A new approach to gravitational clustering: a path-integral formalism and large-N expansions
P. Valageas
TL;DR
This work casts gravitational clustering in an expanding universe within a path-integral framework by deriving an action $S[f]$ that encodes Gaussian initial conditions and Vlasov-Poisson dynamics for the phase-space density $f({\bf x},{\bf p},t)$. It then develops three large-$N$ schemes—the direct steepest-descent, the 1PI effective action, and the 2PI effective action—to produce self-consistent nonlinear equations for the mean $\bar f$ and two-point correlators $G$ (and the response $R$). The analysis shows that standard perturbative results on quasi-linear scales are recovered while the large-$N$ techniques enable controlled resummations that remain well-behaved in the nonlinear regime, provided infrared cancellations are handled properly. Infrared issues favor the semiclassical/2PI approaches over the straightforward $N$-field expansion, and the framework naturally connects to BBGKY-like closures and may extend to non-Gaussian initial conditions. The methodology provides a principled, field-theoretic route to modeling structure formation with potential numerical implementations.
Abstract
We show that the formation of large-scale structures through gravitational instability in the expanding universe can be fully described through a path-integral formalism. We derive the action S[f] which gives the statistical weight associated with any phase-space distribution function f(x,p,t). This action S describes both the average over the Gaussian initial conditions and the Vlasov-Poisson dynamics. Next, applying a standard method borrowed from field theory we generalize our problem to an N-field system and we look for an expansion over powers of 1/N. We describe three such methods and we derive the corresponding equations of motion at the lowest non-trivial order for the case of gravitational clustering. This yields a set of non-linear equations for the mean $\fb$ and the two-point correlation G of the phase-space distribution f, as well as for the response function R. These systematic schemes match the usual perturbative expansion on quasi-linear scales but should also be able to handle the non-linear regime. Our approach can also be extended to non-Gaussian initial conditions and may serve as a basis for other tools borrowed from field theory.
