Probabilities in the landscape
Alexander Vilenkin
TL;DR
In an eternally inflating multiverse with a vast string landscape of vacua, predicting observed constants requires a gauge-invariant measure. The paper advocates a pocket-based approach that computes a local volume distribution $P^{(V)}(X)$ inside a representative pocket and couples it to bubble counting via a bubble abundance $p_j$ and a reference length $R_j$, yielding $P_j^{(V)} \propto p_j R_j^3$. It provides concrete methods to define $p_j$ (via a comoving-size cutoff or worldlines) and $R_j$ (via early-time universes or curvature scales), and generalizes to continuous variables $X$ with $P_j^{(V)}(X) \propto p_j \hat{P}_j(X) R_j^3(X)$. The approach aims to deliver a practical, largely gauge-invariant framework for predicting observer distributions across vacua, while acknowledging non-uniqueness and inviting empirical tests and comparisons with alternative proposals.
Abstract
I review recent progress in defining probability distributions in the inflationary multiverse.
