Searching for something

"We are all here to do, what we are all here to do"
---said by the Oracle in Matrix

I have again not updated my blog daily, but now I have given up. I will update it only once two days or something.

Today was a fantastic day. It started well, it starts with me sitting in Computer Center(CC). You know, technically a day starts at 12 midnight. Ya, so I surfed along the net. Did a lot of work for my research. I am beginning to understand how deep this rabbit hole goes. Then went back to hostel at 2:30 am. This lab shuts down at 2 am ("We dont want to waste ACs for a few people", my mentor, CC head said).
Then I got up at 12 noon. Today there was supposed to be a get together for all the SURGE people. So I was generally feeling very good. Chatted along with Manas, Raja, Gaurav after lunch. ( I generally prefer to have breakfast in my dreams). Then came here to CC (My first home, the other is my hostel room, where I just goto sleep). Then I got a few friends online. Sandhya was online, so I chatted with her for 2-3 hours. The chat went about girls, relationships, NITT, etc. I washed Isli's (Nitin Isloorkar) dirty laundry in front of her. I told her how I cunningly tricked Isli in FTP to find out that his password was "Ilove**104**" which was a girls roll number. But I spared her (or rather Isli) the exact details. Felt very good. Then I chatted with Muayuri Bhavsar ( an arbit friend I added on orkut, because she was from Kopergaon where I studied till 5th std). She was nice. As I was busy chatting with these damsels I forgot that I had to attend the get-together. So I hurried and exited.

Manas & co were all ready, so I slipped in to something, and off we were hurrying for the get-together. There we met CVR ( Murthy) Sir. He is the chief co-ordinator of SURGE. Manas had predicted that Sir was going to crack some joke. And he did. He commented on Raja (wearing a cap at 6:00 pm) that the weather was really sunny. We all had no option but to laugh. So we waited for everyone to arrive. Sir decided that we will wait till 6:15, and dump the people coming later than that. We had 3 sumos for travel. He hinted that he did not like that we had formed local groups: One ours, NITians ie: me, Manas,Raja, Gaurav,Vikas,Abhay ; second IITians Sidhant, Gaurav & co and the third girls - Manpreet, Shwetaleena & co. So he decided that he will randomly choose who will go in which sumo. And he chose wisely all groups scattered. I got into one which had Manpreet. I was the unfortunate ( or rather fortunate ) guy who had to sit along with her. So I talked to her. About hobbies, computers, her NIT, reading, singing, etc. I had a good time. I mentioned to her that I sing. But later confessed that only Bathroom singing... She took it as a clue, and asked me to sing something, but I flatly refused. I am a solitary living being (or I like to live wiht people who like to live a solitary life), and I intend to be that way.

So then we reached Kanpur city, the chaat shop. I had aloo-tikki, pani-puri, dahi vada and then mazaa. It was very good. Then we went to have ice-cream. I had Hot chocolate fudge. It was delicious. My tummy was almost full (It rarely is..). So then we got back to our sumos. This time, CVR sir suggested that girls get to choose whom they want company with. So then all the girls chose and there were 4 girls and some 12 boys. So I was left out, and chose myself. Then we went to the Temple here. It was very good, sir explained us the architecture (CVR sir is a civil engineer BTech & MTech at IITM 10 pt CG ) . He was very good with it. I being an atheist just looked around the temple and the idols. But I did not express my atheism. OK So were back again to our sumos, ready to go back to IITK, this time we got our freedom. We were asked whom to sit with. I chose Manas. So we happily sat together. Behind were Shwetaleena and Manpreet. We just chatted along, and there came IITK (Relativity, huh!!! ).

So, I am back here in CC, writing this blog. Tomorrow I have to go to Lakhimpur (Kheri). I am leaving tomorrow morning at 5:00 am. So I better go and sleep fast. I live in 3 or more Multiverses.. There is my one universe in NITT, one at my home in Ahmednagar, one here in IITK, and one more in Lakhimpur. It is where my parents were born and grew up. I have all my relatives (chacha, mama, mausi, giya, etc every blood relation you can think.. there). So I am going there. My brother Sonu got engaged recently. So I have to meet my hone-wali-bhabhi. The wedding I suppose will be in January.

