Choosing
research topics and advisors
Literature Searches
To do research in a subject, you must become an
expert in it. Therefore, research projects must always be
accompanied
by very thorough surveys of the scientific literature. If you
find an
unexplored avenue, unjustified assumption, or unexplained phenomenon in
past work, this can be the basis for a new project. Once you've
chosen
your project, map out the sequence of concepts and arguments about it
by
tracing the important citations backward to their source. At the
same
time, you should know of any recent developments that might impact
(e.g., scoop!) your research. Here are some primary tools:
Your local library – the right
place to start for finding books, journals, and electronic resources
otherwise hard to find.
Here is a large
repository of on-line texts, which may cover your topic of
interest.
The on-line Handbook of Space Astronomy &
Astrophysics by Zombeck has a broad discussion of useful
facts and formulae.
CDS's AstroWeb – a large, comprehensive collection of astronomy-related internet sites and resources
NASA ADS –
get to know it. It can find and often display most astronomy articles,
display lists of their references and citations and 'related' and
'also-read' articles to the one you've chosen. It also generates
automatic LaTeX and BibTeX citations to the articles (using a
standardized citation format); see the bottom of any abstract page.
These, in turn, can be used to construct the bibliographies of your
scientific reports and articles. It is often updated with new,
useful
features. Recently, it has gained the capacity to generate email updates
on topics of your choice, and to generate, organize, annotate, and
share "private
libraries" of selected articles.
The arXiv
preprint server - A forum for the circulation of papers before
publication. You can sign
up
for daily emails of all astrophysics (astro-ph/) abstracts,
for instance, but be ready to spend half an hour each day reading
through them. Most astronomy departments have a daily or
weekly
meeting to discuss the most interesting of these. You may also submit refereed
publications or conference proceedings. Some people stick to the
rule
that articles should only be submitted after they have been submitted
to journals or after they have been through one referee report.
There
is no injunction against this, but it's probably not a bad idea,
especially for a first publication. Note: some, but not all,
arXiv
preprints are searchable via ADS if you click "arXiv" in the query
form.
In addition, here are several excellent pages compiled by the Astronomy &
Astrophysics Library at the University of Toronto:
Computer Literacy and Useful Scientific Software
Start with the Astrobites
guide to astronomical software and the AstroBetter Wiki
(computer section) for quick introductions. Some of the languages
discussed below (Python, HTML, etc) can be learned rapidly from sites
like CodeAcademy; for a
low-level introduction, Khan
Academy's programming pages are not bad at all.
Linux and Unix – Preferred operating system for scientific work. Note, Mac OS X is Unix-based (open the Terminal).
A collection of links to online books on related topics (courtesy Anna Lubchnko)
My “Blitzkrieg Introduction to Unix/Linux
Commands” written circa 1990.
LaTeX – a typesetting language for writing scientific documents, including means to typeset equations and formulae and to include figures, tables, etc. This is far faster and more flexible than Word, believe me. In fact, it's practically the only way to produce scientific papers in astronomy.
A directory of my LaTeX style files, including a 'defs.tex' file that can be included (illustrating command definitions) and a 'mybib.bib' file (illustrating the construction of a BibTeX bibliography database), and useful style files like “emulateapj.sty” and “mnras.sty”. When included, these make your document conform to the appearance of major journals (ApJ, MNRAS, etc).
The ADS Abstract Service provides LaTeX references -- both in \bibitem format and in BibTeX form.
LaTeX Symbols
(useful for constructing fomulae); see also this very
long list [PDF] of LaTeX symbols.
The AASTeX
Macros for Manuscript Preparation and its latest update, AASTeX
6.0. Note, AASTeX now includes all of the useful citation commands
– like \citep[e.g.,][and refs therein]{label} to make (e.g.,
<authors> <date>, and refs therein). A sample.tex file is
available from AASTeX.
The LATEX2HTML Translator (converting latex documents to websites)
Emacs – a sophisticated text editor with many commands and “modes” that are extremely useful for editing specific types of documents, viz. C or FORTRAN progams, LaTeX documents, etc.
