Dr. P J Narayanan Advice to research students

       "But, I have been a student all my life ..."
       or
    Informal Advice to Research Students

       P J Narayanan, IIIT Hyderabad
       April 19,  2005

Congratulations on becoming a research student and choosing a path of
enquiry. You were a serious and studious undergraduate or bachelors
student just yesterday. Since you continue to be a student -- and in
some cases in the same institution -- life can't be all that different,
can it? Turns out it is!

The emphasis in undergraduate education is on comprehending the concepts
and developing the necessary skills in a broad area.  A course had
fairly well-defined syllabus and contents. Examinations and assignments
contained questions to which answers existed and were known to the
professor. There was only a little -- if any -- of tricky components to
the courses. The emphasis was on learning to a certain level; effort
usually guaranteed learning. Those who had the opportunity to go through
the research-focussed honours programme during the bachelors may have
had glimpses of deviations from the above model: glimpses into the
research-land!

The world changes when you become a research student. The problems are
neither well-defined nor necessarily solvable. The professor may know
only a little more than you about the specific problem. S/he often knows
less. The emphasis is not on achieving the defined levels of learning.
The goals are to understand the intricacies of the topic through an
enquiry into the various aspects of the given problem. The method of
enquiry is as important as its results.

Thus, finding a solution to the problem is not the end of matters
anymore. Finding better solutions than existing ones assumes critical
importance. Sometimes, establishing that a satisfactory solution does
NOT exist can be a more significant result. The question is no longer
one of finding a solution to the problem, but one of understanding the
associated issues, techniques, and solutions to master the problem
landscape. Your work does not stop at finding a solution to your
problem. Rather, it begins then. The enquiry is into the structure of
the problem (and the solution) and constraints thereof. That often
begins after one has a solution already.

So, what does it mean in terms of what one does every day? While it is
impossible to give a definitive manual on how to do research, there are
certain points worth reminding yourself continuously.

 - Have goals: Have (vague) goals for the next year, (reasonable) goals
 for the next month, and (very lucid) goals for the next day!
 This is easier said than done. Many of us are in an exact
 opposite situation. We have lucid goals about next year ("get
 the best paper award from IEEE") but only hazy ideas on what
 needs to be done today to reach there.

 Plan your time properly based on what you need to do. Do work 24
 hours when a paper deadline is around, but don't wait for them
 to psyche you into action. Your own goals should drive you
 steadily under peacetime conditions. It will not suffice to work
 VERY hard a week or a month before the thesis submission
 deadline (unlike what you did with exams or projects in your
 undergraduate days).

 - Immerse in your work: Tentativeness or inattentiveness is highly
 lethal to a successful research career. When you are into a
 problem, live it! Breathe it! Be disturbed by suboptimal aspects
 of it.

 Many students think of their research problem in the few minutes
 they may get between emails, chat sessions, tea-breaks, and
 general other "important" activities. This will lead you
 nowhere.  Five or six hours of concentrated work per day is more
 than sufficient if you are living and breathing research.
 Without that, 24 hours per day "spent" on research are unlikely
 to yield anything.

 This is where many students falter. They do spend enormous
 amounts of time in the lab, with a paper or program in front,
 and mistake that to be time spent on the research. No, sir.
 Quality time without distractions is absolutely necessary for
 research.  The goal is not merely to comprehend something or
 explain it in an exam any more. Deep understanding is the goald
 during research.

 - Look around: Pay attention to all aspects of the path you take.  This
 means, read and analyze every aspect of papers you read.  It may
 be sufficient to have a vague notion of a method used (e.g.,
 merge-sort algorithm or least-squares technique) to understand
 the main concept of a work. And that is the right approach for
 the first reading. You should come back and understand all
 nuances of the algorithm or the technique subsequently. To
 marvel at its mere beauty, perhaps. To realize it is not the
 right fit for *your* problem, often. No stone on the way should
 be left unturned by you.

 - Question everything: Marvel at everything you hear on the outside.
 Doubt everything you hear deep down! Nothing is sacred and there
 is nothing like a stupid question. Question each step. Argue
 against it. Find better options. Once you have found better
 options on your problem, doubt yourself deeply! Recurse in this
 manner till the right solution is found.  That is the way of
 research. Remember, your goal is not to solve a specific problem
 but to master the problem landscape.

 - Discuss: Share your ideas, doubts, anguishes, etc., with others
 including your advisor, other faculty members, folks from your
 area/lab, folks from other areas/labs, your dog, whoever!
 Others may give you good advice or may reflect your own ideas
 back to you, only for you to realize their worth/folly. The mere
 act of explaining your problem can (and often does) result in
 you finding the solution yourself. Explaining the problem to
 yourself can be tried but explaining it to others is likely to
 be more effective.

