26-hour syndrome

Let me explain. There are days when I want to go to bed at 12 midnight, and I get up at 7 in the morning. The next day, I would still want to hit bed at 12, but I'd get up at 9am. The next night I just keep tossing and turning for a couple of hours if I hit bed at 12. I feel if the day was a couple of hours longer, or if I could happily live a longer cycle - sleep more, wake more - things would work out much better. Isn't it possible that God made random errors in manufacturing human beings, and while many were aligned to the 24-hour cycle, some were not?

Amarok 1.4 on newer *ubuntu

If your newer (k)ubuntu comes with Amarok 2 by default, here's an easy way to install Amarok 1.4 even if you just have KDE4.

Research+ w/ Zotero & Jungledisk

Sometimes waiting a while is the best option. I didn't wait, I started to write my own pdf-manager but never really finished it. The goal was to keep track of all my research sources (read online papers in pdf form) with copies available offline whenever possible. Additionally, I wanted to keep notes on these resources as I read this. All bundled together, instead of notes and actual pdf's flying all over the place.

How I designed my to-be manager and what fate it met is a story for another day. Fast forward 3-4 months and find me falling back to Zotero. Why? Because now it's version 2.0. They have an excellent demo video on zotero.org, so I won't detail it here. The thing to note, though, is that I came back to it because they added a server-sync functionality in 2.0. All I needed, now, was a WebDAV server storage, which Jungledisk provides off the Amazon S3 storage for $2/month + 15c/GB-month - peanuts. WebDAV is just a protocol on top of http for generic storage access. Again, jungledisk.com is very intuitive and works right out-of-the-box. (No, I don't use Windows or Mac).

Conclusion: Firefox + Zotero plugin + Jungledisk a/c make it a breeze to keep track of your resources from anywhere, anytime. It's worth your time exploring these.

Caveat: It may not be possible to include your existing resources into the Zotero library. If you think of your research resources/projects as music albums, just switch to Zotero from your next album.

pulseaudio(amarok, firefox, xmms)

This article explains nicely the myriad of linux sound projects and where pulseaudio fits or why to use it. The pulseaudio wiki tells how to set up the alsa driver to route audio to pulseaudio server so that all applications using libalsa (alsa API) can work seamlessly.

I have youtube, xmms and amarok playing simulateneously right now. Did three tracks together sound sweet ever before?

P.S.: I'm not suggesting pulseaudio - it just happens to be the default sound server in most distributions today and it's easier to work with it than around. Things worked even before pulseaudio.

Universal playlist interface

I want an anytime-anywhere playlist interface, so that no matter which music player or which machine or which device I'm playing music from, if I "favorite" a track, it is magically recorded to my has-it-all, omnipresent, cloud-like personal music database, such as one on last.fm or iLike.com. It should also know which source file I was listening from and how I can get back to it. I should then be able to pull up all my favorites into my iPod or like with a single click, and keep it up-to-date as I listen to more new music.

That was from the user side. With current proliferation of server-side music databases like last.fm and musicbrainz.org, this shouldn't be too hard to do. I see two hurdles: 1 getting all client-side software to talk on a single interface, and 2 getting all server-side databases to share information. The latter should be easier to handle as long as the services are co-operative and willing to provide an API, which any reasonable service today does. The first problem also shouldn't be too hard to solve either as long as you're working with open-source and modern software, but legacy commercial players like Windows Media Player and Winamp may be harder to write plugins for and, unfortunately, form a large fraction of user-base.

The sleek part of this exercise would be the universal interface, which must be designed such that it can make it's way incrementally and unimposingly into a user's music experience. Do you know of any such existing solution? Or do you want to build one? Life is too short and music too big to forget and lose a track you liked in the first listen.

Recommended: virtualBox!

No more VMWare or other commercial, hefty virtual machines. Use Sun's FOSS VirtualBox. It's light, streamlined and it works! The configurations I have tested are XP-host, Ubuntu-8.04-guest Mandriva-2009.1-host, Ubuntu-8.04-guest. The latter because the cisco-vpnclient for linux always hangs my dual-core pentium machine (known issue), even on disabling one processor. It ran just fine on the virtual machine.

