Skip to Main Content
 

Global Search Box

 
 
 
 

ETD Abstract Container

Abstract Header

Bayesian Model Mixing for Extrapolation from an EFT Toy

Abstract Details

2021, Bachelor of Science (BS), Ohio University, Physics.
I test methods for extrapolating from limited data through Bayesian modeling on a simplified model of Effective Field Theory, or "Toy." I fit finite polynomial models not only based on data, but on the Bayesian prior understanding of the parameters in this polynomial having "naturalness," or being of order 1. These polynomials are represented by probability distributions conditional on the degree of the polynomial and the naturalness hyperparameter. After generating a set of polynomial models, I perform Bayesian Model Averaging to create a "mixed model," which is a weighted summation of the individual models where the weights are the evidence that each model fits the data and prior. I test how these mixed models extrapolate when compared to each individual non-mixed model. Given a set of models that are useful for extrapolating, the mixed model performs better than a naively-selected non-mixed model.
Daniel Phillips (Advisor)
65 p.

Recommended Citations

Citations

  • Connell, M. (2021). Bayesian Model Mixing for Extrapolation from an EFT Toy [Undergraduate thesis, Ohio University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ouhonors1619122381888487

    APA Style (7th edition)

  • Connell, Matthew. Bayesian Model Mixing for Extrapolation from an EFT Toy. 2021. Ohio University, Undergraduate thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ouhonors1619122381888487.

    MLA Style (8th edition)

  • Connell, Matthew. "Bayesian Model Mixing for Extrapolation from an EFT Toy." Undergraduate thesis, Ohio University, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=ouhonors1619122381888487

    Chicago Manual of Style (17th edition)