nags
Research Fellow
CSE Dept
IIT Bombay
email: nags@it.iitb.ac.in, nags@cse.iitb.ac.in
About me:
I put myself as hungry for deep knowledge, thirsty for ultimate happiness, passionate for different thoughts, and curious for various objects.
My broad Re-search interests are
Mathematics and its connection to visual art and music
Theorem Proving and Puzzle Solving.
Application of Category Theory and Domain Theory to CS in general and to model Software from the Algebraic point of view.
Scalability amd Speedability ussues in Massively Parallel Database Architectures
Granular Computing and the emergence of 3rd generation search engines in p2p contexts
Control Theory and System Dynamics to model non-linear system behaviour.
Quantum Computing and its realizability. How much of this mechanics is actually a reality?
And finally to use some of these above things (not all) to solve real life problems. I am open, but presently not working in this.
*The above things may look vast, but they are limited by the amount of knowledge avaliable in the universe
My interpretation of Re-search: Re-search means writing an altogether different history book, understanding the beauty of poems and its different interpretations, and argumenting like advocates for justice in the court of spiritual knowledge
High School
I always thought I could do different than others thinking differently. This I realized when I was in class 7th, where I proved a geometry problem (10+2 class problem) wrong. This problem was exclusively given to the topper (considering me as an average pupil) of our class, from whom I stole on the same day. I solved overnight (sat till 0600 hrs morning) only to prove, but later found it was wrong. I also gave an alternative problem definition along with a different proof. I had a great applause from my own teacher which he never expected. Those days I never understood the word "research" in that context of problem solving.
My quest for doing something different grew by posing more mathematical problems and giving solutions to my friends until I discovered the first IBM PC with 8088 processor, having 640K RAM and 20 MB hard disk storage space. I was astonished by its incredible power of displaying stored data from packages like WordStar and Lotus, much more than the high-end calculators could do. I thought it had the power to think and store data more than human beings. So, the only way to know that creature is to dissect it as we would do for the biology class. Opening it and dissecting would not be useful if something had to fail and never work after that. The other alternative was to know some internals before opening or doing anything with it. I frequented the library then, to search for books which would talk about it. I found some basic books and also others which talked about digital IC's and architectures. I started figuring out what was written (though I did not understand everything) and slowly learnt the required details until I could say there is something which I know. Later I found a book on Assembly Language Programming(ALP) and some connections from the previously read books made me think it was the right book to go for. I learnt the tricks of writing some programs on the MASM which I got from some indirect friend's contact. Though I started with errors after execution, as most others have encountered in their programs, later I could assemble and link my programs to generate an executable without any error. My next goal was to know how data was stored on the hard disk. I explored more further till a stage I could write directly on it, playing with the FAT structure manipulating cluster information, writing the file data to the appropriate cluster, only to view the created file using the dir command. Later, I wrote different small tools to access different peripherals all in ALP. A small breakthrough (those days) happened when I directly wrote a .com and .exe file structure using executable mnemonic codes on the hard disk. This was indeed a big success to me. At later stages, I could create graphics on that green and black monitor using ALP and it had become my favorite language. I was just 13 then.
Undergrads
During engineering, I kept of switching different languages to get a feel of each of those and started writing software by taking projects from companies, later realizing it was really boring. So, I decided something different as usual. In the 3rd semester end of my engineering, I wrote a paper on "infinite primitive concept for knowledge representation and learning" which unfortunately never got published. Anyway, for me that paper did not look great (though the idea was different). It was during that time I shifted to Databases from AI and finished final semester of my engineering scoring just first class. The number of paths that I had chosen had a bad impact on my ACADS, but I would never like to repent. I started my own venture soon after my B.E. After a year or so going in loss, I thought more research is what is required. So, turned towards academics. Those days, I never liked the way code was written as most of the programmers had no understanding of the "beauty of the code".
After graduation
I took up a job as a visiting faculty in my own college. Slowly later, I thought I could teach GNU/Linux professionally in and out. I started my own classes in the college premises in the late evenings after the regular college hours, which were 1900 hrs to 2200 hrs everyday where I covered GNU/Linux shells, editors, scripting, device drivers and kernel programming. The syllabus which I framed was impressive and caught attention of 30 odd students. I would demonstrate on the fly using a PC and a projector, and looked like I had changed the way of learning to a certain extent in my own college. I even tried with computer illiterate friends of mine and the response was good. I would take for 3 hours continuously with a break of 5 minutes in between. This developed my communication skills enormously. The outcome of this was a 150 page incomplete book on GNU/Linux and its advanced usage.
