I am doing a Masters in theoretical physics. It's been half a year since I've begun my Masters.

Recently, I find myself intrigued by machine learning and would like to do a PhD in machine learning after I finish my Masters in theoretical physics.

I am rather afraid to break it to my supervisor that I plan to move into machine learning after my Masters. I am also afraid how he might react if he finds out that I am taking courses in the Computer Science department. I plan to work with machine learning profs to get two recommendation letters, but I guess I must have a third letter from my physics supervisor. I am afraid that he will not write me a good letter for PhD admission into machine learning.

What should I do?

  • 2
    Just a thought, but can you shift your physics research to focus on machine learning? A physics PhD would be highly competitive in this field and showing that you can apply ML to a different field would be great.
    – Hobbes
    Jan 31, 2017 at 16:36
  • 2
    What would a PhD in machine learning allow you to do is the real question. I would highly recommend you find ways to integrate machine learning into a field of study and not the other way around. Machine learning is after all, common knowledge at this point, it is a bandwagon, but physics at the PhD level is rare, even rarer are people who find use of ML in physics related applications
    – Fraïssé
    Feb 1, 2017 at 3:18
  • The machine learning is also an issue in theoretical physics, for example in [quantum setting][1] [1]: uibk.ac.at/th-physik/qic-group/research/…
    – Agnieszka
    Sep 4, 2020 at 15:31

1 Answer 1


I would suggest that you make as smooth and timely a transition as you can. The more dramatic a change you make, the more damage you risk doing to your network of relationships. Tell your advisor about your interests while making clear that you remain committed to the program you're in.

Now, there are some advantageous tactics for studying several fields at once. First, modulate your commitments -- if half your courseload is in another department, that might be too much. Second, combine them wherever you can. Certainly you could train a neural network on solid models to recognize favorable heat dissipation, say.

  • Well, my supervisor's on sabbatical, so I'm planning to ask him for permission by email if, this semester, I could take a course or two in machine learning in the computer science department to pursue my interest in machine learning in my free time. I plan to talk to him in detail about my desire to switch to machine learning during summer when he comes back to campus. What do you think?
    – user22613
    Jan 1, 2017 at 9:48
  • Also, I plan to tell him in the summer that I intend to work with machine learning profs in my free time to get a paper published to help with my admission to a reputable PhD program in machine learning. In the meantime, I plan to work on my supervisor's project in physics and get a paper published because that's what he would like. I plan to get two letters of recommendation from the machine learning profs I intend to work with. Do I have to get a letter of recommendation from my physics supervisor for PhD admission in machine learning?
    – user22613
    Jan 1, 2017 at 9:51
  • 1
    @failexam - How many physics courses do you plan to enroll in for spring semester? Two non-physics courses in one semester seems like a lot. // One machine learning letter may be enough; yes, it would be helpful to have a letter from your physics supervisor; are you finding your physics studies and research sufficiently rewarding to make finishing out your master's worthwhile? // Your fear of alienating your physics supervisor makes me wonder whether you are thinking of him in the wrong way. I can imagine someone's father reacting badly to news of a planned switch... Jan 2, 2017 at 19:55
  • 1
    of fields, but an advisor/mentor would normally have more detachment than a family member. But I don't know your supervisor. Maybe he is not sufficiently detached.... // Once you have your game plan, you may want to ask for a phone or Skype conference and discuss your plan with him live rather than by email. If he asks for the topic or reason for the phone appointment, keep the plan veiled (e.g. say, "I'd like to talk over my spring course choices with you"; if he says, "Nay, let's do it by email," tell him you'd prefer a phone or Skype conference, if he would be so kind; repeat as needed). Jan 2, 2017 at 19:56
  • 1
    You might also want to check in with the dean of graduate studies in the two departments before and after the phone call. Note, getting the spring ratio of ML courses to physics courses right is an important part of getting your game plan ready before you speak with your supervisor. Jan 2, 2017 at 20:02

You must log in to answer this question.