I haven't got around to testing an XP-guest, but my next configuration will be Mandriva-host, XP-guest. It might not be as clean because of a greater kernel-replace for a Windows-guest but it's too early to say. I'm simply loving VirtualBox so far. It's worth the... effort?

corkscrew

Excepts from man corkscrew:

CORKSCREW(1)

NAME
corkscrew - Tunnel TCP connections through HTTP proxies

SYNOPSIS
corkscrew proxy proxyport targethost targetport [ authfile ]

DESCRIPTION
...
corkscrew is a simple tool to tunnel TCP connections through an HTTP proxy supporting the CONNECT method. It reads stdin and writes to stdout during the connection, just like netcat. It can be used for instance to connect to an SSH server running on a remote 443 port through a strict HTTPS proxy.
...

COMMON USAGE
The common usage of corkscrew is to put the following line in your ssh_config:

ProxyCommand corkscrew proxy proxyport %h %p

This will let your ssh connection go through the proxy with the help of corkscrew.

Butter slices?

They have cheese slices. But I eat bread-and-butter in the morning. And I want to eat it almost everyday. And it's so hard to scrape and spread the butter when I take it out from the fridge. Why can't they just have butter slices too? Imagine your warm toast that just popped out from the toaster with a crisp butter slice sandwiched between them. And now think about digging your teeth deep into it... as opposed to still being stuck scraping butter and somehow getting it to spread uniformly on that toast. Wouldn't you love to have butter slices? I wonder why we don't have them. Or am I missing something?

New grad students to US - housing

The first part of this post talks about admits/admissions/funding and such.

Now that you have decided to join UCLA, let's get you settled here. The primary resource is the Bharatiyam website. It contains a lot of useful information. Below is a collection of questions freshers asked me.

Bharatiyam:
http://www.studentgroups.ucla.edu/bharat/
http://groups.yahoo.com/group/bharat_ucla
bharat DOT ucla AT gmail DOT com

Co-op housing:
http://www.geocities.com/uchaonline/ucha_home.htm

UCLA off-campus listing:
http://www.data.cho.ucla.edu/cfapps/mainpublic.cfm

Craigslist:
http://losangeles.craigslist.org/

UCLA fall 2009 Indian community:
http://groups.yahoo.com/group/ucla_fall2009/


I plan to get down there by the end of August. How do i find a place to stay? Do i post it in the bharat UCLA community and ask for help. One of my friend would be joining with me. How do i go about in deciding on a place to stay?

Options:
- Weyburn Terrace grad housing
- University apartments
- Co-op housing
- Privately-owned apartments



The university offers accommodation to some grad students at Weyburn Terrace for (at most) first two years (you would know if you have been offered a spot - not everybody gets one). This is the cheapest private-bedroom option you will find close to campus. It's about $900/month per person. Even if you have been offered one, you can choose to decline it. Sharing bedrooms is not allowed at Weyburn.

University apartments are more suited to married grad students. They are also economical if you share bedroom with a roommate, which is allowed in this case. These apartments are farther from campus (about 4-5 miles) and harder to get due to a long waiting list. If you're looking to live close to campus early on, this is not the best option.

The "Co-op" housing is not owned by the university, but they allow only students to board. Find out more about it on the website. It's a cheap housing option close to campus if you are fine with the trade-offs.

Privately-owned apartments are the most common accommodation and allow you the most flexibility. Accommodation is rather expensive around UCLA. This are is called Westwood/Westwood village. If you look at the map of UCLA campus, you will notice it's shaped like the mickey-mouse face. There is a pocket of housing to the west of it next to the left ear of the mickey-mouse. This is also the area that includes Weyburn Terrace. You can usually find a place here for $450-$650/mo per person for a shared bedroom (3-5 people in one apartment), or double that for your own bedroom. This housing area provides the easiest access to UCLA. There are also a few houses to the east of the campus (Hilgard Av).

The next pocket is to the south of the campus, south of Wilshire Blvd. Houses will become cheaper as you go farther from the campus. This pocket (about 2-3miles from the campus) is a comfortable distance for biking to the campus or taking a bus. Commute time either way will be about 15-25min (excluding bus-wait time).

You can live farther if you want yet cheaper accommodation. Culver City is a nice neighborhood with bus access. At this radius, there are too many residential communities to list. Refer the map.


