Lots of free time. Things to read/study before grad school?

[Lavabug]

So I'll be starting a Physics PhD program in the fall, August to be precise. Now I'm the type of person that would go and self-study whatever topic I found interesting (currently working through a reading list in GR, even though it's not a field I'm going into), but I am wondering if there are any specific topics in math or physics that are worth going into deeply that would be of any immediate benefit for a PhD across physics disciplines.

This question is mostly geared towards PhD grads or post 1st year grad students: in retrospect, what subjects do you wish you had spent time self-studying before your 1st year in grad school that would have been of practical benefit to you and your work? Ie: getting a rigorous foundation in statistics/probability, learning advanced techniques for PDE's, diff. geometry or getting a head start with Jackson's E&M?

Some ideas I've thought of/already acted upon are:
-picking up grad/qualifying exam problem collection book(s) and/or upper level undergrad books and working through them
-Improving programming skill, something I am still working on sporadically now. Maybe find a course or project with lots of programming challenges at my level to solve?
-Tutoring math/physics here and there, so that's already helping me to not forget the basics (got a few students already)
-Taking apart and fixing electrical appliances around the house (done 2 and counting), maybe find an interesting diy project to work on my electronics/soldering skills (I've built a few tranny amps, I am interested in building a valve/tube amp, but not with scientific applications in mind).
-I'm also looking for a job of any sort for the coming months and though I'm not picky by any means, I am keeping my eyes peeled for anything that would teach something valuable (data analysis/entry, clerical, but no luck). I've already been in the retail grind years ago and don't really think I have anything left to learn there (but would obviously take the job if I get called).

 

[micromass]

I wouldn't be able to help you much, since I'm in math instead of physics. But perhaps it's a good idea to tell us what specialty you're going into? Is it like theoretical? Computational? Experimental?

Computer programming is always a valuable skill though, you'll definitely need it somewhere later in life, even if it's not in grad school.

 

[Lavabug]

I wouldn't be able to help you much, since I'm in math instead of physics. But perhaps it's a good idea to tell us what specialty you're going into? Is it like theoretical? Computational? Experimental?

Computer programming is always a valuable skill though, you'll definitely need it somewhere later in life, even if it's not in grad school.

Possibly all three, as I intend to experiment in my first and second years as much as I can. I never got any experimental research experience as an undergrad so I am looking to remedy that in particular, but I'm also interested in general math topics that would benefit anyone in the physical sciences, ie: not just diff. geo which is only immediately relevant to relativists and so on. My limited research background/experience is on the numerical/computational simulation side of things.

 

[ModusPwnd]

I would study for the quals.

 

[cristo]

Possibly all three, as I intend to experiment in my first and second years as much as I can.

You can't specialise in everything (or else it wouldn't be called specialising...)

 

[Lavabug]

You can't specialise in everything (or else it wouldn't be called specialising...)

I am not planning this. PhD programs in the US are not a commitment to a specific project once you get there, as it is in the EU, the first 2 years are taken -apart from coursework and quals- as a time for getting to know the faculty and doing small projects to pick a thesis topic by the end of the 2nd or 3rd year. I should have been clearer, I am not 'speacializing' until I pick a thesis topic..

I don't know if I would like strictly experimental or strict pen+paper theory topics because I've never done either of these before. So I am trying to keep an open mind in that regard. I've already done numerics.

But back to the topic at hand:
What subjects do you wish you had spent time self-studying before your 1st year in grad school that would have been of practical benefit to you and your work?

 

[Vanadium 50]

Work out every problem in Halliday and Resnick.

1. You will likely be TAing, and this will reduce the time you need to prepare.
2. On the qualifier, you don't want to lose any easy points.

 

[Lavabug]

Work out every problem in Halliday and Resnick.

1. You will likely be TAing, and this will reduce the time you need to prepare.
2. On the qualifier, you don't want to lose any easy points.

I had this in mind, I have a 70's edition laying around. Thanks. But do you literally mean every problem? What about spending time on more challenging, original problems like in K&K or Purcell?

 

[heatengine516]

You should check the syllabuses for the undergrad classes at the school you're going to and see what books they use before working out every problem in a book.

 

[Vanadium 50]

I had this in mind, I have a 70's edition laying around. Thanks. But do you literally mean every problem?

Yes. If they are too simple then it won't take much time to finish them.

What about spending time on more challenging, original problems like in K&K or Purcell?

If you have more time, why not?

 

[Choppy]

If you've been accepted, it's not too early to contact potential supervisors and start some background reading that will help you determine a project.

Or, I would also advise just reading up on what interests you. The time to do this tends to fade quickly once you get into graduate school.

