|
12/24 |
2013/5/1-18 [Computer/SW/Languages/Java, Computer/Theory] UID:54669 Activity:nil |
5/1 What's the difference between CS and Computer Engineering? http://holykaw.alltop.com/top-ten-paying-degrees-for-college-graduates \_ One is science and the other is engineering. \_ From http://en.wikiquote.org/wiki/Computer_science 'A folkloric quotation ... states that "computer science is no more about computers than astronomy is about telescopes." The design and deployment of computers and computer systems is generally considered the province of disciplines other than computer science. For example, the study of computer hardware is usually considered part of computer engineering' Here in Cal, though, I think an EECS major is free to take a whole bunch of CS16x and even CS17x theory courses (maybe that's why it's called EECS rather than CE) while a CS major can take EE courses. As if it's not confusing enough already, check this out: http://www.eecs.berkeley.edu/education/degrees.shtml BTW, another set of seeming confusing majors in Cal: BA Chemistry, BS Chemistry, BS Chemical Engineering. --- L&S CS class of 1993 |
2012/12/19-2013/1/24 [Computer/Theory, Computer/SW/SpamAssassin] UID:54564 Activity:nil |
12/19 http://www.extremetech.com/extreme/143382-programmer-creates-800000-books-algorithmically-starts-selling-them-on-amazon More books baby!!! |
2012/2/15-3/26 [Computer/Theory] UID:54309 Activity:nil |
2/15 New problems found with RSA keys: http://preview.tinyurl.com/6ondkw6 [ars technica] |
2011/12/9-2012/2/6 [Computer/Theory] UID:54255 Activity:nil |
12/9 Improved approach for matrix multiplication: http://www.cs.berkeley.edu/~virgi/matrixmult.pdf http://preview.tinyurl.com/cdf5u32 [new scientist] |
2011/4/26-7/13 [Computer/Theory, Health/Women] UID:54095 Activity:nil |
4/26 Is it correct to say that Godel's work on the incompleteness thm proved the Principia Mathematica wrong? \_ It didn't exactly prove it wrong; it proved that the true goal of PM (a complete and consistent set of mathematical truths) is unattainable. -tom \_ Ah cool, no this is good. See ok yeah so the main goal of PM was to "be complete" but obGodel so that fails. However as a piece of logic, that is, a system build up from first principles is PM is still valid, inspite of Godel? etc.) --OP \_ What Godel showed is that you can use PM's language (or any other complete mathematical language) to create a paradoxical statement. That means that you can't assert that any statement described by PM's language is true. But in practice it doesn't change much. (Although why you would use PM's language in practice is unclear; it's rather a theoretical exercise). -tom \_ Would you use it to teach logic to a someone? Is it at least good for that? \_ Sure, it could work for that, but it's pretty complex. -tom \_ Would it be correct to say, "If you studied logic and understand how to follow logic, then you should be able to read the PM." (as a milestone marker)? \_ It proved that PM is incomplete. DUH!!! \_ Well, more precisely, it proved that a symbolic logic cannot be both complete and consistent. It would be more accurate to say that PM is inconsistent than that it's incomplete. -tom |
12/24 |
2010/8/29-9/30 [Recreation/Humor, Politics/Foreign/Europe, Computer/Theory] UID:53940 Activity:nil |
8/29 http://www.google.nl/trends?q=ramadan,+porno&ctab=0&geo=ma&date=all&sort=0 Funny graph. -- linkpusher |
2010/8/23-9/7 [Computer/Theory] UID:53933 Activity:nil |
9/20 Why does everyone talk about Turing but nobody talks much about Babbage? \_ arithmetic vs algorithms |
2010/1/8-19 [Computer/Theory] UID:53619 Activity:nil |
12/8 Will she be as hot as Pris from Blade Runner? http://www.truecompanion.com/about.html |
2009/12/8-26 [Computer/Theory] UID:53581 Activity:nil |
12/8 http://www-cs-faculty.stanford.edu/~knuth/musings.html SOUNDS EXCITING The next talk in the series will be entitled Spanning Trees and Aspects (The 15th Annual Christmas Tree Lecture) The spanning trees of a graph with n vertices are the sets of n-1 edges\ that connect the graph. In this lecture I'll discuss the remarkable relation bet\ ween the number of spanning trees and the aspects of the graph, which are define\ d via matrix theory. For example, I'll explain why the number of spanning trees\ of an n-dimensional cube is equal to exactly $2^{2^n-n-1} \prod_{k=1}^n k^{n\choose k}$. \_ Cool find, thanks. \_ Haven't you guys seen all these before? I saw these when I was in high school in Wisconsin (yay online). |
