|
Hello folks,
I'm going to write blog here because I have not set up a good blog for myself yet, but I want to keep a record of the experience, and maybe it can be of use for others.
Writing the paper for publication is the most genuine graduate school experience. I am sure some of you will find this process an interesting read.
all the blog entries will be inside this thread . I don't want to make a new blog entry each time, it might be annoying so they'll all be in here somewhere.
Overall, graduate school for me is a engaging but not a painful environment. With a shitty pay comes great freedom. I can get up whenever I wanted, enter / leave the lab whenever I wanted, and I can take a day off just to play Dota if I feel like it as long as I get my work done. My day at work consists of 5-6 hours of work, which is a mixture of focused thinking, coding, talking to lab-mates for inspirations, and meeting with my advisor. I prefer to think more than code, so most of my time I'm with either a book or a white board.
All of this changes when there is a looming paper deadline. Like all the computer science related endeavors, we drag it until the last minute and go for utter collapse.
I have 10 days left to make the deadline for POPL 15 starting tomorrow, and this will be a documentation of the entire process.
Prelude: Day 0
Counting on my calendar, I have 11 days left still to spent for my paper deadline. I was extremely worried because I don't think I'm ready to start writing yet. As good writing takes time, this can be catastrophic.
A CS paper, or more specifically in my field (programming language) consists mainly of these following components:
1) Motivation for a problem we would like to solve
2) Your general idea of how to approach the problem
3) Your algorithm and formalism to back-up your general idea
4) An implementation of your algorithm as a proof of concept
5) Demonstrate your algorithm on a set of benchmark problems to show that it works
6) Of course, you have to put all 5 things on the paper and present them well!
As it stands I'm stuck at 3 / 4.
Yesterday I was extremely happy because I thought I was done with 4. To my dismay, late night around 5pm I discovered a nasty bug in my code that made the algorithm unusable. Further investigation led me to realize that maybe my algorithm was incorrect altogether, which by then is already 10pm.
I was "rescued" from work by my girlfriend who says we're joining some of her friends to see transformer 4. I didn't want to stop working but I already promised so I went. The entire movie is a huge advertisement for all sorts of Chinese products btw, from hard liquor to soy milk, and I never laughed so much each time I discover an advertisement for which the American audience would have no idea. Her friends told me that China actually owns AMC theatre now, fancy that.
After the movie it was already 1:30am, and I had to sleep. I had dreams about my algorithm, and I woke up at a good time of 9am and I went to work already at 10am.
Except that was just a dream. No. Today I woke up at 11:30am and felt depressed because it seems half the day is already gone. Wouldn't it be nice if I dreamed I was working from 10 to 11am and actually woke up at 8am? I would have worked in my sleep! My girlfriend and I went to get a burrito and watched half of Brazil vs Chile game, and I nearly spilled my burrito yelling GOoooooooooooooooooooooooooooooooooooooooooooooooooooooooal. After that I went straight to the lab.
I use ubuntu for coding, and it has multiple workspace. One of those of course contained the second half of the Brazil vs Chile game. Needless to say not much work was being done even though I only try to play the game as back-ground noise. By the time it was the penalty shoot out there is no choice but to just watch the game.
I spent 2 - 3 hours meticulously printing out the states of my program and drawing it all out on the white board trying to discover what was wrong. At the same time I couldn't really focus well and I kept playing clash of clans and doing goblin raids to get elixers. I need them elixers to research my goblins so I can get more elixers so I can...
![[image loading]](http://i.imgur.com/2Wt2F5M.jpg)
When I finally discovered the problem with my algorithm through series of assertions and prints and drawing, I actually had little idea of how to fix it. It almost feel like I need an entirely new algorithm. That felt too much to do and I got scared. I went to the restroom and came back, and did few more goblin raids.
Deciding goblin raids are no good. I forced myself another hour to really understand what was fundamentally wrong with my algorithm. And I decided I'm going to get dinner.
On the ground floor I found out that there are all these iGem kids around. I looked at them and wondered they looked so happy without a paper deadline! Luckily, the organizer over-ordered food and i was able to grab 2 boxes of sandwitch (turkey craneberry) and I didn't have to buy food. I went outside and sat on the stairs. It was great weather I thought, warm and breezy. I sat there for an hour munching my sandwitch and thinking about my algorithm.