 

[Lavabug]

Yes. If they are too simple then it won't take much time to finish them.
If you have more time, why not?

Alrighty.

You should check the syllabuses for the undergrad classes at the school you're going to and see what books they use before working out every problem in a book.

I see your point, but chances are any problem in a 1st year physics course exists in some shape form in Halliday-Resnick. Plus it's the only intro physics book I have in my possession (got it for pgre review for a bit of pocket change), so I might as well use it.

If you've been accepted, it's not too early to contact potential supervisors and start some background reading that will help you determine a project.

During my visit to my top choice school, I was told this is a possibility, even going there early in the summer with a 'mini RA' before coursework and TA assignments start, to get myself settled and getting a taste of their research program. I will be contacting them about this as soon as I give them my almost certain 'yes', as I still have to visit one last school.

 

[Ben Niehoff]

I think Halliday & Resnick would be a complete waste of your time.

You haven't really mentioned what kind of research you want to do, specifically, so it's difficult to give you decent advice. But learning to program is good...use a compiled language like C. Or if you think you'll be using Matlab or Mathematica (and have access to them), mess around with those and learn how they work. Messing around with electronics is also a great idea, if you think that's the direction you'd like your research to go in.

Aside from that, read Goldstein, Jackson, or Sakurai, or pick up a book on stat mech (Chandler is decent).

 

[Lavabug]

I think Halliday & Resnick would be a complete waste of your time.

You haven't really mentioned what kind of research you want to do, specifically, so it's difficult to give you decent advice. But learning to program is good...use a compiled language like C. Or if you think you'll be using Matlab or Mathematica (and have access to them), mess around with those and learn how they work. Messing around with electronics is also a great idea, if you think that's the direction you'd like your research to go in.

Aside from that, read Goldstein, Jackson, or Sakurai, or pick up a book on stat mech (Chandler is decent).

Why would H&R problems be a waste of time? Granted I should and probably do know how to do most of the problems in it without much or any revision, but wouldn't it help at all? I am getting a good review already just by tutoring AP physics and college students so maybe that covers a good portion of intro material, maybe I should focus on the topics I am not touching/haven't touched in a while to keep them sharp.

I had two courses as in Goldstein and Landau mechanics as an undergrad so I do feel prepared in that regard. Jackson is in my sights, as is C and/or python programming, but I'm trying to just focus on 1 or 2 things at a time and actually learn something. I have tried to learn 2 languages at once before and did not succeed at either, I found I do better when just focusing at one at a time.

I have done a bit of fortran for numerics in both linux and windows, so C sounds like an easy transition (got myself a numerical MHD package yesterday and compiled it), but I would like to learn something that makes array-based programming really fast and easy.

I found enthought python distro is very good at this (and can speed things up with cython), but I also have access to Mathematica and IDL, both of which I know only a little. Should I focus on the former since it's freely available and more common, or should I continue using what I already know and get better at it?

 

[TomServo]

Would you peeps recommend reading Schwartz to prepare for Jackson?

 

[nonequilibrium]

My personal answer to the question "what do you wish you had learned about before starting grad school": differential geometry

 

[TomServo]

My personal answer to the question "what do you wish you had learned about before starting grad school": differential geometry

Why is that? GR? Do you recommend a text?

 

[Lavabug]

Schwartz is an interesting read but I think it is of little practical value for graduate coursework in EM. Working boundary value problems like the type you find in a PDE+special functions course will probably be much more useful.

Well as far as programming goes, I made it through a couple of video seminars and tutorials from enthought on python+numpy+scipy for scientific computing and will be moving on to some of the more challenging data analysis exercises next. Since there aren't that many, could anyone recommend a page, course, or w/e with exercises that include data sets to solve fun and challenging problems in scientific computing? Something that doesn't require a very deep field-specific background (or at least anything a physics graduate could jump immediately into without consulting journal papers, as I really just want to get more skillful at cracking data problems).

 

[nonequilibrium]

Why is that? GR? Do you recommend a text?

GR is an obvious choice, but differential geometry even underlies much of QFT! The nice thing about differential geometry is that it underlies gauge theory, which on its turn underlies both QFT and GR :) And it also features a lot in more recent advances like string theory etc. I think a good knowledge of differential geometry (including basic differential geometry (manifolds, forms, etc), Riemannian geometry and bundles) will keep on paying off (in theoretical physics). Gauge theory is even penetrating condensed matter physics!

An amazing text is "Gauge fields, knots and gravity" by John Baez, but it's not really a textbook (although it will teach you so much!). For a more standard textbook, I'm not the person to ask. Simple but thorough lecture notes I like are https://www.dpmms.cam.ac.uk/~agk22/teaching.html, although I wouldn't use them as a first encounter.