2009/9/18-29 [Computer/Theory] UID:53378 Activity:nil |
9/18 I forgot my math. Say the probability of a bug happening is the unknown fixed value p in [0,1]. I attempt to reproduce the bug until it happens once. Then it happens at my n-th trial, so I stop. Now, what is the expected value of p? Is it E(p) = 1/n? Thx. \_ Did a quick program to test. Looks like E(p) = 1/n (given assumption n(p) = sum x = 1 to inf of x*(1-p)^(x-1)*p where n(p) = expected n for a given p) \_ Consider the case n=1 to see why E(p) = 1/n can't possibly be right. You need to specify a prior distribution and then apply Bayes' theorem. If your prior distribution is the uniform distribution on [0,1] (i.e. before running the experiment you think all values for p between 0 and 1 are equally likely), then if you see your bug for the first time in the nth trial, you'll get a posterior distribution of (n)(n+1)(p)(1-p)^(n-1). The expected value of this is the root r of the equation (1+nr)(1-r)^n = 1/2. For n = 1, E(p)=1/sqrt(2). For n = 2, E(p) = 1/2. For n = 3, it's around .3857. This sequence probably has some name, but I don't know it. [In practice that might be a stupid prior to use for your problem. But you have to have some prior, otherwise the problem is not well-specified mathematically.] \_ Doy, wow I'm an idiot. I'm wrong, this person is right. -pp \_ That is a fine answer above. The part about Bayes Rule and having to make some assumptions about priors/distribution properties is hard to explain. All I have to add is OP may enjoy reading about the NEGATIVE BINOMIAL DISTRIBUTIION. (while you are at it, you might read up on POISSON DISTRIBUTION/ PROCESS ... which can be quite useful in a lot of engineering contexts). |
2009/8/30-9/9 [Computer/Theory] UID:53309 Activity:low Cat_by:auto |
8/30 I thought this was kinda interesting. Monty Hall Problem, a classic: http://en.wikipedia.org/wiki/Monty_Hall_problem \- one of the first profs to write on this is from ucb pub health the newcomb paradox [from livermore] is even crazier. \_ This is now a classic problem in cs70. I totally didn't believe it for a while, but I think understanding why you always want to switch doors gives a lot of insight to probability theory. -mrauser \_ What about the monty python problem? \_ See also: Principle of Restricted Choice -tom http://en.wikipedia.org/wiki/Principle_of_restricted_choice_%28bridge%29 |
2009/7/14-27 [Academia, Academia/GradSchool, Computer/Theory] UID:53139 Activity:nil |
7/22 (redux) To those in academia, how do you organize the journal articles you keep (hard copies, that is)? By date? Author? Subject? Something else? Thanks. \_ I am not in academia, but author makes most sense. If you want, you can group by subject (Sun Spots, Solar Wind, Solar Flares) and then by author within that. Depends on how broad the topics span and how many papers you have. For the well-known papers within a field people are going to know them as "Moore, 1972" which is why I suggest authors rather than topics. \_ I assign each subject a prime number, e.g. Sun = 3, Flare = 7, then I file Solar Flares in the 21st spot in my library. It is left to the reader to see why this works \_ High school math about uniqueness of prime factorization. (But then how many slots are there in your library?) |
2009/4/30-5/6 [Computer/Theory] UID:52923 Activity:nil |
