Pedagogic Null Tests of Dynamical Dark Energy Hints: Reconstructing LambdaCDM with Consistent BAO, CMB, and SNe Mocks

TL;DR

This study uses realistic mock datasets to perform a null test on dark energy models, demonstrating that apparent deviations can arise from dataset degeneracies and likelihood geometry rather than true physics.

Contribution

It introduces a controlled null test with comprehensive mock datasets to clarify how dataset degeneracies can mimic evidence for dynamical dark energy.

Findings

Individual probes show apparent deviations in (w0,wa) space.

Two-probe combinations with supernovae recover fiducial parameters but lack absolute scale.

Full multi-probe likelihood breaks degeneracies and accurately recovers the true cosmology.

Abstract

Hints of a dynamical dark-energy equation of state have appeared in several combined cosmological probes. However, such indications may instead arise from the intrinsic likelihood geometry of individual datasets, residual inter-probe tension, or restrictive priors. These factors can mimic evidence for dynamical dark energy. To clarify these issues, we perform a controlled null test. We use realistic mock BAO, CMB, and Type~Ia supernova datasets generated from a common fiducial LambdaCDM cosmology. These mocks include the DESI~DR2 BAO covariance, the Planck~2018 distance-prior covariance, and the full Pantheon+ SH0ES supernova covariance. This setup isolates physical information from geometric or statistical effects in the CPL parametrization. We find that individual probes and most two-probe combinations show apparent deviations in the (w0,wa) plane. These could be mistaken for phantom crossing or evolving dark energy. Two-probe combinations including supernovae (BAO+SNe, CMB+SNe) recover values near (w0,wa)=(-1,0) but fail to reconstruct (Omegam0,H0), because SNe do not determine the absolute distance scale. Combinations without SNe (BAO+CMB), as well as any single dataset, retain strong degeneracy directions. These produce significant shifts driven purely by likelihood geometry. These behaviors arise because BAO, CMB, and SNe each constrain only one principal direction in (w0,wa) space. Their degeneracy ridges are misaligned due to distinct redshift sensitivities. In contrast, the full BAO+CMB+SNe likelihood with proper covariance breaks all degeneracies simultaneously. It cleanly recovers the fiducial cosmology, including (w0,wa)=(-1,0) and (Omegam0,H0). Our results provide a transparent benchmark for assessing future claims of omega(z) neq -1. They emphasize the need for complete multi-probe analyses with flexible H0 and rd priors.