I then walked for an hour and thought some more then I went back to my lab and spent 3 hours to code it up. (goblin raids too, of course, damn these addictions).
Miraculously It worked, or so I hope! If I jynxed myself with this I'll be really sad but the assumptions are strong, and the benchmarks are at least working. I cannot say about the performance but it looks correct.
I sent an email to my advisors summarizing my reports and sat down to write this blog. When you talk to your advisors in email, always keep it brief and keep the good news while never ever complain about the bad news. Keep it cheerful. I promised that I can finally starts writing by tomorrow. One thing I read recently tells me that, nobody actually care much about the facts that you tell them (rarely) but most likely they'll care about how you made them feel. So make others feel you're positive and happy and they'll treat you the same.
Anyways, that's it for today. Starting from tmw it'll be the official 10 days countdown right to the meltdown. Again I apologize for my bad writing, but this is mostly for book keeping on my behalf. If you stayed to read this sentence, I have nothing but gratitude for you. Thank you so much.
Until tomorrow then.
--evan
p.s. for those who are curious, I do research in programming language, with a focus on probabilistic programming. I don't want to discuss the details for now because I don't know yet how to present it in a nice way so everyone can appreciate it. I will definitely do so if you guys want to see it, and when I finish the paper.
 
|
Great read man, that sounds really stressful.
Damn goblin raids are a wonderful distraction from everything around you.
|
United States10328 Posts
waaaait a second do you work with my friend Vlad F? (he does probabilistic programming language stuff and is entering grad school in Course 6)
|
Very interesting read bro, I enjoyed it.
|
You can present an algorithm with out a proof of correctness?
|
@ 343: No I don't think so. I'm primarily still in programming language, my advisor work on program synthesis and the venture into probablistic proramming is a recent one.
@ n.DieJokes: No we cannot. We'll need to prove it.
|
First Blood: Day 1
I was fairly happy yesterday since the algorithm worked out. I promptly wasted all my time from 1am to 3am playing dota and went to bed at 4am.
By the way earthshaker is amazing hero. Maxing out enchant totem first and u can do combo with enchant, blink fissure, hit, enchant, hit, for 700-ish damage. Once you have echoslam that's easily 1k+ dmg. I see people doing it but they all think they're carry that needs crit and shadowblade. That is entirely false. You are still support, except you change your build to max either enchant or aftershock first. You still buy wards and support items. The only thing is you would want a fast blink and if you can get away with it skip manaboot for a faster blink. It's absurd to think an earthshaker who is playing support can have the same playstyle as a templar assassin, with same amount of damage, with a stun to boot, as a support! That's insanity.
My dream was about my algorithm but I also had a dream with playing this video game except different factions in the game are identified not by different costumes or colors but by how they're rendered. There's a 8-bit minecraft faction, a high resolution FFX faction, and a fucked up faction where their entirely body is rendered with paint's spray can brush.
I woke up to a buzz on my phone showing somebody has raided my base. I looked and he only got the collectors and not my storage. I was hitting the revenge button but I found out he's actually still online so I can't do it.
Girlfriend sent me in gchat saying she's found a landlord for me to talk to for an apartment we're looking at. So I called.
Landlord, "Hello?"
Me, "Hi, I'm a graduate student and I'm looking for an apartment, is it open house today can I come and look?"
Landlord, "Can you introduce yourself so I know who you are"
Me, "Uh... my girlfriend and I are thinking of renting this place, she's already sent you an email"
Landlord, "Your daughter is renting this place? what's her name?"
Me, "No my girlfriend is renting this place with me"
Landlord, "Your broker is renting this place?"
Me, "Forget it it's not working"
I hanged up.
I told my girlfriend that my working on the paper has given me an italian accent and the dude can't understand me for some reason. She got mad. I spent 10 minutes apologizing saying my mind wasn't at it and I was being an immature asshole. I drank some chocolate milk, fried 2 eggs, and come to the lab.
Mexico got fucked up by Netherlands. It was so close too. South america players are exciting to watch but they have no discipline and logic. It's depressing because if Europe were to win it'll be a triumph of logic over passion, and I'd hate to see that.
My girlfriend messaged me that she has arranged a meeting with the landlord at 6pm today and she informed me that the landlord is actually Chinese. She demanded me to speak chinese to him because in her words "you can't speak chinese w italian ascent im confident". So I'll have to take off at 6 today I guess.