How do I find a privately-owned apartment?

I found Craigslist and UCLA community-housing (off-campus) database helpful in my housing search for an off-campus apartment.

I wanted to look at the apartment before I signed the lease and so, found one only after arriving at LA and going around looking. That's what most of my classmates also did. I stayed with some senior grad students while looking for an apartment. The process was to look at the listings, call and arrange appointments to look at the house, apply if to liking and finally sign the lease once settled, before moving in. I fixed my roommates before starting the house search, naturally.

Unfortunately, the Weyburn apartment is offered on an 'unfurnished' basis, except for the basic equipments in the room. It would have been so much easier if all the rooms were furnished.

"Unfurnished" still includes the basic utilities - refrigerator etc. I don't know if you'll get a mattress or not - ask someone who lives in Weyburn.

I wonder if you could let me know about general availability of furniture around that place, rental or purchase, how to lug it into the room and where to store it when I vacate the room at the end of the academic year.

That's the good part - with all the students moving out, there is an abundance of used stuff on the streets. You are welcome to pick up anything you like. If you would rather prefer a new mattress, you could order it from a store and they'll deliver it. Moving around stuff isn't easy without a car, but usually you will find stuff within walking distance. You can always promise us seniors a treat and we'd be happy to help you move in. Putting stuff at the end of the year etc is not an issue - you'll be able to work that out.

Is there some Desi Community of UCLA which helps the Indian students?

Did I already mention Bharatiyam? How about ucla_fall09?


This isn't the most help FAQ. What do I do now?

Are you kidding me?! Well, ok, I know. Finding an accommodation is a daunting task. It appears like a jungle in the beginning. There's no easy way out. Absorb the geography, lay down your preferences (close to campus or not, own-bedroom or not, budget, roommates), then see what fits your needs best. And be on the lookout for postings on the groups above - it's easiest to move into an already settled house.

I hope I can call you up for any assistance when we reach there.

Wouldn't you if I said no? :P

Welcome to Los Angeles, babeyh!

Good die young - Rajeev Motwani

The following was circulated on the LAN of IIT Kanpur in honor of Rajeev Motwani. Some people are very familiar without ever having met them. He was one of those few to me.

Remembering Rajeev Motwani, 1962-2009


IIT Kanpur deeply mourns the tragic and untimely demise of one of its most illustrious alumni: Rajeev Motwani. Rajeev did BTech in Computer Science and Engineering in 1983 from IIT Kanpur, after which he went to the University of California, Berkeley to do his PhD which he completed in 1988 under the supervision of Professor Richard Karp. Rajeev then joined the faculty of the department of Computer Science at Stanford University. He was awarded the IITK Distinguished Alumnus Award in 2006.

Rajeev was a brilliant researcher who made fundamental contributions to many areas: randomized algorithms, computational complexity, approximation algorithms, data mining, mathematical modelling of the world-wide web etc. Rajeev was one of the group of five researchers who gave the world the PCP theorem, one of the most stunningly beautiful and immensely deep results of the modern era. It is this contribution for which Rajeev was awarded the prestigious Godel Prize in 2001. Rajeev was a great teacher too; he has co-authored two books, one on randomized algorithms and another on theory of computation which have taught these two subjects to students all over the world.

Those in IIT Kanpur who knew Rajeev as an undergraduate, remember him not only for his academic brilliance but also for the good cheer that he always exuded. He was a very friendly person, and remained so all through his life. His office door at Stanford was ever open – to students, to young entrepreneurs, to academicians from all across the world, and to his friends from IIT Kanpur. Rajeev was also closely involved with the Department of Computer Science and Engineering at IIT Kanpur as a mentor in its research program. He was on the board of Research I Foundation of the department since 2003.

Rajeev possessed the rare genius of using the most abstract of theories to solve practical problems impacting society at large. For example, he made use of his expertise in randomized algorithms in building a drug design system for Pfizer. Rajeev was an early “unofficial mentor” of Google founders Sergei Brin and Larry Page. He helped them through the initial years when the basic page ranking algorithms were implemented and continued hand-holding the young company when they went in search of venture capitalists. Google founder, Brin, described Rajeev as his “friend and teacher” and said that “... his legacy and personality lives on in the students, projects, and companies he has touched. Today, whenever you use a piece of technology, there is a good chance a little bit of Rajeev Motwani is behind it.”

