I will graduate this semester with a bachelor's degree in physics (in the US). I applied for PhD programs in physics, but have come to realize (rather late in the process) that I want to pursue a PhD in mathematics. However, I was accepted into two physics PhD programs and received the NSF GRFP. These two schools have respectable math departments.
As for my background in math, I have taken much of the undergraduate curriculum (linear algebra, algebra, complex variables, Lie groups, two differential geometry courses, algebraic topology); I have also taken around 10 graduate courses in physics, including quantum field theory. But I am missing core coursework in analysis. I am interested in studying low-dimensional topology, gauge theory, mathematical physics, and related topics in math grad school.
Technical details about the NSF fellowship (more here):
- It can be transferred between institutions (e.g., from master's at one school to PhD at another).
- Changes in the field of study can be proposed after the first year, and require approval by the NSF. My awarded field of study is condensed matter physics.
- It cannot be deferred; once accepted, I must enroll in a graduate program this fall. Or I can decline it and re-apply in the future.
- Once accepted, I can choose to use the funding for any 3 of the next 5 years.
Given these constraints, I am considering the following options:
- Accept an offer at a physics PhD program. Re-apply to mathematics PhD programs after 1-2 years, leaving the physics program with a master's degree. Defer the NSF funding for two years, until beginning a math PhD program.
- Apply to math master's programs still taking applicants. Use part of the NSF fellowship for the master's. Apply after 1-2 years of the master's program to math PhD programs.
- Decline the NSF fellowship and take a gap year before applying to math master's or PhD programs.
Given my circumstances, do you think it is wise to pursue one of the options above for graduate study? Any other suggestions are welcome. (I am also not completely savvy with the NSF guidelines sketched above, so please correct any misconceptions.)