A non-Gaussian landscape
Sami Nurmi (Helsinki U. & Helsinki Inst. of Phys.), Christian T., Byrnes (Sussex U., Astron. Ctr.), Gianmassimo Tasinato (Portsmouth U., ICG)

TL;DR
This paper explores how large-scale primordial perturbations influence local observable properties in inflationary models, especially affecting non-Gaussian statistics and their detectability, with implications for interpreting cosmological data.
Contribution
It establishes systematic links between observable primordial perturbations and the global inflating space, highlighting effects of multiple fields and non-Gaussianity on observational signatures.
Findings
Non-Gaussianity predictions can vary significantly across different patches.
Detection or non-detection of non-Gaussianity constrains inflationary models.
Novel methods to assess the naturalness of observational configurations.
Abstract
Primordial perturbations with wavelengths greater than the observable universe shift the effective background fields in our observable patch from their global averages over the inflating space. This leads to a landscape picture where the properties of our observable patch depend on its location and may significantly differ from the expectation values predicted by the underlying fundamental inflationary model. We show that if multiple fields are present during inflation, this may happen even if our horizon exit would be preceded by only a few e-foldings of inflation. Non-Gaussian statistics are especially affected: for example models of local non-Gaussianity predicting |f_NL|>> 10 over the entire inflating volume can have a probability up to a few tens of percent to generate a non-detectable bispectrum in our observable patch |fNL^{obs.}|<10. In this work we establish systematic…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
