Abstract:The frequentist method of profile likelihoods has recently received renewed attention in the field of cosmology. This is because the results of inferences based on the latter may differ from those of Bayesian inferences, either because of prior choices or because of non-Gaussianity in the likelihood function. Consequently, both methods are required for a fully nuanced analysis. However, in the last decades, cosmological parameter estimation has largely been dominated by Bayesian statistics due to the numerical complexity of constructing profile likelihoods, arising mainly from the need for a large number of gradient-free optimisations of the likelihood function. In this paper, we show how to accommodate the computational requirements of profile likelihoods using the publicly available neural network framework CONNECT together with a novel modification of the gradient-based $basin$-$hopping$ optimisation algorithm. Apart from the reduced evaluation time of the likelihood due to the neural network, we also achieve an additional speed-up of 1$-$2 orders of magnitude compared to profile likelihoods computed with the gradient-free method of $simulated$ $annealing$, with excellent agreement between the two. This allows for the production of typical triangle plots normally associated with Bayesian marginalisation within cosmology (and previously unachievable using likelihood maximisation because of the prohibitive computational cost). We have tested the setup on three cosmological models: the $\Lambda$CDM model, an extension with varying neutrino mass, and finally a decaying cold dark matter model. Given the default precision settings in CONNECT, we achieve a high precision in $\chi^2$ with a difference to the results obtained by CLASS of $\Delta\chi^2\approx0.2$ (and, importantly, without any bias in inferred parameter values) $-$ easily good enough for profile likelihood analyses.

Calculating Bayesian evidence for inflationary models using CONNECT

Fast and effortless computation of profile likelihoods using CONNECT

CONNECT: A neural network based framework for emulating cosmological observables and cosmological parameter inference

The future of cosmological likelihood-based inference: accelerated high-dimensional parameter estimation and model comparison

Bayesian model comparison in cosmology with Population Monte Carlo

Cosmic Inflation at the Crossroads

Limitations of Bayesian Evidence Applied to Cosmology

The distribution of Bayes' ratio

Bayesian evidence and model selection approach for time-dependent dark energy

Profile Likelihoods in Cosmology: When, Why and How illustrated with $Λ$CDM, Massive Neutrinos and Dark Energy

Evaluating extensions to LCDM: an application of Bayesian model averaging and selection

Observational predictions of some inflationary models

Bayesian evidence for $α$-attractor dark energy models

Field-level simulation-based inference of galaxy clustering with convolutional neural networks

Cosmological Parameter Estimation with a Joint-Likelihood Analysis of the Cosmic Microwave Background and Big Bang Nucleosynthesis

Cosmological inference from Bayesian forward modelling of deep galaxy redshift surveys

Cosmoglobe: Towards end-to-end CMB cosmological parameter estimation without likelihood approximations

Comparison of Bayesian inference methods using the Loreli II database of hydro-radiative simulations of the 21-cm signal

EFTofLSS meets simulation-based inference: σ 8 from biased tracers

Exploring the likelihood of the 21-cm power spectrum with simulation-based inference