Advice for Graduate Students
Your mileage may vary.
Becoming a Researcher
- Ask questions. Now is the time.
- Take care of yourself. See a doctor regularly.
- I expect you to perform your work ethically. You can
expect the same of me. For details, see
American Physical Society web pages under the headline
"Ethics and Values".
- Always come to my office with something to write
on, either a few pieces of paper or a laptop.
- Make sure you know how to look up articles at
given for example, the first author's last name and a
year of publication.
- If you would like to do research with me,
be prepared to tell me why.
- If we haven't seen each other for a few days, when
we meet again, you should either have a question to
ask or something to present (preferably both). If you
haven't made any progress, that's ok, just be prepared
to tell me what you tried (even if it didn't
- Sometimes I give you papers to read. When you come
back, be prepared to tell me something you learned.
You don't have to understand the whole paper, or be
able to derive all of the equations, but you should be
able to tell me about some detail that you learned
(even if it's not a detail that is central to the
- Part of being a graduate student is learning to
manage your time. Make sure you're not stagnant and
always making progress. Make sure you're never just
waiting on me to do something. Efficiency is
important. If you're having problems with this, let me
know and we'll get it sorted out.
- Impostor syndrome is
feels, at some point, like they're not smart
enough. (Me too.) More often than not, you are a
lot more capable than you think. In any case, I'm
always happy to give you an honest assessment of how I
believe you are doing and how you can improve.
However, I can tell you the most likely answer
already: you are smart enough and yes there are things
you can do to improve (and you probably already have a
good idea what they are), but I can't make you any
guarantees about your future.
- Go to the department colloquia, the nuclear physics
seminars, and the astrophysics seminars. If you're not
getting anything out of them, then try to read the
relevant papers authored by the presenter the day or
night before the seminar.
- Success in the future requires a solid understanding
of the scientific background of your thesis topics and
other related topics in the field. I will be able to
help you with a small part of this background
material, but it is important for you to do a fair
amount of reading of papers (and sometimes computer
code) on your own. Make sure that reading is an active
not passive learning process. For example, try to
derive some of the equations in the papers you read,
reproduce some of the plots, write some related
computer code, or download a related open-source
computer code and run it, etc.
- Plots are the scientific currency which our field
uses. You'd be surprised how far you can go with one
really great plot. Learn how to make good plots
quickly, using whatever software works best for you. I
recommend starting with python3 and matplotlib, and I
recommend against using Microsoft Excel, but in the
end I leave the final answer up to you. I have my own
for this sort of thing, but it's not well documented
- There may be times which I am on travel and I
cannot answer your email messages quickly. This means
that, in the mean time, you may need to either try a
different approach to your problem, or switch
focus to a different task. Sometimes this
means you need to try again to solve the problem
Worst Advice Grad Students Get",
you want to be a grad student?", and
To An Academic Conference? Here Are Some Tips".
Letter of recommendation?
- Let me know, preferably a month in advance, that
you would like a letter.
- Send me your CV.
- Provide for me, as soon as possible, the
- the nature of the position you are applying for
(postdoc or faculty job?),
- the name of your future boss or the head of
the search committee,
- the institution you are applying to,
- the deadline, and
- and the way in which the letter
is to be sent.
- Remind me 24-48 hours before the deadline just
Resources for UTK graduate students
My Advice on Talks
First, understand that all of this advice comes from a
particular perspective which may not be the same as
Everyone says this, but it's worth saying again:
understand your audience.
I'm less picky about looking at the audience versus
your slices than I used to be. As you progress in your
career, the most important thing here is to communicate
that you're excited about the material that you're
Timeliness is important for for graduate students
because many of your first talks will be at larger
meetings where session chairs are likely to be rather
strict. I am of the belief that, if you haven't
convinced me of your point in 20 minutes, an extra 2
minutes isn't really going to help your cause. It takes
a certain skill to watch the clock and extemporaenously
reframe your talk as necessary. In any case, you
should always make sure there is some slide in your talk
at which you plan to look at the clock and re-evaluate
your talk. Then, you can present the following slides
faster or slower accordingly. Also, I recommend always
having a few extra slides at the end in case you're
faster than you expect. Practicing your talk also helps.
When I'm practicing, sometimes I like to actually type
out what I intend to say. Some other method
may work better for you.
Be careful about apologizing too much. Act as if
you're busy and important and you carefully chose to
spend exactly as much time and energy on the
presentation as the situation warranted.
Be intentional about how you use mathematics, and
don't apologize for it. Also, don't skip over
mathematics claiming that you did so because it was
too difficult. Many people (even those outside
academica) do a lot of mathematics, and it's always
disappointing to find people giving more weight to the
common myth that math is particularly hard, or just
too hard for some kinds people, or just too hard for
them personally. Math requires practice, just like
reading. That's it. Also, when you say you're skipping
something because the mathematics is complicated, I
typically interpret that as meaning that you didn't
feel like taking the time to present the mathematics
in a comprehensible manner.
In my experience, the nuclear physicists and
astronomers tend to be completely different about
mathematics. Astronomers feel no shame in writing an
entire article without a single equation, yet the
nuclear physicists don't feel like you've done any
real work unless there's a very complicated equation
up there some where. [I have received a lot of
conflicting advice about this in the past.] Maybe my
assessment is completely wrong, but either way you
should make sure to understand what is
important to your audience and act accordingly.
- Steiner's first rule of scientific computing:
If at all possible, avoid asking a computer to compute
a quantity you don't already understand.
- Steiner's second rule of scientific
Read the first rule again.
- Comment your code. Otherwise you'll regret
it when you try to come back to it 5 years later.
If you don't think you'll come back to it 5 years
later, then you're probably writing code to
solve the wrong problem.
- Keep a backup of your files. You never know
what will happen. Let me know if you need
storage space somewhere.
- You may spend 10% of your time writing
code and 90% of your time debugging it. Obviously,
planning ahead is a good idea, but even then,
you will spend a lot of time debugging.
- For debugging, divide and conquer almost always
works. Break the problem up into small pieces and test
each individual piece and then progressively put them
together and test each time you combine the pieces.
It's frustrating to me if you come to my office with
complaints about debugging when you haven't tried
divide and conquer first. Another strategy is to take
the function, class, or program which you are
attempting to debug, and write it again from scratch
(either in the same language or in a different one).
It sounds onerous, but it's often faster (especially
if you're spending 90% of your time debugging code and
10% of your time writing it)!
- Make a log somewhere of how
you install packages to your computer so that you can
fix it when it goes wrong or reproduce it in the
future. This is particularly important on Mac laptops,
as there is potential confusion between native OSX
packages, Fink, MacPorts, homebrew, pip, etc. Not all
of these tools play nice with each other.
Back to Andrew W. Steiner at
the University of Tennessee.