Anything but what I should be doing

I was sure I was not the only one who always wanted to do any- and every-thing other than what I was supposed to be doing, but never really understood why, until it suddenly dawned on me. It's the professionalism that every damn activity has reached in this world today. You would not be doing something unless you were a pro, or were in the process of becoming one. Were it otherwise, you would surely be paying an opportunity cost because everybody else in the world would be making better use of their time. (Comment: the learning process itself is a well-defined activity too, and you would be doing it only if you had the pro-level learning aptitude).

The problem then is that you don't allow your mind to breathe. You don't let it run free and land somewhere. You beat and mold it into what you "should" be doing now. You kill down all scope of creativity. You take out the element of choice. While you should have been the master, you become a slave. If you're lucky, you may still be doing what you wanted, but at any rate, it's no longer "you" doing it. The effect is that when do finally get down to doing it, it demands the very professionalism you were trying to escape when it was that "other" thing you wanted to do. That's not how it was when I was a child.

I beg. I beg to be free. Let me free. Please, let me free. Let me choose. Let me soar. Let my mind breathe. Let my heart and actions be one. Let my life be mine, and yours yours. Once again.

Obsession is bliss.

You learn about life as you live, and the other way round. Learning and living are in bijection. One doesn't beat the other. Living probably stays a step ahead most of the times, but not always. This figment is one such case in my perpetual discussion on life and its purpose.

You are born innocent and ignorant, though programmed with some "fore-knowledge". Then comes your first acquisition of worldly knowledge in the form or material perception. You learn about a few "things" and, unknowingly, obsess with some. In the next stage, you learn about "obsession" as an article itself, along with another multitude of non-materials. That's exciting, and you also know you're "excited". You're probably an adolescent now. You spend the next couple of years tearing the hair from all metaphysics and, in varying choices, teach yourself to master some of these meta-articles. Desire was definitely one such for me.

I taught myself, or so I think, to make my desires voluntary, and to be indifferent. Voluntary, I said. That is the demon in hiding. When you finally do encounter it, things begin to fall apart like never before. You doubt all that learning and principles you developed on your first meet with the meta-world. You doubt the very foundation of it. And if you happen to read "Zen and the Art of Motorcycle Maintenance" somewhere around this time, things only get worse.

You don't know anymore why "reason" should be superior to "feel". You're not sure why you wanted that control over your desires really. The control is not the bother. It's what to do with it. There isn't a straight direction to take. In fact, I don't know if there is any objective to life at all on which to base your decisions with that sacred control you so worshiped. And you think, think and think but it gives you no lead. All you wish now is to be that innocent and ignorant self again, governed by involuntary nature-programmed emotions; to be no longer holding that leash to yourself. And be obsessed once again, without reason. And to live in it than the void.

I "rode" the MINI!

The one we drove today is called Cooper MINI. It's bio-data looks something like:

owned by BMW
manufactured in the UK
1.6L 16-valve engine
112bhp at 6000rpm
0-60mph in 9s
37mpg on highway, 28 in the city
starts at just $18.5k!

As if that wasn't enough, the real deal comes when you rev it on the road in the sports mode. There are cars to your left and cars to your right, but one kick to that pedal and right after the intersection, your rear view mirror is spotless. All you see is some blonds and some brunettes, their necks turned and craned (even men, for ye girls driving). The MINI is like sex. It can only be experienced, not explained.

Note: The car likes to run comfortably at 1700rpm, but in sports mode, it doesn't want to run below 3200 ever. Like it just took the pill.

It's one bomb! Oh MINI....

"Multi-core" car engine

The Green Multi-core

I was talking to my roommate who is a Mechanical Engg major. I learned that the car's combustion engine, say of 1600cc combustion capacity, does not have all this volume in one chamber. It has multiple "cylinders", each with two valves, which share this capacity equally. So if the engine is label V8, it has 4 cylinders, each of 400cc volume. Each of these cylinders also has a dedicated spark plug which ignites it's fuel injection, and is, in turn, connected to the crankshaft through a piston.