As regular teaching in the college was not that interesting (no modifications to syllabus -- University rule), I decided to leave everything and go to a place where it would be different. It was IIT!
IIT Bombay
I came to IIT with a strong intension of research in CS, especially Databases. I longed to understand the deeper mathematical connections in CS and some years back discovered Category theory as a vehicle for analyzing those deeper abstractness which are inherently natural properties of theories(example, set theory) underlying those assumptions of axioms in a language and which cannot be proved using the same language. Very familiar ones could be in English if you ever wondered to answer "what is what?".It directly follows from Gödel's Incompleteness theorem that such an axiom is not questionable and you indeed need a higher superset language to actually prove those axioms. Category Theory(CT) is one such language and it is a language of functions on objects only. It does not deal with object internals but it gives a totally different point of view to things which are familiar in conventional theories. In that, the vaguely connected things are suddenly observed at systematic places, thus revealing the true natural properties. CT gives a single view of the same concepts in different theories, thus treating them same e.g the product of two topological spaces, the product of two sets, etc. Thus, looking at the deficiencies and inadequacies in other theories for example, in representing the basic primitives (say in any programming language), it is very easy to see a new language based on a categorical style where no primitive data types nor primitive control structures are essential. This magic of no data types can be achieved using parametrized FG-Dialgebras. These are nothing but generalizing the notion of adjunctions, which is nothing but natural transformation of covariant and contravariant functor and this is possible because the basic underlying machinery of CT allows us to have duality(eg. duality between cartesian products and disjoint sums) which is not visible in any other theories like set theory, domain theory, etc.
Using CT, data types can be associated with its own control structure, for example, boolean with an if statement, natural numbers with primitive recursion, disjoint union with a case statement, etc. We have realized the importance of data types through programming languages,. The more richer it is, the more the better. These categorical connections have far more impact in realizing the discovery of new and rich data types with their associated control structures. In brief, it looks the application of CT would give rise to a new language whose underlying semantics will allow us to write a far more lesser code in size for any complex software systems for its manageability.
I am working(this is my hobby) on the above idea to actually develop a compiler for supporting CT semantics which may be a useful work to explore in CS. It would be a major task where everything is "abstract nonsense" with magic, that is what every categorist say.
As I came to IIT for Database research, I still work in it but has been in Parallel Databases. I have been involved in the different stages of the project under Prof. Phatak along with his M.Tech/DD students, understanding its applicability to massively parallel architecture in an affordable way. There are many problems which I have personally investigated, some small and solvable, and others big enough to be given as thesis topics. Some of the other areas that can be applied to Databases is control theory, which gives you finer control over the system.
I have also been involved in e-learning with Prof. Phatak and am trying to understand the fundamentals of it.
Apart from all these activities and my hobbies of achieving something, I am also involved in CS 101 to give necessary help when required. Much of my activities would be to coordinate with TAs and I would like to take the whole responsibility if something goes wrong as it is of prime importance and want to understand their(next generation) problems that they face so that it can effectively be translated for e-learning also. I would like to contribute to CS 101 and has been thinking a lot as it is not trivial. May be some of the problems can be translated for e-learning in a different fashion all together.
It is the passion in doing something which keeps my perseverance to achieve those goals in my life which I have in my dream list. May be I have failed many times in understanding my life, but my failures are nothing as compared to Edison who failed 1000 times before inventing an electric bulb. This is my inspiration which keeps me alive and running, making me more enthusiastic about future outcomes. I like to work hard because India is way behind and to catch up with Europeans or Americans we have to work harder(I hear from Prof. Phatak who is more passionate than what I could be). I consider every work as a gift from GOD though I never pray for anything. It is in us to look for what we can do in every alternate possible ways, to serve the nation for whom we are travelers/nomads in our spiritual world where no lunch comes free.
Note: There are some activities which are not listed like writing poems(two diaries of poems on love, life and nature), artistic writing of love letters in a poetic way which I thought of compiling as a book titled "Illusionary love", music especially drums(but presently no time), learning any new language in about 3 weeks to make myself more attentive, to improve self-learning, training my mind etc. I also designed and fabricated a jacket which can give a weight from 1.5Kg to 45kg on the shoulders. This can be used by runners to increase their endurance. I am looking for collaborator(especially from bio-mechanics) who would be interested in measuring the optimal performance of a runner based on some computational techniques. This can have some commercial value also.