Latex-mode in Emacs (invoked by 'Esc-x latex-mode' if not automatic) has many useful properties: if your 'font-lock' settings are correct, you'll see Latex commands colored according to context. (If not, issue the command 'Esc-x font-lock-mode'.) In this mode, ctrl-c ctrl-f opens a shell that compiles your latex document automatically.
Reftex mode (invoked by 'Esc-x reftex-mode')– an extremely useful mode for editing LaTeX documents, in which Emacs can bring up menus of references, equations, tables, and figures and inserts references to these tags once you've chosen from the list. This saves immense amounts of time. See Reftex in a Nutshell from the Reftex link. Examples: 'ctrl-c [' = find reference to cite (using a string match); 'ctrl-c )' = find reference to eqn, table, figure, section, etc.
A few other very useful Emacs minor modes: auto-fill (word wrap), flyspell (spellcheck), and show-paren (highlights matching or mismatched parens next to cursor). Invoke with escape-x and then 'auto-fill-mode' etc.
My .emacs and .mytcshrc files, which define, for instance, where Emacs should look for a BibTeX database while I'm typing a LaTeX document. (Note TEXINPUTS, BSTINPUTS, and BIBINPUTS environment variables set in in .mytschrc.) These should be edited to point to the correct directory.
vi
– a text editor with a completely
different set of rules from Emacs, and a fierce following. Note, there
is a vi-mode in Emacs, so you can get the best of both worlds.
Matlab
– an interactive language for
data
manipulation and visualization. Unlike Python it is neither free
nor open source; on the other hand, it has the advantage that functions
follow a standard protocol andand documentation is easy to access.
Matlab's syntax for doing operations on
arrays and matrices has some commonalities with Fortran 90. It is
especially useful if augmented with functions and routines
(<function>.m) that perform common tasks (differentiation,
trapezoidal integration, zero-crossings) on two-column arrays that
define “functions” y(x).
Matlab has the capacity to produce publication-quality figures,
which are easily generated and whose properties can be edited from the
command line or via GUI tools. It major drawback is that its TeX
interpreter is limited to a range of symbols that does not
include, for instance, the Sun symbol (\odot).
Inside Matlab, type “help” to see a list of topics to learn about, or choose topics from the Help menu.
A directory of useful Matlab routines. Copy these to a new directory and then issue 'addpath <directory>' at the Matlab command prompt. This allows you to query the routines using 'help <routine>' if <routine>.m is one of the filenames. Especially useful is the 'units.m' script that, when run, defines (capitalized) physical and astronomical constants in cgs units (see 'units' and 'unitsB' in the Mathematica section). Also very useful are the integrate.m, integral.m, and mydiff.m (calculus) functions.
Octave – a free package (included in recent RedHat Linux releases) that is very similar to Matlab (to the point of using the .m filename extension) but which (for now) lacks a number of useful features. It employs GnuPlot for its graphics output.
IDL –
similar in style to Matlab, but with
a
different (I say worse) syntax; more commonly used among astronomers.
It
therefore has available a wider range of pre-existing routines
specifically for astronomical use.
IDL Astronomy Libraries and Resources from NASA
IRAF – a package specifically for astronomical data reduction.
An IRAF tutorial
Mathematica – an environment for the symbolic (and numerical) manipulation and solution of equations. For instance, Mathematica shares with Matlab and IDL the ability to solve ordinary differential equations given sufficient boundary conditions (although the degree of user control is generally lower with Mathematica); however, only Mathematica (or Maple, or a few other symbolic math packages) can symbolically solve a differential or algebraic equation in closed form.
Mathematica files 'units' and 'unitsB' that define physical constants in cgs units, or with explicit units, respectively. Include with '<<units' at the command line. Also, Mathematica file 'xOne' that defines a function which solves equations of a common power-law form assuming its first argument equals unity, e.g, 'xOne[a b^c, b]' yields 'a^(-1/c)' as the solution for b.
Maple
– a mathematics package
similar to Mathematica. More Canadian.
