The paper, "Nonparametric Inference for the Cosmic Microwave Background," was published in the journal Statistical Science. The paper describes statistical techniques that Genovese, Wasserman and colleagues developed to analyze recent detailed measurements of the CMB by the Wilkinson Microwave Anisotropy Probe (WMAP). The techniques in the paper allow scientists to assess the strength of the conclusions that can be drawn from the data about cosmological questions and can help in fine-tuning the design of future data collection.
"The WMAP data provide a very solid confirmation of the Big Bang, but uncertainties remain about some of the more specific inferences that might be drawn from the data," Genovese said.
The primordial photons of the CMB permeate the universe at a temperature less than three degrees above absolute zero, cooled from billions of degrees by the expansion of the universe. Measuring the temperature of this radiation across the sky is as close as we can get to a snapshot of the early universe. Understanding the CMB offers insights into the shape, composition and eventual fate of the universe and offers clues to how the distribution of matter and energy in the universe became so clumpy--what cosmologists call large-scale structure. The latest analysis indicates a universe that is roughly 13.7 billion years old with a flat shape and composed mostly of mysterious "dark energy" (73 percent) and "dark matter" (23 percent). These conclusions are consistent with the results from studies of supernovae and galaxy clusters.
Theory predicts that at the largest scales, the CMB will appear the same in every direction. But we know that it cannot be completely uniform because otherwise there would be no irregularities to seed the formation of structure (e.g., galaxies, clusters of galaxies) that we see today. Until the early 1990s, however, this irregularity, or anisotropy, in the CMB eluded detection. Recent observations using more sophisticated instruments have picked up fluctuations in the CMB temperature with unprecedented precision. But translating the data into conclusions about cosmology requires complex data processing and statistical assumptions. Physicists have based their conclusions on procedures, models and other data sets whose effect on the conclusions is unknown.
That's where Genovese, Wasserman and their astrostatistics group come in. Genovese, Wasserman and their team applied refined statistical methods to determine which models were accurate and what could be inferred about the unknown CMB spectrum and in turn, the underlying cosmology.
"One advantage to our approach is that it allows one to separate the information in the data from the information in a model. We could use the cosmological model to generate spectra, but then test which spectra are consistent with the data," Genovese said.
The other researchers who contributed to the paper were Christopher Miller, assistant astronomer at the Cerro Tollolo Inter-American Observatory in Chile; Robert C. Nichol, a professor of physics at the University of Portsmouth in the United Kingdom; Mihir Arjunwadkar, a postdoctoral researcher at the University of Pune in India; and Larry Wasserman, a professor of statistics at Carnegie Mellon.