I know that sometimes the car seems less powerful (and may drive slightly bumpy) because a cylinder went down (the spark plug, usually). If the car can still operate almost fine with a cylinder down, why cannot the engine be designed to operate in dual (or multiple) modes where different number of cylinders are active in the two modes, say half and all. The all-mode can be the regular, full power mode while the half-mode can be the less powerful but more fuel-efficient mode, analogous to a multi-core processor chip.

I have not tried to do a technology/literature survey. Such a design may already be underway. I would be happy to learn about it.


The Green Hybrid

Some time back, while in India, I had the idea of a car that could operate in both the fuel-mode and the electric-mode, so that at traffic crossings and jams, it would operate in the pollution-free electric-mode and in sparser outskirts with higher speeds, would automatically switch to the regular fuel-mode. While I was discussing this with a Mechanical Engg friend, I came to the US, only to see hybrid cars already on the streets. Rest is just implementing a speed-mode transition policy, which will eventually happen as the need becomes more prominent here. (The need is surely more prominent in India, where pollution is more severe. A low-cost implementation would thrive as a business in one of the biggest markets of the world.)


The Green Recycler

It's interesting how I have "Green" in all three parts of this post. For two of these, that was not a part of the original idea at all. I was just trying to design something more efficient and economical. It's rewarding to see Green and Economical rhyme. Or maybe not. One could as well ask why Green is the most important of our economic problems today. It's a punishment to one and all. But I would like to stick with the reward-view, since that's future and hope.

The Green Recycler goes like this. Every time we take pickup from the engine, we burn fuel, generate energy and pump it into the car. Some of it is lost as heat, but a major portion is used to accelerate the vehicle. Why then, when we brake, do we not recover that energy, only to reuse to pickup again? All we need is to apply brakes electromagnetically instead of mechanically, using magnetic force to brake instead of friction. The reverse current thus generated can be stored and reused. The answer again came from my roommate, who told me that the MagLev train in Japan does exactly this. Perhaps it's not yet economical enough to do this for vehicles. But yeah, the idea is already there, and will make it's way as we develop. It seems we (our generation, that is) are now educated enough so that our ideas are at par with or leading the state of the art/development. We are ready to contribute.

Tar stains on the carpet?

It'll give you nightmares, I'm sure. It's not charcoal or juice stains. It's tar. The thing used to seal plugs in your roof, or to pave the road. It's supposed to stick. Stick strong and hard. And this time it chose your carpet.

I was trying to save six hundred dollars to pay my landlord to clean the carpet when I would vacate my place, only till I read on a forum that WD-40 dissolves tar. And, lo and behold, it did! WD-40 is that bluish bottle of lubricant with the characteristic yellow tubular folding nozzle attached to it's spray mouth. You can find it on any hardware store for $6-7, and it will leave your tarred-patches spotless. Literally.

Sleep well tonight.

New grad students to US

The second part of this post talks about housing and such, once you have decided to join UCLA.

Here is an FAQ for new admits to graduate school in the US, especially those from India. Some parts may be specific to UCLA/LA. Feel free to discuss and contribute. Also, a lot of these issues could be very subjective, depending on your background, your disposition and your goals. What I write here is just my opinion.


Q. I saw approximation of the fee on the UCLA site and the i20 is 44k. However, what does the actual fee turn out to be? They say that MS CS is a 5 quarter course...so is the fee / quarter. In the nut shell, how much will the tuition come to for the entire MS?

UCLA fees and tuition amount to about 24k/year for the MS program. That is for three quarters a year, excluding summer, therefore about 8k/quarter. Living costs included in the calculation are estimated to be about 2k/month, adding another 20k for I-20. Tuition + fees alone are, therefore, 24k/year. MS program usually will take 5 quarters of university. (The spread will be 6 quarters because it will include a quarter of summer). It is not hard to finish in 4 quarters if you plan out your curriculum well.


Q. What is the approximate duration of the course? Is it 1.5 years or 2 years?

1.5 years, as explained above.


Q. What does your monthly expenditure come to, staying in LA + misc..

Obviously, that subjective, but LA, especially Westwood (the area where UCLA is located) is one of the most expensive areas in the US. If you live close to the university campus and share bedroom, you will pay about $600/month for house rent (including utilities). A personal bedroom would be anywhere from $900-$1100. Assuming you share, $1000/month is a moderate, comfortable expenditure, but even the stringent estimate would be atleast $850/month.

