R&D Projects
Supervisor : Dr Krishna S
Reports : R&D Project-I | R&D Project-II
I was always fascinated by theoretical aspect of computer science and hence did majority of my courses in theory. After doing the course "Natural Computing", I got initiated to the area of formal languages and automata. My R&D Project-I was a reading seminar, in which I studied P Systems .
In R&D Project-II, I studied in detail the computing power of a variant of the basic P systems, called P Systems with worm objects. The best result so far in this regard was that P Systems with worm objects having 6 or more membranes were equivalent to Turing Machines, i.e. universal. But proving universality with fewer number of membranes is of special theoretical interest because number of membranes is a measure of descriptional complexity. In fact, determining the minimum number of membranes needed for getting universal model is an open research problem.
In this project, I was able to prove an improvement to this result, as in, only 4 membranes suffice for getting universality.
Systems with 2 membranes are already known to be strictly sub-linear, so to close this open problem, one now needs to define the computing power of systems with 3 membranes. Master's Thesis
Supervisor : Dr Krishnas S Report | Slides
The affair with P systems continued even after the two R&D Projects. In my MTech Project, I started out with studying a new variant of P systems called the SN P systems. They borrow ideas from spiking neurons as in Petri Nets. Under the assumption of synchronization, they become universal, but synchronization plays an important role in proving this. My objective was to define the computing power without assuming synchronization, ans hence find out if there is a loss of power due to loss of synnchronization. And if there is such a loss, then which system features can make up for this loss? Cavaliere et.al. proposed the notion of reliability of asynchronous SN P systems as a measure of there ability to correctly simulate Turing Machines (i.e to become unoversal) against growing asynchronism (controlled by a probability distribution) and suggested ways to improve reliability.
So now there were two ways to proceed --
One, Come up with more reliable models of asynchronous SN P systems and define what maximum reliability can be acheived with maximum asynchronism (And hence conclude that purely asynchronous systems can not be more powerful than that). To simulate SN P systems and measure their reliability, the Mobius Modeling Tool was used.
Two, try to prove that the basic model of asynchronous SN P system is strictly sub-universal and adding features like extended spiking rules makes them universal.
Although in conclusion, I got only partial success in both cases, I gained significant insights regarding the computing power of SN P systems and the role of synchroonization.
Seminar
Supervisor : Dr A A Diwan Report | Slides
The even cycle problem for directed graphs is stated as follows --
What is the necessary and sufficient condition for a given a directed graph G = (V,E) to contain a cycle of even length.
Some cousins of this problem are the odd-cycle problem for directed graph and the even cycle problem for undirected graphs. Although both of these problems are polynomial-time decidable, there aren't any known efficient algorithms for this particular case. So what is so special about the "directed" and "even cycle" case? Thomassen et.al. proved necessary and sufficient condition for even di-cycle case. I studied this proof as well as the connection to other known hard problems like sign-nonsingular matrices and how they become easy because of this proof.
Maintaining services like File (NFS and RAID), Mail (Qmail-Ldap, courier-imap), LDAP (OpenLDAP), News (INND) and proxy (squid) for the students and faculty of CSE department.
Maintain and troubleshoot the CSE building network (including one Juniper router and 4-5 IP-layer switches).
Daily incremental backup of the file servers (nearly 500 users with 750M quota at least) using rsync.
Setup and configure nearly 250 machines in two labs for various UG and PG lab courses (like OS, DBMS, S/W Lab etc).
Some tips and tricks:
Arbit (that's again acronym for arbitrary, IIT-B lingo). This is unorganized content. I dumped it here whenever I learnt something. Thanks Uma, and Shyam, my seniors, for this.
A more organized, and much more helpful collection by Vaibhav Gupta, again a senior sysad.
Another one by Suresh, my friend and senior sysad. I learnt a lot from him.
Finally, sysadgiri @cse.iitb was awesome, I enjoyed it a lot.
Geodesic Ltd:
I work for Geodesic Ltd as a software trainee engineer since 3rd August, 2009 till date. Geodesic is an
internet company which has a spectrum of products under the head "Mundu". I work in the Mundu-WebIM team, wherein
we are developing a web based instant messaging client for various protocols like Yahoo!, Jabber, MSN, AOL, etc.
with features similar to those of a full-fledged desktop client like Pidgin. I am responsible for the AOL component.
Kritagnyata Social Welfare Association:
"Kritagnyata" in hindi means remebering (and trying to give back) favours done on you. Kritagnyata Social Welfare
Association is such a society formed by we, the students of CSE, IIT Bombay, in an attempt to "give back to the society" (being "kritagnya").
We are trying to work for the education of underprivilleged children. To this end, we have started making video-lectures for computer-education of students of 5th standard to 10th standard. We are following the syllabus prescribed by Maharashtra State Board. In association with Vanvasi Kalyan Ashram (VKA, for short), we plan to distribute these videos to schools like those run by VKA, who work for the education of Adivasi children. The task involvs writing the lecture transcript, actually recording it (using screen-capture software) and finally, editing the video to reduce noise, etc and produce the .avi file.
Currently, I am mainly involved in recording the videos in Marathi language. There are several other activities like teaching in the IIT campus school (in association with Vidya), clothes collection drive, cleanliness drive, etc. that the group is undertaking regularly. I have contributed to some of these activities.
C-DAC Mumbai:
Center for Development of Advanced Computing Mumbai (formerly National Center for Software Technology) is a scientific society of Department of Information Technology, Government of India. I worked for C-DAC as a Staff Scientist from 31 July, 2005 to 4 June 2006. As part of my work, I taught the Computer Organization module to students of Post-Graduate Diploma in Software Technology (PGDST). I was also associated with Open Source Softwar Resource Center (OSSRC), wherein, we worked with the content management system Drupal for its integration with other open source software like TWiki and CAS.
Dr. B.A.Tech. University, Lonere:
Dr. Babasaheb Ambedkar Technological University, Lonere (Maharashtra State) is a unitary (autonomous) university formed by Government of Maharahtra in 1989. After graduating from CSE Department, Dr.B.A.Tech. University, I joined as a lecturer in our department (July 2003 to Nov 2004). I taught the courses Database Management Systems, Theoretcial Computer Science and Operating Systems to the studets of BTech.