Numerical Methods
SuperMongo – a popular graphics package that can read data files, perform simple manipulations on them, and graph their output. Generally it is not as powerful an analysis package as Matplotlib, Matlab or IDL; nevertheless, its graphical output can be very clean.
GnuPlot – a less powerful, but
perhaps more straightforward, graphics package
VNC –
Not a software package. Instead, VNC is an efficient means to turn a
window on your machine into a 'distant monitor' on an external
machine. VNC uses compression to speed the transfer of mouse movements
and keystrokes from your computer to the distant one. VNC comes
standard with recent RedHat Linuxes, and versions are downloadable for Windows and for Mac OS X. You can get by with just the viewer on your
own machine. You run 'vncserver' on the machine you want to log into,
and 'vncviewer' on your local machine.
The first time you use VNC, it will prompt you for a password. Do
not forget this, as it remains your VNC password for all future
sessions. I suspect that a file must be deleted to establish a
new one.
You may have to set up a
'ssh port tunnel' in order to get through the firewall on the distant
machine; under Linux this can be accomplished with 'vncviewer -via
user@<remotehost> localhost:<port>' where
<remotehost> is the name of the distant computer and
'<port>' = '4' corresponds to actual port 5904 (offset of
5900).
To set up port tunneling under Windows, use an ssh program like the
free one from ssh.com to log into the remote machine; define a profile
for that login. Then select
"edit->settings->Connections->tunneling" and add a new
connection of type TCP with the destination port both equal to 5904 (or
5900 + desktop number assigned by the server). The listen port
can be 5901, for instance. Then run VNC viewer; you should be
prompted for your VNC password and the host. For the host, type
"localhost:1" if you have mapped the distant port onto 5901.
SSH tunneling is strongly recommended and often mandatory, as VNC does
not encrypt the
information it transfers.
Science is gaining knowledge by applying logic to the world around us. Every scientific presentation must establish that some important question is at stake, and how logic is used to understand it.
As
communication tools, presentations leave a lasting effect on the
audience -- either that you really know what you're up to and how to
convey your results (and enthusiasm) to them, or that you
don't. For this reason it's a good idea to think
through the process of constructing and presenting a talk. My
tips:
0. Choose an interesting topic and make sure you know what you're talking about.
You should know the subject, the background, and what the important
questions are, as well as the details relevant to your own
presentation. If this is missing, the following stylistic points
won't amount to much.
1. Practice your talk at least
twice, preferably in the presence of someone who can give substantive
critical feedback.
2. Start by giving the audience a sense of who you are, why you're
interested in your subject, and what makes the topic (or your approach
to it) distinctive. That is, establish some rapport with the audience and give them a sense of what's at stake so they'll stay
awake. Along these lines,
3. Explain up front what central
question will be addressed in the talk and why this question is
interesting and important. Never launch into a complicated
argument
until you've thoroughly established why the audience should care about
it.
4. Make sure each element of the talk has a point that builds toward the goal of
the entire talk, and take time to remind the audience of these points
and goals at regular intervals.
5. Make sure you speak clearly, include pauses at important places,
maintain eye contact with your audience, and stop to ask for questions. Treat questions
with
respect and interest, even if they seem derogatory. If possible,
incorporate aspects of the questions into the talk. Do not be afraid to
say "I don't know." Difficult questions should be postponed for
discussion afterward.
6. Strive to be clear and concise.
Use your practice runs to streamline the discussion and to
eliminate
extraneous comments. Be careful not to repeat yourself. Avoid
jargon
terms and excessive detail on overheads; make sure what is shown is
legible and easily comprehensible, and that all graphs and figures (and
their axis labels) and equations get explained in words. Be
careful to
check the units of any equation, and make sure that every variable is
defined as soon as it is introduced. Many tips for clear
expression are
offered in the writing advice section.
7. Interest the audience.
Do
it through analogies, posing questions and puzzles, (brief) relevant anecdotes,
etc...
not jokes, unless you're naturally funny. Act naturally at all
times. Make sure that the audience feels they have learned
something,
even if it is only how to explain something clearly that they already
knew. This takes a lot of work, so start preparing your talk
early.