Q. How is the aid situation? How easy is an RA/TA? I haven't been offered any aid as of now... So, do MS-CS grad students get something at least by the 2nd quarter or so, or there is absolutely no aid?

As of April 2009, the economy's downturn has affected everyone. The aid situation is bad. Even PhD students don't feel secure. Aid for MS students is yet harder to find. In my batch, a few have found TA positions are in the third quarter, but new students without aid should not count on finding one. RA positions are similarly hard to find due to very reduced funding to professors. Apart from RA and TA, there are some "reader" positions that include grading homeworks, and pay about $1000/quarter at best. Usually, you can hold more than one reader positions but, again, don't count on it.


Q. How easy do campus jobs come by, and what do they pay like? I am really hoping that cali should be better than elsewhere...

Campus jobs don't pay enough to buy tuition - best jobs will pay about $15-20/hour (usually coding jobs). Campus jobs like working in stores/restaurants etc will pay about $7-10/hour. Given your work restrictions and available time anyway, this is all peanuts, especially since you're also paying NRT.


Q. How is the internship / job scene there? Is UCLA CS as affected as other colleges?

Quite affected. Some students are deliberately extending their stay in the program because they didn't find a job of their liking.


Q. And right after your admit, did you write to professors, and if so, did it help? I wrote to a few profs before the admit, no response....do u think that wriiting now can possibly help for RA/TA?

I wrote after my admit and got no response. Professors are very busy, so it's no surprise they don't respond. You shouldn't read anything into it. Since the funding scene is pretty bad, there are a lot of students roaming around for funding, so perhaps it's not that much personal/individual attention to student funding right now. My best advice to try find a funding is to arrive early - say 2-3 weeks earlier than most students would, and approach a professor immediately. That's what I did, and it worked. Chances of finding a yet-vacant TA position also improve if you come early. But again, no guarantees.


Q. I know if I should be taking such a heavy loan...so, do you think, taking the job + internship situation there, that UCLA CS will be worth the money I pump in?

Always entirely your call. From my personal experience and exposure, I'd say that the research scene and culture in US is significantly different than in India. This is not to compare professors, students or the quality of work, whatever is being done in either India or US. I'm only comparing the environment, culture and the working of things. If you aim to make research a career, the US experience would be very valuable. As for loan, try to imagine yourself 10 years down the line on either path, and weigh loan today against your place in 10 years, rather than weighing against your/your family's current situation. That should be a more appropriate basis for decision. But, like I said, it's entirely your call.


Q. I have an admit from Cornell MEngg. CS as well now, so I have to choose between UCLA CS and Cornell MEngg. What is your opinion?

That's exactly the situation I was in. The Cornell MEngg program is a rapid 9-month two-semester program with 5 courses each quarter. It is not a research program, be design. It's more of a terminal degree for students looking to enter the industry. MS, on the other hand, is a degree that prepares you for a research career through PhD. MS also allows you more opportunity to do research. It was mainly this that led me to choose UCLA MSCS over Cornell MEngg, because I was sure on continuing for a PhD (either at UCLA or another university). Another factor was that UCLA MSCS to PhD transition is internal to the department and therefore, easy if you perform well enough in the MS, while at Cornell, switch from MEngg to PhD would go through a full central application procedure and is much less common (to the best of my knowledge).

Q. How do you think are the job prospects from UCLA for an MS CS grad...And if they have deteriorated sharply in the past couple of falls, were they great before that? And how do you thing it compares with Cornell job-wise?

I'm not very aware of the job prospect comparison of the two programs.

Define "the Internet"

This is a question. I don't know the answer to it myself so I ask you.

I know what the internet protocol is - it's as defined in RFC791. I, therefore, also know what an internet(work) is. I understand what we mean by the Internet, but I am unable to define it, with all my knowledge of networking and network protocols as a graduate student. As a consequence, I am also unable to define the World Wide Web, the definition of which depends on that of the Internet.

The best definition I can come up with is this:

"Internet is that inter-network which has the most nodes (hosts + gateways/routers) in the world that run the internet protocol (RFC791) to inter-connect."

Please contribute if you can provide a better definition.