Exploiting non-Gaussian information to better understand the first stars and galaxies
Wed 3 July, 2019 @2:15 PM, level 7, David Caro Building
Dr Catherine Watkinson, Imperial College London
The SKA and HERA are expected to detect the 21cm line of neutral hydrogen from the high-z Universe with great precision. This should allow us to learn about early generations of stars and galaxies, as we observe the impact of their radiation through ionisations and heating of hydrogen in the intergalactic medium.
As theorists, the challenge we face is how to best interpret these observations. Much work has already been done to develop MCMC pipelines using fast semi-numerical simulations. The current pipeline 21cmMC only includes the power spectrum as a metric to compare how similar a given simulation is to the data. I will discuss the benefits of including statistics that are sensitive to non-Gaussianity in the data.
To facilitate this discussion, I will provide an intuitive explanation of the bispectrum, why it might be an interesting statistic for many datasets, and how it is easier to measure than is often thought. I will also touch on some work I have been doing on likelihood-free parameter estimation which allows the researcher to concentrate on forward modelling their problem, rather than working out the appropriate form for the likelihood.