Cosmological constraints from the DESI DR1 Bispectrum Full-Shape and DR2 BAO

Abstract

We present cosmological constraints from the combination of DESI DR1 full-shape measurements, including for the LRG bispectrum, and DESI DR2 BAO data. The joint analysis accounts for cross-covariance using mocks, while ShapeFit compression mitigates prior volume effects that hinder beyond-$\hat{I}›$CDM analyses. In $\hat{I}›$CDM, the bispectrum (P+B) shifts $\ddot{I}{ƒ}_8$ up by $1.1\ddot{I}ƒ$ and $S_8$ by $1.2\ddot{I}ƒ$, reducing their uncertainties by $26\%$ and $28\%$, respectively. For $w_0{w}_a$CDM, DESI-only analyses with the bispectrum shift dark energy parameters toward $\hat{I}›$CDM, staying consistent with a cosmological constant within $1\ddot{I}ƒ$. Adding CMB creates a preference for evolving dark energy: DESI+CMB (P+B) shows a $2.8\ddot{I}ƒ$ deviation from $\hat{I}›$CDM. Including DES-Dovekie supernovae alone reduces this to $1.6\ddot{I}ƒ$, while the full combination DESI+CMB+DES-Dovekie gives $3.1\ddot{I}ƒ$, driven primarily by the CMB. The bispectrum consistently weakens evidence for time-varying dark energy relative to power-spectrum-only analyses. The bispectrum also enhances sensitivity to massive neutrinos: in DESI-only analysis, the power-spectrum-only posterior for $\sum m_{\hat{I}}½$ is consistent with zero, whereas adding the bispectrum yields a mean of $0.26\pm 0.17$~eV and a $95\%$ upper limit of $0.57$~eV, shifting the peak into the positive region and agreeing with oscillation lower bounds. For modified gravity, the bispectrum further constrains $\hat{I}{¼}_0=0.12\pm 0.49$ from DESI-only data, consistent with general relativity. Our analysis shows that accounting for cross-dataset covariances and avoiding prior volume effects yields robust constraints, with the bispectrum raising amplitude parameters and tightening their uncertainties.

32 pages, 10 figures, submitted to JCAP