8. Express enthusiasm for your
subject. The inflection of your speech and your gestures and body
language should convey this. Speaking of body language: do not
pace or
fidget; at the same time, don't stand motionless or hide work on the
blackboard with your body; don't read your overheads.
9. Some people think opening outlines are essential; others
eschew
them. Be careful to use an
outline only
as a teaser for what's to come; don't fall into the trap of
discussing material that will be properly introduced later. I
prefer not to
include "Conclusions", "Summary," and "Future Directions" in the
outline: these elements should exist in any talk and it wastes time to
say them. Stick to items that distinguish your work.
10. Finish your talk early
enough to practice it several times.
Did I say this already? This is more impotant than trying to
include last-minute results,
although it may not feel that way at the time. Pay close attention to
the critique of your test audience, and respond to it! Be
willing to cut or reorder material, or start over entirely, if you need
to.
Peter Kenny, A Handbook of Public Speaking for Scientists and Engineers, New York: American Institute of Physics, 1982.
Jeff Radel's Effective Presentations website
Mike Dahlin's How to Give a Bad Talk and How to Give a Good Talk
Cole Miller's Ten Commandments of Viewgraph Presentations and additional hints on giving talks
Talk guidelines by Marie DesJardins at Indiana
Obtaining strong letters of recommendation
Professors are impressed by curiosity, creativity, dedication
and tenacity, interest and enthusiasm, and initiative.
(In short, “spark”.) Very beneficial also is the ability to carry
projects through to their conclusion without cutting corners, and the
ability and willingness to discuss your work and ask questions of
others. A good letter of recommendation contains specific examples
of these traits (and other positive qualities, including how the
student functions within a group or in collaboration). For this
reason, it is vital to create specific instances to refer to! And, it
does not hurt at all to keep a log of your research activities
for later reference (by you or your prof.).
Many good research practices will manifest themselves in good
letters. So, establish a schedule and a set of expectations with
the
professor early in the project, and then provide them with regular
progress reports. Professors will vastly prefer honest
assessments of
what can be done by when, to reasons for slipped deadlines after the
fact.
Visiting
a Research Group or Prospective Institution
Whether visiting for an undergrad project or for a faculty job,
remember that there is only one chance to make a good first impression.
Your hosts are taking time out of their schedule to meet with
you, just
as you are for them. Even if you know that you will not join
their
group, there is ample opportunity for future interactions -- so make
your best effort. This includes listening attentively to their
research
activities, asking challenging questions, and maintaining a friendly
and
outgoing demeanor. Do not feign knowledge or skills you don't
have. If
possible, prepare in advance by looking up their projects, papers,
websites, and collaborations.
Maintaining a Teaching Portfolio/Dossier
Although this site is primarily about research, teaching is an equal
part of the job description for graduate and professorial (and
sometimes postdoc) positions. Teaching well is also a noble end
of its
own. When applying for jobs that include a teaching component, it
is
essential to provide some indication of your past experience.
This is
best accomplished with a teaching portfolio in which you present
curricula, syllabi, assignments, student testimonials and evaluations,
etc., that may be of use in this regard. Writing a portfolio also
stimulates self-reflection on how to teach well!
A strong and well-maintained CV (curriculum vitae, or vita) is
always useful. Here is
the LaTeX file for my CV (as of June 16) and here
is the PDF it
generates.
There is a wealth of advice online at the University of Toronto Writing Centre's site on Application Letters and Résumés (within Writing Advice). Purdue University also has a comprehensive set of advice files.
It is never too early to think
about attending meetings in your field of research; there are often
special financial aid packages from societies like CASCA, AAS, or your
institution,
and the meeting organizers often have extra money to aid young
researchers. (I paid
my own way to a very good meeting on starburst galaxies in the south
of France when I was in the third year of graduate school.) A list
of meetings is available at http://cadcwww.dao.nrc.ca/meetings/meetings.html