Kilonova Light Curve Parameter Estimation Using Likelihood-Free Inference
We present a parameter estimation algorithm on kilonova light curves using likelihood-free inference. Our inference is optimized through a pre-trained embedding network that marginalizes the time of arrival and the luminosity distance of the signal. We find that parameter inference utilizing a pre-trained embedding outperforms the use of likelihood-free inference alone, reducing training time and offering the capability to marginalize over certain nuisance parameters. The model is capable of retrieving the intrinsic parameters of the kilonova light curves with a comparable accuracy and precision to nested sampling methods while taking significantly less computational time. This framework has been integrated into the publicly available Nuclear Multi-Messenger Astronomy codebase so users can leverage the model for their inference purposes. This algorithm is broadly applicable to parameterized or simulated light curves of other transient objects.
PDF Abstract