Astrostatistics Image Penn State University Eberly College of Science Center for Astrostatistics Center for Astrostatistics

 2014 Summer Events | Travel & Visa | Lodging | Registration | Program

Summer School in Statistics for Astronomers X (June 2-6, 2014)
Statistical Modeling of Cosmic Populations (June 9-10, 2014)
Bayesian Computing for Astronomical Data Analysis (June 11-13, 2014)

Map     Restaurants

Penn State is continuing its annual Summer Schools in Statistics designed for graduate students and researchers in astronomy. The tenth summer school is an intensive week covering basic statistical inference, several fields of applied statistics, and the R computing environment.

A repertoire of well-established techniques applicable to observational astronomy and physics are developed. Classroom instruction is interspersed with hands-on analysis of astronomical data using the open-source R software package. The course is taught by a team of statistics and astronomy professors with opportunity for discussion of methodological issues. The topics covered include:

  • Exploratory data analysis
  • Hypothesis testing and parameter estimation
  • Regression
  • Bootstrap resampling
  • Model selection & validation
  • Maximum likelihood methods
  • Non-parametric methods
  • Multivariate methods
  • Clustering and classification
  • MCMC
  • Bayesian Analysis
  • Spatial Statistics
  • Time series

The 2014 Summer School will be modeled on the last nine Penn State Summer Schools and the three Indian Institute of Astrophysics-Penn State Summer Schools; see 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012 and 2013 lecture notes for the Penn State Summer Schools. The summer schools are partially supported by the National Science Foundation.


Summer School Registration is closed

Registration Deadline for the events, Statistical Modeling of Cosmic Populations, & Bayesian Computing for Astronomical Data Analysis: May 9, 2014 or earlier if the enrollment limit reaches.

Statistical Modeling of Cosmic Populations:   This two-day session will introduce participants to statistical methods for demographic modeling of cosmic populations, with an emphasis on Bayesian methods based on multilevel models (MLMs) with separate levels of stochastic modeling that explicitly account for uncertainties due to population sampling and measurement errors. Topics to be covered include:

  • Selection effects and measurement errors in astronomical surveys
  • Eddington, Malmquist, and Lutz-Kelker biases/distortions
  • Estimation of observable number-size/number count distributions ("log N - log S")
  • Multilevel models (e.g., empirical and hierarchical Bayesian methods) for estimation of intrinsic distributions, such as luminosity functions for stars and galaxies or distributions of masses and orbital parameters for exoplanet populations
  • Computational methods: Laplace approximation, cubature, Markov chain Monte Carlo (MCMC), approximate Bayesian computation (ABC)
The computational component will be designed as a lead-in to the following session on high-performance statistical computing. Participants interested in cosmic demographics with large datasets are especially encouraged to attend both sessions.

Bayesian Computing for Astronomical Data Analysis:   High-Performance Computing for Astronomical Data Analysis & Bayesian Computing (June 11-13, 2014) will prepare students to harness modern parallel computing hardware for analyzing astronomical data, with an emphasis on applying Bayesian inference to multilevel models. While lectures will include examples from a variety of astronomical applications, the lab exercises will use astronomical data sets and Bayesian multi-level models motivated by the prior school on Statistical Modeling of Cosmic Populations (June 9-10, 2014), so those participating in both parts of the school will experience a seamless transition. By comparing the performance of common Bayesian algorithms, students will develop intuition to help choose the most appropriate algorithms, architectures and programming model for their own research applications.

Lessons will be targeted at graduate students and postdocs in astronomy, astrophysics and physics. Senior researchers interested in learning about Bayesian computing are also welcome to attend. Prospective participants who are not already familiar statistics at the level of an undergraduate course, should participate in the Summer School in Statistics for Astronomers (June 2-6, 2014). Prospective participants not already familiar with Bayesian data anlaysis and multi-level modeling, should also participate in the Statistical Modeling of Cosmic Populations (June 9-10, 2014).

Preliminary Program:   The Morning lectures will describe:

  • the challenges of performing Bayesian computations on multi-level models,
  • the performance characteristics of modern computing architectures and memory systems,
  • parallel programming tools and practices, and
  • case studies of applications of parallelized codes for astronomical data analysis.
In afternoon lab sessions, participants will gain hands-on experience performing Bayesian computations using multiple computer architectures and programming models, such as multi-core systems, distributed memory clusters, hardware accelerators (e.g., graphical processing units) and cloud computing. Example applications in afternoon computer lab sessions are likely to apply techniques such as importance sampling, sequential Monte Carlo, population-based Markov chain Monte Carlo (MCMC) and approximate Bayesian Computing (ABC) to multi-level models for cosmic populations that many participants will have learned about during the previous workshop (June 9-10). This session is partially supported by the Penn State's Institute for Cyber Science.

In addition, all the three 2014 summer events are partially supported by the Penn State departments of Statistics, and Astronomy & Astrophysics.


  • For the 10th school, Eric Feigelson, Dept. of Astronomy & Astrophysics, Penn State University, or G. Jogesh Babu, Dept. of Statistics, Penn State University,
  • For Statistical Modeling of Cosmic Populations, Tom Loredo, Dept. of Astronomy, Cornell University
  • For Bayesian Computing for Astronomical Data Analysis, Eric Ford, Dept. of Astronomy & Astrophysics, Penn State University

Local information:
The 2014 summer school will be held on the Pennsylvania State University's University Park Campus located in State College, Pennsylvania, USA. The town of State College and the university campus combine to offer a relaxed, college town atmosphere with many shops, restaurants and points of interest. Recreational opportunities abound including fine golf courses, tennis courts, gymnasiums and swimming facilities. For maps and tourism information visit the Central Pennsylvania Convention & Visitors Bureau online.

NSF Department of StatisticsEberly College of ScienceDepartment of Astronomy and Astrophysics