Advice to Physics PhD Applicants
Advice on Preparing to Apply to Physics PhD Programs
This advice will be most useful if you act on it starting in your 2nd or at latest 3rd year of undergraduate studies, but some points are relevant even as you prepare to apply in your final year.
To be a competitive candidate for PhD programs, you'll need to take a demanding set of coursework, particularly in physics and math, and to do well in these classes. At many US universities, this will go beyond the requirements to graduate with a physics major. For example, at Stanford we want to enable students to be able to benefit from the physics approach to problem-solving, whether they will then go into law, journalism, science policy, industrial design, biology, teaching, or many other areas that do not necessarily require the multiple classes on quantum mechanics and advanced math that you'd need to be ready to thrive in a PhD program.
Learning extra math (not just calculus/differential equations and linear algebra, but abstract algebra, topology, differential geometry) is helpful in physics. Even number theory or graph theory can usefully shape your thinking. Beyond the required classes I'd suggest taking classes on material you enjoy and/or with especially excellent pedagogy and/or relevant to research you're doing. Don't worry much about notations on your transcript such as a minor or double-major. If you end up taking enough math to qualify for a double major (or minor), great, and maybe it's worth taking one extra class if you're in striking distance. But that's about it.
Bottom line: Strong grades, and taking classes beyond requirements, are important indications of your capabilities -- necessary, but not sufficient, for admission at the most competitive programs.
Though a knowledge base is important, doing well in the structure of a class says very little about what you'll be like as a researcher. Success in scientific research derives from what you know, what you can do, what you enjoy and are motivated by, how hard you work, how effectively you communicate, ... and luck. Admissions committees are attempting to judge who has a chance of being an outstanding scientist, so they search for evidence of those factors. Prediction of such future success is very hard, and maybe even impossible because of how it depends on future circumstances. A colleague of mine, Carl Wieman, has studied how student scientists learn to become experts. He told me that the best evidence that you can do strong scientific research is having already done strong scientific research.
This requires putting in time and effort, but that's not the whole story. When you start in a research group you'll probably be largely following instructions. But if a recommender is able to say that you eventually came up with your own ideas for what to pursue, or a better way than they thought of to tackle a problem, that's powerful. An equally important reason to engage in research is to get a sense of whether you'll enjoy it, and which type of research you'll enjoy.
3. By the time you apply to PhD programs, you should decide whether you want to pursue experiment or theory -- at most departments, admissions are largely separate for those two, though experimentalists will often do some simulation and simple theory, and will engage with more complex theory done by others. Both areas are highly competitive, but theory even more so. If you think you want to do theory, don't try to game the system by applying to do experiment. If you say you want to do experiment and after you're admitted you try to get a position with a theorist they'll probably say they have no space -- the department already admitted the applicants the theorists could accommodate. Earlier on, if you think you want to do experiment, ideally some of your research experience should be with experiment. If you want to do theory, it's imperative that some of your research experience be focused on theory. But it's also good to get cross experience. And in your first two years of undergrad you're more likely to be able to contribute meaningfully to cutting-edge experiment than cutting-edge theory.
4. Spending 3 years working with one group allows you to come up a learning curve and make meaningful contributions. There's also value in seeing more than one area of research so you learn what activities and ideas appeal to you most, and are likely to sustain your enthusiasm throughout 5-6 years of a PhD. There's a tension between breadth and depth, and no magic formula, but even if at age 18 you think you know what you want to spend your next decade on, don't count on remaining excited about this specific area as you grow personally and intellectually.
5. You will need at least 3 recommenders, of whom at least 2 and ideally all 3 should be able to say more than "The student got an A in my class, having done really well on the exams and problem sets." If you work for an extended time with one group, try to build a relationship with a collaborator and/or do a great job in an open-ended project-based class, to set up other strong recommenders.
6. You should not pursue a PhD simply because you want the credential and the honor (whether the PhD itself, or becoming a professor or famous scientist.) You should relish the process. There will be hard times, and few people can sustain their effort if they're not enjoying it a lot of the time, and really excited to learn the answers to their questions through research.
7. Pretty much every physicist needs to do some coding at some point. With LLMs this is a rapidly changing landscape -- those tools can make coding dramatically faster, but you need to learn how to direct them and how to evaluate whether the program is doing what it should be (e.g. writing and running automated tests; for numerical calculations, checking scaling and limiting behavior). I expect coursework could be useful for this, combined with practice on projects you define. Concrete evidence of your capabilities is more valuable than a mark on a transcript.
8. Standardized tests, particularly the Physics GRE, are also taken seriously. Departments are again starting to require or recommend providing those scores. In my opinion, the problems on the Physics GRE are mainly very simple, with some requiring memorization of trivia. In senior year of a solid undergraduate program at a research university, the main challenges of the test are doing a lot of short problems in a fixed time, learning some of the trivia across different areas of physics, and possibly not having covered stat mech/thermodynamics by the time the test rolls around, requiring you to study this on your own. So a high score cannot demonstrate great promise in physics, but with grade inflation and the rise of LLMs (and the fact that we want to be able to consider students from smaller or lesser-known schools) it's seen as a meaningful data point. Also, doing well on the test typically requires practice, so it demonstrates determination, which is important in research. Hard to predict how this will look 3-4 years from now.
9. Unless you feel you've exhausted the opportunities at your university (unlikely, at a strong research university) I would not advise graduating in 3 years. Instead, take the extra time to do more research, take more advanced courses, study abroad if you wish. Also, a strength of the US university system compared to those in most of the world is the option to take classes outside your major, and I don't just mean math and CS. Pick some classes you're curious about in humanities/arts/social sciences and possibly other natural sciences. This may be your last opportunity to delve deeply in other areas, and it will enrich your intellectual life and maybe even shape your directions in ways you can't predict.
10. In the US -- unlike, for example, in most of Europe -- a student applying to a PhD program does not need to definitely decide which professor to work with. In fact, at any university you're considering applying to you should read webpages and then publications to identify at least a few professors whose work you would find interesting. If you know faculty in a related area, you might ask them for suggestions.