Date of Award


Document Type


Degree Name

Bachelor of Science



First Advisor

Emory Bunn


In the search for effective processes to estimate E and B spectra from polarized data, Gibbs Sampling has proven to be a powerful method. In the search for B modes, it is essential to avoid a false positive detection due to contamination from the larger E component. It is therefore of interest to combine Gibbs sampling with methods to “purify” the B modes, ensuring that a B-mode detection is robust. This goal can be achieved by compelling the Gibbs Sampler to estimate a pure B spectrum. The method we chose to implement involves an artificially inflated E spectrum, which “forces” the sampler to prefer to classify all ambiguous modes as E modes. If this method is to be useful, the resulting bias in the B spectrum must be small, and any increase in the error as compared to the “standard” sampler must be acceptable. We present Gibbs sampling analyses of simulated polarized sky data, in which we quantify these effects.