
Efficient Bayesian synthetic likelihood with whitening transformations
Likelihoodfree methods are an established approach for performing appro...
read it

BSL: An R Package for Efficient Parameter Estimation for SimulationBased Models via Bayesian Synthetic Likelihood
Bayesian synthetic likelihood (BSL) is a popular method for estimating t...
read it

Detecting conflicting summary statistics in likelihoodfree inference
Bayesian likelihoodfree methods implement Bayesian inference using simu...
read it

Bayesian inference using synthetic likelihood: asymptotics and adjustments
Implementing Bayesian inference is often computationally challenging in ...
read it

A Comparison of LikelihoodFree Methods With and Without Summary Statistics
Likelihoodfree methods are useful for parameter estimation of complex m...
read it

On Model Selection with Summary Statistics
Recently, many authors have cast doubts on the validity of ABC model cho...
read it

Approximate Bayesian computation via the energy statistic
Approximate Bayesian computation (ABC) has become an essential part of t...
read it
Parameter inference and model comparison using theoretical predictions from noisy simulations
When inferring unknown parameters or comparing different models, data must be compared to underlying theory. Even if a model has no closedform solution to derive summary statistics, it is often still possible to simulate mock data in order to generate theoretical predictions. For realistic simulations of noisy data, this is identical to drawing realisations of the data from a likelihood distribution. Though the estimated summary statistic from simulated data vectors may be unbiased, the estimator has variance which should be accounted for. We show how to correct the likelihood in the presence of an estimated summary statistic by marginalising over the true summary statistic. For Gaussian likelihoods where the covariance must also be estimated from simulations, we present an alteration to the SellentinHeavens corrected likelihood. We show that excluding the proposed correction leads to an incorrect estimate of the Bayesian evidence with JLA data. The correction is highly relevant for cosmological inference that relies on simulated data for theory (e.g. weak lensing peak statistics and simulated power spectra) and can reduce the number of simulations required.
READ FULL TEXT
Comments
There are no comments yet.