 Again, the difference with UG education is significant. A course
 project was given to you with the idea of you submitting it when
 done. No advisor gives you a research problem expecting you to
 solve it completely before seeing him/her again. In fact, trying
 this is a sure way to get onto the advisor's bad books.
 Advisors expect you to understand different aspects of the
 problem by reading, thinking, discussing about it; s/he expects
 you to present your understanding to him/her periodically. The
 advisor may even learn a thing or two in the process and that is
 quite frankly the reason to have research students.

 - Write: Put your ideas, theorems, algorithms, thoughts etc., down on
 paper.  They may only be a few vague points and not a
 full-fledged paper with introduction, background, and
 references. Writing the ideas systematically is the best way to
 relate them to other ideas, to expose the holes in them, and to
 recognize their true potential. Share what you write with others
 near and far.

 - Record and Summarize: Keep copies of the papers you read. Collate
 them based on the topic or techniques and keep them bound with a
 table of contents! Mothing is more frustrating than the feeling
 that you read "something like this somewhere" recently.  In
 fact, summarize the papers you read in your own words into a
 notebook. (Choose a fat one for it!) Summarize in words,
 summarize the formulae, the techniques, and the references!
 Don't be afraid to let the summary be longer than the paper
 itself. You understand an idea best when you write it down
 yourself in *your own* words.

 It is a great idea to keep notes on your own research thoughts
 and how they progressed. Detailed notes and not just keywords.
 You may revisit them one day and will find your own notes very
 handy. The IPR-evangelists will also tell you that the lab notes
 and discussion notes can be used legally to establish the
 ownership of ideas, should they be questioned.

 - Have diversions: Go away from your work often enough. This does not
 contradict the point on immersion. Have serious hobbies and
 interests. Learn to play a musical instrument. Sing, dance. Tire
 yourself hiking in the hills, Swim to exhaustion. Read
 philosophy, literature, biography, humour. Take your mind
 completely off your work for sometime. A refreshed and relaxed
 mind can produce truly innovative ideas.

 No time is as suitable to develop other interests and hobbies as
 the time spent on research, especially on a PhD.  Repeating an
 year in bachelors or lower classes attract a lot of
 attention. ("So you failed?") Taking an year extra for PhD will
 hardly be noticed.  (Nobody really knows what one does while one
 does a PhD anyway.  Those who do can be convinced that your
 thesis got a lot stronger in that one year!) Doesn't this golden
 opportunity invite exploiting?

 - Have faith: If all the talk about research sounds like magic, nothing
 can be further from the truth. Research is, in the end and for
 most people, about discipline, aptitude, and hard work. It is
 most wrong to think that a bright idea will strike you one day
 and then your thesis will be done. DON"T sit under a apple/mango
 trees with ripe, ready-to-fall fruits and expect to rewrite the
 theory of gravity! You will only have an injured head!

 Research thesis is not about a bulb lighting up in your head one
 fine day.  It is about building the necessary background
 painstakingly, identifying a problem, and working at it with
 devotion.  It is not about being super "brainy" or super
 "genious".  In fact, "ordinary" brains and systematic work can
 produce much better theses than "super" brains and low effort.


These are some of my suggestions for a bright research career. Think of
your research as a deep enquiry into an area. Question continuously.
Hard work will always be rewarded. Smart work will be rewarded
immediately.

What should one do from today, based on the above -- what can be
described as -- principles? Some of it could be:

 * Have a time-table. Fix the time at which you reach the lab, when you
 to take breaks, when to have lunch, play, eat, sleep, etc. This
 really helps. Completely "flexible" timing is a great recipe for
 disaster. Set a time table and start following it *today*. Don't
 put it off for tomorrow!

 * Manage your time well. Have priorities for different activities and
 allocate your time on its basis. This idea is exceedingly
 simple, but one that is perhaps hardest to implement.  You won't
 get anywhere in life without learning the act of time
 management, prioritization, and learning to say no to things
 that you spent time on till yesterday, you still enjoy, but is
 placed low on your priorities today.

 * Setup time to speak to others. Make it a habit to discuss your
 difficulties with others and help out theirs! Setup Special
 Interest Groups in narrow or broad topics and thrash things out
 among yourselves. You will be amazed how much you can teach
 others and others can teach you.

 * Never let the ball stay in your court for long. Work hard on the
 problem you have and get back to the advisor or colleagues or
 whoever. Continue to think about it and don't remove it from
 your thinking ever, but put pressure on others to solve your
 problem. Let the world provide solution to your problem; you
 only have to improve on it slightly at the end.

You can easily see that these points are not specifically addressed
towards research. They are golden habits that will take you ahead,
whatever it is that you do.

Research is highly rewarding, if you get through all of the above!  You
have the opportunity to choose your own problems and work your heart out
on them! You will have the thrill of creating: Be it theorems, machines,
products, solutions! You may experience the thrill of penury too
:-). But no longer. Few other jobs will leave you smiling in spite of
working 20 hours each day. This is a great time to be engaged in
IT-related research in India.

I wish you all joyous and productive research careers!