Inflation at the End of 2025: Constraints on $r$ and $n_s$ Using the Latest CMB and BAO Data

Problem Statement
Inflation provides the initial conditions for the standard cosmological model (ΛCDM), solving the horizon, flatness, and magnetic monopole problems while predicting small, Gaussian, adiabatic perturbations. A critical aspect of testing inflationary models involves constraining two primary parameters: the tensor-to-scalar ratio ($r$), which quantifies the amplitude of primordial gravitational waves, and the scalar spectral index ($n_s$), which describes the scale dependence of initial density fluctuations. While previous data have constrained these parameters, the release of new datasets in 2025 from the South Pole Telescope (SPT), Atacama Cosmology Telescope (ACT), and the Dark Energy Spectroscopic Instrument (DESI) necessitates a re-evaluation of the current state of constraints. The paper aims to present the latest constraints on $r$ and $n_s$ using these datasets combined with Planck and BICEP/Keck data, serving as a reference for the community rather than a comprehensive review of the entire inflationary model space.

Methodology
The authors analyze the ΛCDM model extended with a varying tensor-to-scalar ratio $r$ (at a pivot scale of 0.05 Mpc$^{-1}$). The tensor power spectrum tilt is fixed to obey the self-consistency condition for single-field slow-roll inflation. The model is parametrized by the amplitude of scalar perturbations ($\log(10^{10} A_s)$), the spectral index ($n_s$), physical baryon and dark matter densities ($\Omega_b h^2, \Omega_c h^2$), the Hubble constant ($H_0$), and the optical depth to reionization ($\tau_{reio}$).

The analysis utilizes the following datasets:

• CMB Data: A "SPA" combination of primary CMB (TT/TE/EE) and CMB lensing ($\phi\phi$) measurements from Planck (PR3/PR4, restricted to $\ell < 1000$ in TT and $\ell < 600$ in TE/EE to avoid double counting), SPT-3G (D1), and ACT (DR6).
• Polarization Data: The latest BICEP/Keck B-mode (BB) likelihood.
• BAO Data: Measurements from the second data release (DR2) of the DESI collaboration.

The authors perform Markov Chain Monte Carlo (MCMC) analyses using the Cobaya sampler and the Boltzmann solver CLASS. For $n_s$, a Planck-based prior on $\tau_{reio}$ ($N \sim (0.051, 0.0062)$) is imposed due to the lack of large-scale E-mode data in the CMB combination. Uniform priors are applied to other cosmological parameters. The authors report 95% upper limits for $r$ and mean values with 68% confidence intervals for $n_s$.

Key Results

1. CMB + BICEP/Keck Constraints: Using the SPA+BK combination, the authors derive:

   • $r < 0.034$ (95% upper limit).
   • $n_s = 0.9682 \pm 0.0032$.
     These results are consistent with the official BICEP/Keck results, with the slight difference in the upper limit attributed to the imposition of the self-consistency relation for the tensor tilt and numerical precision. The addition of SPT and ACT data does not measurably impact the $r$ constraint.

2. Impact of DESI BAO Data: Adding DESI DR2 data to the CMB datasets yields:

   • $r < 0.035$.
   • $n_s = 0.9728 \pm 0.0029$.
     The addition of DESI data causes a statistically significant upward shift in the central value of $n_s$. This shift arises because the CMB constraint on $n_s$ is correlated with parameters ($\Omega_m$ and $h r_d$) that are tightly constrained by BAO data. The CMB and DESI results show a marginal tension (2.8 $\sigma$) in the $\Omega_m$-$h r_d$ plane, which acts as a lever arm to shift the inferred $n_s$.

3. Model Comparisons:

   • CMB-only: The constraints favor regions compatible with Starobinsky $R^2$ inflation ($N_\star=51$) and Higgs inflation ($N_\star=55$) at $\le 2.0 \sigma$.
   • CMB + DESI: The shifted $n_s$ value favors monomial inflaton potentials with $N_\star \sim 50$ (specifically $V(\phi) \propto \phi^{1/3}$). Under these conditions, the prediction for Starobinsky $R^2$ inflation differs from the data at the 3.9 $\sigma$ level, while Higgs inflation remains compatible at 2.9 $\sigma$.
   • $\alpha$-Attractors: The authors demonstrate that a class of polynomial $\alpha$-attractor models ($V(\phi) = V_0|\phi|^k/(\mu^k + |\phi|^k)$ with $k=2$) can predict $n_s$ values consistent with both the CMB-only and CMB+DESI results, depending on the number of e-folds ($N_\star$).

Significance and Claims
The paper positions itself as a reference for the current state of inflationary constraints at the end of 2025. Its primary significance lies in:

• Synthesizing New Data: Integrating the latest releases from SPT, ACT, and DESI with Planck and BICEP/Keck to provide updated constraints on $r$ and $n_s$.
• Highlighting Tension: Explicitly noting that the shift in $n_s$ upon adding DESI data is driven by marginally significant differences between CMB and BAO data that remain unexplained within the standard model. The authors caution that while these differences may be statistical, they could also indicate systematic errors or new physics.
• Resource Availability: The authors make the data products, visualization tools, and code publicly available to allow the community to compare these constraints against any inflationary model.
• Future Outlook: The paper notes that while future CMB missions (South Pole Observatory, Simons Observatory, LiteBIRD) will improve sensitivity to $r$ (targeting $\sigma(r) \approx 10^{-3}$), robust constraints on $n_s$ are equally crucial for differentiating between models. The authors suggest that future data from DESI and the Euclid satellite may either resolve the current CMB-BAO tension or elevate it to statistical significance, thereby refining the analysis of inflationary models.