CLASS GOALS AND ADVICE ON RESEARCH PROJECTS AND PAPER PRESENTATIONS
    -------------------------------------------------------------------


I have designed this class in such a way to give you some fundamental
lessons in how to do research in grad school.  Both lectures and paper
selection introduce some concepts without exhaustively exploring the
topics. Additional material will be provided, so you can explore the
introduced topics yourself.

In particular, the research papers were selected such that to give you
a sense of how beneficial it is to have an interdisciplinary approach
to research. You will see ideas and patterns that appeared originally
in one domain (e.g., parallel hardware architecture or databases)
being taken up and applied in a different domain (e.g., OS or
distributed systems) with a lot of success.  In the same vein, most
assignment problems and especially projects will be incompletely
specified.  This is to prepare you, in a minor way, for undertaking a
research topic, where almost everything is unspecified. Try to become
independent and take charge of your project. I welcome discussions and
questions about your projects, assignments and paper selection. On the
other hand, when talking to you, I will try to establish whether you
have tried to answer your own questions and whether you have analysed
the problems you are asking an answer to.

Finally, I hope that being in and working for this class will teach you 
some very essential things that cannot be taught directly, but will
have a profound impact on your future success. This is: how to work
with your colleagues, your professors, your advisor, how to behave in
a professional manner for your larger audience that will be there for
you once you start doing research, publishing, going to conferences,
delivering conference presentations, starting to have your own network of 
professional contacts.

First and foremost, impeccable work ethics is one of the most
important aspects of professional behavior in grad school. There is no
surer way to destroy your academic career without possibility of
redemption than if you cheat during a grad course or show consistent
lack of seriosity in other ways (e.g., exhibit repeated disruptive
behavior). This is not undergraduate anymore where some courses and
topics may have been forced down your throat.  Presumably you are in
grad school because you want to learn something and you want to do
research. This applies when you come to class as well.  Your presence
in class is not mandatory. I am very open and flexible to suggestions
from you regarding the class format, the paper selection and so
forth. In particular, for the project, I will give you a lot of
latitude so you can maximize the benefit for your own research topic;
for the papers, I will give you some extra selections, listed under
"Other" to choose from.  You are welcome to pick one of those for
paper discussions instead of the main papers suggested if you like.
The grade you get in this class, in itself, should be less important
to you than what you can learn from the class. And this is, in a large
part, up to you.  

Remember this: Your success is not defined necessarily quantitatively
anymore.  As an example, it does not matter to me whether you get a
good speedup in performance for your application after you have
parallelized it in your class project or assignment.  What matters is
whether you have tried to analyze your experiments and you have looked
for sources of overhead and you have tried to eliminate them.  It does
not matter to me whether you are a naturally born-speaker or not and
whether you make mistakes or have questions yourself during your
presentation because you could not understand the paper very well. It
is natural for you to still have questions after reading a paper or
going to a conference and listening to a talk, especially for papers
and talks that are not in your direct area. Being open to learning
from other areas that you are not an expert in, or even know nothing
about is the strength of doing interdisciplinary research. You will
understand the main idea, or the general approach from a paper in a
different area (details not necessary) and you will try to see how it
fits your own research area, if at all.  Just remember to look for the
key-points of each paper and not get bogged down by its details.  It
is natural to be slightly uncomfortable when reading a paper in a
completely new area. If you are very comfortable in your work and you
know all the answers, it means you are simply not doing research.
Also, the time is past when people that simply work in a very narrow
research niche of their own and never look at the big picture can
continue to be successful forever. Whether we like it or not, the
computer field is changing very rapidly. The architects need to know
about the software running on their architecture, the software
developpers need to know about the applications and architectures that
they are developping software for, the application developpers need to
know something about how to program the application for performance
using newer software packages and so on.  Expect to potentially have
to broaden your research topic, integrate orthogonal research
prototypes, or go from one research topic to another.  There will be
at least some transitional discomfort in the process.  This class
tries to prepare you for that.

In spite of this and for the purposes of your paper presentation for
the class, what matters is whether you have prepared your presentation
to the best of your abilities, that you have thoroughly read and tried
to understand the paper, that you have understood the key idea of the
paper and you can express it. Details cannot be covered usually in a
short presentation anyway and you are speaking to an audience with
very varied backgrounds.  It is also a good idea to read suggested
extra readings, and follow some pointers to related work from the
paper itself.  It is important to try to have clean and uncluttered
slides with key-points on them and not a lot of text without a point
to it.  It is important to work towards delivering a logical flow in
your talk.

The bottom line is: it should not matter to you what letter grade you
get in my class. What should matter to you and what ultimately also
matters to your advisor, is whether or not you are on your way to
becoming an independent thinker, a thorough and enthusiastic
researcher and a true professional.

Good luck with the class and I hope you'll enjoy it !

Your instructor (Cristiana)