My johari window, shows that no one knows that I have the quality of "searching" in me. Search is the most difficult when one does not know what he is searching. I remember my Lucky Ali's favourite song:

" Jaane kya dhoondhta hai yeh mera dil,
Tujhko kya chahiye zindagi? "

meaning:

" Dont know what my heart searches for,
Life, what do thee want? "

Started off finally

So I havent blogged now for 2-3 days..

Nowadays mostly whiling away my time chating (chaating) Varshita, isli, et al. Two movies were screened here in IIT K: Hulk (which reminded me of my ex-roommate Tushar Bihani) and Pirates of the Carribean.
The auditorium here is very good.. as good as a multiplex.. and no sex discrimination is inflicted here as in our college (NIT Trichy). The environment here is very free... No restriction for anyone to goto any hostel. That is very good. Had a long debate on this with Varshita. Chatting, orkutting,..

But in the middle of everything I am also researching. Today I met PG Sir. He is going out of station tomorrow.. IIT Delhi and then mauritius I think. I asked him if I could have the permission to go home.
He told me that I was free to do whatever I wanted, whenever I wanted. He wanted only the final result. I am feeling very good. Free. So I am going to live upto his trust.

I coded the Cholesky algorithm. It works in O(n^3). I chatted with Prassanna ( the genius, intership@google banglore) about this. I told him that this algo worked in (n^3)/3 FLOPs while the LU decomposition requires 2*(n^3)/3 FLOPs. So this is twice better. But he told that sometimes a factor of 2 is not worth it. It is possible that the LU algo works more efficiently than Cholesky because of Compiler adjustments. I tested the Cholesky algo on some SSPD matrices of 2,3,4 orders. I want more test data of N=100 or so.. I searched. Found data with N > 10000. Too large. My program will not even take it in.

The ACM portal here is very good. I downloaded all the research papers related to this topic. I have to read through them. I now wonder why I downloaded all of them.. Free net, they are always here...

So here starts my research ...

This is my research proposal:

SURGE Research Proposal

Name: Abhishek Verma
College: National Institute of Technology, Trichy
Department: Computer Science and Engineering
Email id: cs10464@nitt.edu, vermaa@iitk.ac.in
Mentor: Dr. Phalguni Gupta

Topic: Solver for sparse Symmetric Positive Definite(SPD) system of linear equations.

I intend to implement a Solver in C for sparse symmetric positive definite system of linear equations. Large sparse linear systems occur frequently in engineering and scientific applications, and the solution of these systems is an important and costly step. Common sources of SPD matrices are finite element analysis of structures, the pressure correction phase of a segregated fluid dynamics simulation, and the analysis of electrical networks. The Cholesky decomposition of SPD matrices is faster than other methods for solving these linear equations. It requires about n^3/ 3 Floating Point Operations (FLOPs), where n is the size of the matrix. Hence they are twice as cheap as the LU Decomposition, which requires 2 n^3/ 3 FLOPs.

Solution of Linear equations of higher orders (N>1000) can be found by implementing Distributed Computing. I intend to implement this in Cilk. It is a language for multithreaded parallel programming based on ANSI C. Cilk is designed for general purpose parallel programming, but it is especially effective for exploiting dynamic, highly asynchronous parallelism. It can run on a wider variety of platforms, including massively parallel computers, symmetric multiprocessors, and networks of workstations.

A number of serial and parallel software packages for solving more general systems are available. Some of them are Watson Sparse Matrix Package (WSMP), Multifrontal Massively Parallel sparse direct Solver (MUMPS), Unsymmetric MultiFrontal (UMF) Package, etc. The algorithms involved in these shall be studied in detail and their performance will be tested and compared on several standard matrices from various scietific areas.

IIT Kanpur

Let me first introduce myself. I am Abhishek Verma. I am doing my BTech in CompSci in NIT Trichy.

