Flat tree-level inflationary potentials in the light of cosmic microwave background and large scale structure data

Guillermo Ballesteros, J.a. Casas, J. R. Espinosa, Roberto Ruiz de Austri, Roberto TrottaJournal of Cosmology and Astroparticle Physics 2008 · DOI

TL;DR

This paper explores a class of "flat tree-level" inflationary models, motivated by supersymmetry, that predict a specific logarithmic dependence for the inflaton potential. By analyzing cosmic microwave background (CMB) and large-scale structure (LSS) data, the authors find these models fit remarkably well with very few parameters, predicting small tensor perturbations and a reasonable number of e-folds, making them strong contenders for a standard class of inflationary models.

What Problem Does This Paper Address?

The early universe's inflationary epoch is a crucial theoretical framework for explaining the homogeneity, flatness, and structure formation observed in the cosmos. However, a concrete, physically motivated description of the inflaton potential – the field driving inflation – remains elusive. While simple monomial potentials (like $V(\phi) \propto \phi^n$) have been explored, this paper investigates a broader, well-motivated class of models arising from supersymmetry. These models feature "flat tree-level potentials" where the crucial features emerge from radiative corrections, leading to a logarithmic dependence on the inflaton field. The challenge is to test these more complex, yet theoretically grounded, models against the precision of current cosmological data (CMB and LSS) and compare their predictive power and physical viability against simpler phenomenological models.

Methods

The authors employ a Bayesian model comparison framework to analyze the data. They derive the theoretical predictions for the primordial power spectra (scalar $P_s(k)$ and tensor $P_t(k)$) for two main classes of models:

  1. Logarithmic Regime (LOG): Models with flat tree-level potentials where radiative corrections lead to a logarithmic dependence of the potential on the inflaton field.
  2. Logarithmic Regime + Non-Renormalizable Operator (LOG+NRO): An extension of the LOG models that incorporates new physics beyond a certain energy scale, manifesting as non-renormalizable operators in the inflaton potential.

They use Markov Chain Monte Carlo (MCMC) methods to sample parameter spaces and derive constraints. Model comparison is performed using Bayes factors, which reward good fits while penalizing unnecessary model complexity. They also analyze the spectral index $n(k)$ and its running, $dn/dlnk$, which are key observables from cosmological data.

Datasets

The paper utilizes a combination of observational datasets:

  • Cosmic Microwave Background (CMB) data: WMAP 3-year data, ACBAR, CBI, and BOOMERANG.
  • Large-Scale Structure (LSS) data: SDSS galaxy survey.
  • Hubble Constant measurement: Hubble Space Telescope measurement of $H_0 = 72 \pm 8$ km/s/Mpc.

These datasets are standard for cosmological parameter estimation and power spectrum analysis.

Key Insights and Ideas

  • Predictivity of Flat Potentials: Models with flat tree-level potentials, driven by radiative corrections, are surprisingly predictive. They require very few free parameters to fit the data, making them strong candidates for a fundamental description of inflation.
  • Small Tensor Perturbations: A robust prediction across these models is the smallness of tensor perturbations ($r \lesssim 10^{-3}$), which is a significant constraint for future observational tests.
  • NROs and Running Spectral Index: The inclusion of non-renormalizable operators (NROs) in the LOG+NRO models can naturally produce a sizeable running of the spectral index, particularly at larger scales (smaller $k$), while still achieving a sufficient number of e-folds.
  • Bayesian Model Comparison: The use of Bayesian model comparison is crucial for distinguishing between these physically motivated models and simpler phenomenological parametrizations, accounting for both goodness-of-fit and model complexity.
  • E-folds are Natural: These models naturally predict a reasonable number of e-folds (around 50-60), which is required to solve cosmological problems, without needing to be artificially imposed.

Key Results

  • Both LOG and LOG+NRO models provide excellent fits to CMB and LSS data with very few free parameters.
  • The LOG scenario, with essentially two parameters, implies a nearly constant spectral index $n(k)$.
  • The LOG+NRO scenario can accommodate a running spectral index, with the magnitude and behavior depending on the order of the NRO ($N$).
  • Tensor-to-scalar ratios are predicted to be very small ($r \lesssim 10^{-3}$), making them difficult to detect with current technology but providing a strong test for future observations.
  • Bayesian model comparison suggests that the LOG scenario (with a flat prior on the number of e-folds) and the LOG+NROF scenario (with $N=10$) are the most viable cases based on current data.

Limitations and Future Work

  • Parameter Degeneracies: The paper notes that strong degeneracies exist among the potential parameters in the LOG+NRO scenario, making it difficult to translate constraints into parametrization-independent results for the potential parameters themselves. Principal Component Analysis (PCA) is used to identify well-constrained directions.
  • Model Selection Complexity: Comparing phenomenological parametrizations with physically motivated models using Bayesian evidence is challenging due to the different nature of their parameter spaces and priors.
  • Future Observations: The authors emphasize that future CMB and LSS data, particularly from missions like Planck, will be crucial for further constraining these models and potentially distinguishing between them. The small tensor perturbation predictions also highlight the importance of future B-mode polarization experiments.
  • NRO Order: While the paper explores $N=2$ and $N=10$ for the NRO order, a more comprehensive analysis across a wider range of $N$ values could be beneficial.

Summary updated

Abstract

We use cosmic microwave background and large scale structure data to test a broad and physically well-motivated class of inflationary models: those with flat tree-level potentials (typical in supersymmetry). The non-trivial features of the potential arise from radiative corrections which give a simple logarithmic dependence on the inflaton field, making the models very predictive. We also consider a modified scenario with new physics beyond a certain high-energy cut-off showing up as non-renormalizable operators (NRO) in the inflaton field. We find that both kinds of models fit remarkably well CMB and LSS data, with very few free parameters. Besides, a large part of these models naturally predict a reasonable number of e-folds. A robust feature of these scenarios is the smallness of tensor perturbations (r < 10^{-3}). The NRO case can give a sizeable running of the spectral index while achieving a sufficient number of e-folds. We use Bayesian model comparison tools to assess the relative performance of the models. We believe that these scenarios can be considered as a standard physical class of inflationary models, on a similar footing with monomial potentials.
BibTeX
@article{ballesteros2008flat,
  author = {Guillermo Ballesteros and J.A. Casas and J. R. Espinosa and Roberto Ruiz de Austri and Roberto Trotta},
  title = {{Flat tree-level inflationary potentials in the light of cosmic microwave background and large scale structure data}},
  journal = {Journal of Cosmology and Astroparticle Physics},
  year = {2008},
  doi = {10.1088/1475-7516/2008/03/018},
}
Figure