It's already 2pm and I have not started working. I'll definitely regret this.
...
--to be continued later today as I make some more progress
It is now 4:50pm. I spent the last 3 hours writing up a benchmark to test my algorithm. As it stands the algorithm takes a long time to run. This was just an excuse to watch youtube videos and do more goblin raids. My friend messaged me asking how's the research going, and insisted that, "You should start writing".
I have currently 3 objectives for today.
1) Run more bunchmarks to make sure algorithm is not broken
2) Transcribe the monty hall problem into a benchmark
3) Write up
I've done kind of 1 and 2 and I will be needing to write more. Goblin raids are proven to be distracting though.
I almost had a panic attack when I realized that while the result from running the algorithm looks correct, my encoding of the problem was actually wrong. (i.e. giving high probability when it should've been low in the assumption). Luckily when I fixed this issue my result remained to be correct. I kept having to tell myself that I've thought about this algorithm from first principle and from ground-up, so that it really should have no problem except for bugs, and not a fundimental flaw in my logic. Each time the algorithm pass another benchmark my confidence increases.
...
-- to be continued when I get back
It is now 7:41pm. Went to the apartment but it looked too expensive and dim. Ate bunch of food including some stinky tofu, hope I don't smell bad.
It is now 11:24pm. I've just finished writing a report on my benchmark. I'm happy to find that our language is able to show that in the monty hall problem, it is more adventageous to switch the door rather than to keep the same door. When a result like this comes in you are really confident that your system is working. It's a good feeling.
I've been burping for the past few hours. It reeks of stinky tofu.
Will try to bike home now and get some rest. Overall not a bad day except I should be doing more writings soon.
Until next time!
|
Day 2 already tired
Woke up in the morning, went to my office and found this...
![[image loading]](http://i.imgur.com/5MawwbQ.jpg)
My mexican lab mate is trying to each me spanish. Estoy Cagando entiendo espanol.
It's already 4pm and I have not done anything except watch monty python skits.
I'm going to regret this.
It's now 7:40pm. I spent awhile talking to a friend who is doing something similar, went to swim and completely missed the over-time of german vs algeria, and yet to do my benchmark. I ought to pull my acts together. I'll get some food and come back. I'm sure I'm a horrible student at the moment...
It''s now 11pm and I'm just starring at my whiteboard thinking what I could do. I felt as if all the motivation has left me. I don't know... Maybe I'll drink some hot water and start over.
It's now 12am and I have not done shit. I'm calling it a day. I'm gonna go home and play some dota and jerk off or whatever. Hopefully tomorrow is better!.
|
Dude I need to hand in the first 20 pages of my master's thesis (in Philosophy) this Friday and until now I wrote 1 and a half. I'm in deep shit lol, so I feel you completely.
|
United States10328 Posts
the pedestrian shopping street (not sure what to call it in English...) I walked down yesterday reeked of stinky tofu
bleh
Gogogo good luck! You're gonna end up pulling like two straight all-nighters at the deadline at this rate 
|
day 3 unexpected salvation
Professor(s) and I had a meeting today and we decided to submit to a different conference in Sept 15 instead! Another 2 month!
I'm free to do w/e I want again. Yay! It was a good idea to attempt to submit to POPL though, I think I've done a lot of work under pressure, hope I continue (not likely but I'll try very hard to keep the motivation).
cheers!
|
i dont understand why you dont start 20 days before the deadline and save yourself a lot of trouble.
Also, they pay you to play PC games? I would get smacked :D
|
No more blog :[ It's so honest, don't stop!
|
On July 02 2014 14:38 n.DieJokes wrote:
No more blog :[ It's so honest, don't stop!
Well... it'll be here again in 2 months I'm sure!
@ LaNague
I really tried to start 20 / 30 days before but my advisor was even less pressured than I was. He kept saying "alright let's see what we have" and "sounds like we have a plan!" without actually defining a plan haha. Everything comes down to the wire eventually, I cannot say it is a good thing.
And well I wouldnt' exactly say they pay me to play video games. I play games on my own time just like everyone else who plays video games. I think with graduate school you get slightly more time for yourself and a lot less pay.
|
|
|
|
|
|
EPT