May 11: I am finally in IIT Kanpur. The dream place to be for an engineering student. Here all the facilities are the best you can get. The mess is the best. Hostel is very good. I have 2 chairs, a fan, a computer table and a LAN slot, free net proxy account(I have the SQUID proxy account),.. but no computer. I have packed it and kept it at my senior's room (Sameer Gupta) in Emerald Hostel in NIT Trichy. I am missing it dearly.

I have joined in IIT Kanpur under SURGE - Summer Undergraduate Research Grant for Excellence.
Toady was a tough day. I have been frustrated for the last two days. I was given the topic of Reconfigurable Microprcosessors by my HOD Mr. Ramasubramanian Sir. I filled it and sent it here in IITK. I got selected here mainly on my GPA which is good.

My mentor is Dr. Phalguni Gupta(refered to further as PG). He is the head of the Computer Center here. He suggested to change my topic to Biometric Image Processing, which I gladly did. I came here and was asked to join the Biometrics team here working under a Central Government funded project. I was happy. The head, as it seemed, is Hunny maam. She is intelligent. There are other people I dont know their names. One paticular maam is Richa, who did CS graduation from a college in Lucknow. I could see about 8-9 people working on the project. All doing different traits of image processing.
There were 4 major sections:

1. Face Recognition
2. Iris Recognition
3. Fingerprint Recognition
4. Signature Recognition

I was happy to see all the charts, presentations in the Biometrics Lab. But the team is in the middle -> ending stages of their project. On the first day morning, I was given the job of testing a piece of java code for image recognition. I had to note down the HAAR(An algorithm for face recognition) values and Hamming Distance values in my notebook for each and every image seperately. I thought it OK in the beginning. But then I thought to myself, what have I come here for? do this peon work (I call it hamali (in marathi) or coolie's work). I came here to do research. Then I just told myself to relax, it was my first day. After lunch, I thought there will be light at the end of the dark tunnel..

Ya, there was light indeed, but of the incoming train... I was given a GUI in Matlab for Face recognition. All I had to do was to run it, Click on few buttons, copy paste a few files here and there, then click on the center of the eyes and the lip of the photograph.. The same thing in a seemingly infinite loop. So the whole day, I was doing the testing part of the project. While going out of the lab, I met PG Sir. I told him that testing was really boring. He replied that it was like ragging on the first day and will get better.

Next day arose, and I had to complete the left out work of yesterday. Same clicking on the eyes and lip of each photo. I told Hunny maam, that I did not want to do testing anymore. She is very kind and understanding. She asked me what I would like to work on... I was given Offline Signature analysis.. I thought that it will be a good challenging thing. A Sir.. (I dont recollect his name) gave me a fine introduction to it, and explained that I had to do the identification part. He explained the Range Search Algorithm and the Dynamic Time Warping Algorithms. I understood most of it.... and the rest is left to the reader as an exercise... The next day I was given the source code of the Verification part. I read it and found a flaw in the implementation of High Pass Filter code. Even sir agreed to it, but that library wasnt written by sir and that filter was luckily never used anywhere,..I guess. So my job was left to understand it and make it. I met PG Sir. He asked me how everything was going on. I told him everything frankly. I also told him that I wanted to do something challenging. He said,"you have just completed 2nd year, and you are here for just two months. You cannot do much. It is better if you just assist these people in their work." He suggested some work, but then he disagreed with himself, and said that, it was too young for me. Those were the words that would keep ringing in my ears the whole night. He asked me to come on to Face detection, ie automating the part I was doing manually, finding the face in a photograph. I thought that will be challenging and I slept off.

Next day, the 10th, i.e- yesterday, I was given 2 algorithms for face detection, one was based on the 2D Discrete Wavelet Transform and the other was some Averaging algorithm. Both already implemented in two Matlab files. I also had this GUI of clicking. My work was to pipeline one or both of these algorithms into that GUI. in short, I was made a plumber here... I read some stuff on Discrete Wavelet Transform. It was very interesting and I tried to understand it. But then, Hunny maam said that I need not understand the Algorithm, I just need to do the plumbing part. I continued to do the work. Had a bit trouble with it. At night I got frustrated of all this work.

