Predictions in eternal inflation
Sergei Winitzki
TL;DR
This work analyzes how to make predictions in eternally inflating spacetimes by formulating a stochastic inflation framework based on Langevin dynamics and Fokker-Planck equations for $P(\phi,t)$ and $P_V(\phi,t)$. It highlights gauge dependence issues, the distinction between diffusion- and tunneling-driven self-reproduction, and the role of regularization in deriving observable distributions; it also contrasts volume-based and worldline-based measures, including their behavior in the string theory landscape. The paper provides both general analytical structures (largest eigenvalues $\gamma,\tilde{\gamma}$, stationary distributions) and specific prescriptions (spherical vs equal-time cutoffs; comoving vs worldline weighting) for extracting predictions like the distribution of fields $\chi_a$ at reheating. Overall, it clarifies how eternal inflation can generate statistically diverse observers and outlines concrete methodologies to translate this diversity into testable predictions, even in highly inhomogeneous, multi-vacua settings.
Abstract
In generic models of cosmological inflation, quantum fluctuations strongly influence the spacetime metric and produce infinitely many regions where the end of inflation (reheating) is delayed until arbitrarily late times. The geometry of the resulting spacetime is highly inhomogeneous on scales of many Hubble sizes. The recently developed string-theoretic picture of the "landscape" presents a similar structure, where an infinite number of de Sitter, flat, and anti-de Sitter universes are nucleated via quantum tunneling. Since observers on the Earth have no information about their location within the eternally inflating universe, the main question in this context is to obtain statistical predictions for quantities observed at a random location. I describe the problems arising within this statistical framework, such as the need for a volume cutoff and the dependence of cutoff schemes on time slicing and on the initial conditions. After reviewing different approaches and mathematical techniques developed in the past two decades for studying these issues, I discuss the existing proposals for extracting predictions and give examples of their applications.
