•February 22, 2013 • Leave a Comment
Today a very interesting paper focusing on discovery of relationship between complex astronomical data attributes in large dimensional space has been posted on astroph.
To find a coherent structure in large data volume, a clever machine learning technique is a must and this paper seems to be a very nice starting point.
•January 16, 2013 • Leave a Comment
In the last summer, we (Martin, Neal and I) have tested various methods to improve the accuracy of marginal likelihood for Bayes factor computation. What we’ve found then came out as a paper submitted to a journal and has been posted on astroph today (http://arxiv.org/abs/1301.3156). The methods is simple but provides a remarkably accurate marginal likelihood value. It has been tested up to 16 dimensional space and put a solid foundation for GALPHAT galaxy model selection in the coming papers.
•September 21, 2012 • Leave a Comment
I became a postdoc and start to work on the ALFALFA+SDSS project. Jessica Rosenberg is building up a cross-correlated catalog from blind HI survey ALFALFA and SDSS. Main goal of this project is to study a role of gas in galactic star formation and galaxy evolution, using a large number of HI and optical sample. Today Jessica gave a talk and showed some of preliminary results from the ALFALFA+SDSS. Although it was interesting to see that the gas-to-star ratio for galaxy pair is less than that for the entire ALFALFA+SDSS sample on average, a careful investigation of the sample selection function should be done for a reliable inference.
•April 9, 2012 • Leave a Comment
I’ve found a minor typo error in GALPHAT paper (2011, MNRAS 414, 1625).
In equation (7), “k = 0.01945-0.8902n + 10.95n*n – 19.67*n*n + 13.43*n*n*n” should be “k = 0.01945-0.8902n + 10.95n*n – 19.67*n*n*n + 13.43*n*n*n*n”. The code is safe from this typo.
•March 30, 2012 • Leave a Comment
Sean Carroll speaks very clearly and concisely about a modern view of our Universe. Worth to spent 15 min
•March 28, 2012 • Leave a Comment
As a continuing work based on the heat bath paper that I published, I am working on a simple toy model of self-gravitating system at the centre of a deep gravitational potential well, which behaves as a thermal heat bath and increases a velocity dispersion of the embedded system. Then this system is not bound by the system’s self-gravity but bound by the surrounding external potential well. I derived an equilibrium density profile of the system and compared to the Coma cluster galaxy radial distribution. Based on a ‘chi^2 by eye’, it gives an equally good fit to the data as King model does. However this self-gravitating system surrounded by a deep potential well is more realistic description of the cluster galaxies than King model. I am preparing a paper to introduce the new model with some theoretical backgrounds and observational applications.
•March 20, 2012 • Leave a Comment
A new Bayesian inference tool (BIE) has been published on astro-ph today (http://arxiv.org/abs/1203.3816). It has several advanced MCMC algorithms to sample the posterior and some stand alone programs to test a convergence and to compute a marginal likelihood for Bayesian model selection. It is designed for a general purpose Bayesian inference and is possible to add any user likelihood module to solve one’s own problem. It has potentially many interesting applications not limited to astronomy. Two case studies (Bayesian semi-analytic galaxy formation model and GALPHAT) have been introduced in this paper.