Correlation Classes on the Landscape: To What Extent is String Theory Predictive?

TL;DR

This paper investigates the predictivity of string theory within the landscape, proposing a statistical method to quantify correlations among low-energy observables across different regions.

Contribution

It introduces a novel, robust statistical approach to assess and quantify the predictivity and correlation structures in the string landscape without prior assumptions.

Findings

Different landscape regions exhibit distinct correlation sets.

The method effectively quantifies overlaps and sizes of correlation classes.

It applies broadly to the entire landscape or specific subsets.

Abstract

In light of recent discussions of the string landscape, it is essential to understand the degree to which string theory is predictive. We argue that it is unlikely that the landscape as a whole will exhibit unique correlations amongst low-energy observables, but rather that different regions of the landscape will exhibit different overlapping sets of correlations. We then provide a statistical method for quantifying this degree of predictivity, and for extracting statistical information concerning the relative sizes and overlaps of the regions corresponding to these different correlation classes. Our method is robust and requires no prior knowledge of landscape properties, and can be applied to the landscape as a whole as well as to any relevant subset.