In this dissertation I will start with the broad research question of what does problem solving in upper division physics look like? My focus in this study is on students' problem solving in physics theory courses. Some mathematical formalisms are common across all physics core courses such as using the process of separation of variables, doing Taylor series, or using the orthogonality properties of mathematical functions to set terms equal to zero. However, there are slight differences in their use of these mathematical formalisms across different courses, possibly because of how students map different physical systems to these processes. Thus, my first main research question aims to answer how students perform these recurring processes across upper division physics courses.

I break this broad question into three particular research questions: What knowledge pieces do students use to make connections between physics and procedural math? How do students use their knowledge pieces coherently to provide reasoning strategies in estimation problems? How do students look ahead into the problem to read the information out of the physical scenario to align their use of math in physics?

Building on the previous body of the literature, I will use the theory family of Knowledge in Pieces and provide evidence to expand this theoretical foundation. I will compare my study with previous studies and provide suggestions on how to generalize these theory expansions for future use. My experimental data mostly come from video-based classroom data. Students in groups of 2-4 students solve in-class problems in quantum mechanics and electromagnetic fields 1 courses collaboratively. In addition, I will analyze clinical interviews to demonstrate how a single case study student plays an epistemic game to estimate the total energy in a hurricane.

My second research question is more focused on a particular instructional context. How do students frame problem solving in quantum mechanics? I will lay out a new theoretical framework based in epistemic framing that separates the problem solving space into four frames divided along two axes. The first axis models students' framing in math and physics, expanded through the second axis of conceptual problem solving and algorithmic problem solving. I use this framework to show how students navigate problem solving. Lastly, I will use this developed framework to interpret existing difficulties in quantum mechanics.