thus the relation between two reduced forms becomes transparent even in the context of quantum theory. Furthermore for $g=1$, a factor coming from the zero-modes of a differential operator $P_1$ can appear in the path-integral measure in the reduced representation of $Z$. It depends on the path-integral domain for the shift vector in $Z$: If it is defined to include $\ker P_1$, the nontrivial factor does not appear. On the other hand, if the integral domain is defined to exclude $\ker P_1$, the factor appears in the measure. This factor can depend on the dynamical variables, typically as a function of $V$, and can influence the semiclassical dynamics of the (2+1)-dimensional spacetime. These results shall be significant from the viewpoint of quantum gravity.