So I decided to meet PG Sir today.

"Man cannot discover new oceans unless he has the courage to lose sight of the shore"

Well I am in turbulent waters now. I did not beat around the bush while talking to PG Sir. I said that I did not want to work in the Biometrics team, and that they were in their ending stages. I could not find proper work. All algos were already implemented. Just the joining part was remaining. He made two valid points:

• The team was not in their final stages, they had been working for 3 years, and more work lay ahead
• All algos were not already implemented. There is no perfect algorithm for eg: finding where the face of a person is in the photo.

I agree to these points. But I find it unlikely that I will do, or atleast help in doing these works. So I asked him to give me a try, give me a new topic. He warned me that I will be losing my seat in the Biometrics Lab if I quit this. I will have to work in the Computer Center(cc).. (He is the head of the cc). I agreed to it. So he gave me the topic of massively parallel sparse direct matrix solvers for linear equations. I told him I was happy with that. I asked him what the procedure was to get a login ID. He was the cc head. He could have given me an acc there itself. But he insisted on first completing all the formalities.

So here is where my Frustration adventure kicks off. He asked me to get a letter from Mrs. Vibha Gupta (She is the coordinator of SURGE) stating that I required a Login ID so that money could be transferred from SURGE to cc account. So I came out of the office to go to her. But I did not remember her office as she had taken me to her office in her car last time (after the inauguration ceremony of SURGE). I remembered it was related to Dept of Resource Planning and Generation(drpg). I wandererd around there searching for Mrs. Vibha. Then somehow I found the correct location in another building. I went to her and told about this. She asked me who my mentor was. ( I was in a soup, PG Sir had asked me not to mention his name in this). I replied PG Sir... She smirked a bit. And then asked me to meet Shakeel sir in drpg. So I went to his office. He was nowhere to be seen in it. I later found out that he was in the front office, busy with his work. I went to him and explained him.. Vibha maam had already phoned him. So then he asked someone on the phone the formalities to be completed and then got to know that I had to meet Mr. Savitha in cc. I went to cc and explained him everything. He gave me a form, I filled it, I had to fill in the Project Investigators name and get it signed. I went to PG Sir and got it signed from him. I attached xerox of my offer letter with it, and gave it to Mr. Savitha. He said that the drpg has to forward it so that the money transfer can take place. He tried to call up Mr.Shakeel but he wasnt picking up the phone. I again went up to drpg to Mr. Shakeel. He talked with the dean of drpg. He signed my applicaion and asked me to tell PG Sir to give me an account immediately while the dean would talk to him and settle matters with him. So I again went to the cc and spoke with PG Sir. He had an important long meeting next up and was leaving. He said that he could not give me the account unless all the formalities were completed. He asked me to go to the library. I did not have a temporary I Card to get into it. I went to the library. I showed the library guard my offer letter. He asked me to get a pass from Madhu maam. I went to her, and she told me to go to Deputy Librarian Bose. I went to him and asked him about the pass. He said that he wanted a written application from PG Sir, only then he will give me a pass for a fortnight, depending on my behaviour, he will extend it. I asked him to speak to Vibha maam, maybe she could explain. He shouted, who is Vibha?? She is nothing..Ok..Now get out. So I exited the library, and went to Vibha maam. She couldn't do much except agree with me that Bureaucracy in IITK is very rigid.

She tried all means to help me and asked me if I wanted to change my mentor. She asked me to come back at 3 pm. I went back to room, had my lunch, slept, and went to her at 3 as she had told. She couldn't do much and then asked me to wait for a day or two. So I went back to room.

Later around at 4, Manas came to my room( from home) and I explained to him everything. He gave me the solution. Our SQUID proxy ID did not work in Windows Lab, but worked in Linux Lab( and that is what I like more..). So I am here, hapily blogging and chatting for 4-5 hours... I took so much pains to get a LoginID while I already had it...
(I wonder isn't the cc head supposed to know that the SQUID proxy ID works in Linux lab..)
Bye.
(I must leave or else I will miss my dinner.)