Title: New constraint on the tensor-to-scalar ratio from the $Planck$ and BICEP/Keck Array data using the profile likelihood

Abstract: We derive a new upper bound on the tensor-to-scalar ratio parameter $r$ using the frequentist profile likelihood method. We vary all the relevant cosmological parameters of the $\Lambda$CDM model, as well as the nuisance parameters. Unlike the Bayesian analysis using Markov Chain Monte Carlo (MCMC), our analysis is independent of the choice of priors. Using $Planck$ Public Release 4, BICEP/Keck Array 2018, $Planck$ CMB lensing, and BAO data, we find an upper limit of $r<0.037$ at 95% C.L., similar to the Bayesian MCMC result of $r<0.038$ for a flat prior on $r$ and a conditioned $Planck$ lowlEB covariance matrix.