4/30 Sorting question! I have n sorted arrays of doubles. What's the fastest way to sort them into 1 big sorted array? \_ as mentioned below: you are describing one half of mergesort \_ You really have to ask this question? \_ You don't know either, huh? \_ If three are n sorted arrays of m doubles each, I think the fastest way is O(m * n log n). \_ Read up on mergesort, now do the logical next step. \_ Mergesort doesn't take advantage of the arrays being sorted. You only need to consider the first unprocessed value per array when choosing the next to insert. \_ The second half of mergesort takes advantage of the arrays being sorted. That's the whole deal with mergesort. You merge two sorted lists, starting with the trivial case of a single element list. \_ If each sorted array has m doubles, is the complexity for this O(n^2 * m)? -- yuen \_ No, it depends on you make your max function work. I can think of a pretty simple way to do it in O(mnlogn). \_ Hell, you could just jam the lists together and do a mergesort with the starting list size of m (if the lists are all the same length, add a null list terminator if that's the case.) There's tons of ways to do this all close enough to the same big O that the ultimate "fastest" is going to depend on real limitations like is your data so large that it spans multiple file systems. \_ I don't exctly have a pattern handy, but it should be pretty easy. \_ http://en.wikipedia.org/wiki/Bogosort |
2009/4/3-5 [Computer/Theory] UID:52788 Activity:nil |
4/2 Math gangsta rap link:xrl.us/benmir \- wow, i know all of those books ... those are the "cheep asian editions". in the us, those books together would be about $350. [i am assuming the one in the back is Herstein ... "Herstein is hard"]. actually i am not sure the front book is the DOOB Measure Theory book ... maybe there is another GTM called Measure Theory ... can anybody read the name of the author? "Herstein is hard"]. |
2009/1/13-22 [Computer/Theory] UID:52367 Activity:kinda low |
1/13 I am writing a commandline parser for a class and I could use some tips for algorithms to use. (The project is over and done so I am not cheating, but I am dissatisfied with my end result.) I STFW and didn't come up with too much I liked. I read the source for some shells like tcsh and that is *WAY* too complicated and relies on a lot of other code. I know that browsers and other apps have parsers, too. Is there some simple algorithm they use? I am more interested in algorithms than any implementation itself. \_ Are you looking for something like getopt? Or is this more "read this line and figure out what programs to run/ what outputs to pipe into other outputs"? And what language are you using? \_ C language. I am interested in how something like getopt is designed in terms of its algorithm, particularly for multi- character options. Like how it distinguishes -v from -vv from -vvv for example? Does it read ahead? I am trying to do this without using other functions like strtok() but actually parsing a character at a time. I know I can go read the getopt source, but this is not that complicated. There must be some simple parsing algorithms these parsers implement. \_ You probably want a trie and a simple state machine. \_ Why would you do it a character at a time? Do you mean you don't know how to implement something like strtok? \_ Yes, that's what I mean. What algorithm does it use? Are there better choices? What if I wanted to implement a better strtok or a better getopt? \_ Well, that's not really the interesting part. If you just look at strtok's header it wants you to specify what constitutes a token delimiter. So -v versus -vv is pretty obvious I think... \_ I was once interested in parsing mathematical expressions. However, rather than rolling my own, I just made use of python's expression parser. This proved a good way to parse expressions with very little coding on my part: all I had to write was a little glue/wrapper. \_ Greed -- for lack of a better word -- is good. \_ why does gordon gecko get in more trouble than madoff? \_ If you're interested in this from a theoretical perspective, either take CS164 or try reading the Dragon Book. It's way overkill for basic command line parsing, but you said you were interested in the theory of parsing... that's what a lot of CS164 is about. \_ What is the title of CS 164? Compilers? \_ What is the title of CS 164? \_ http://letmegooglethatforyou.com/?q=CS+164 \_ Back in Spring '92 it was "Programming Languages and Compilers". I took it under Fateman. We only learned a lot about compilers, but very little about programming languages. Other schools have separate classes for the two. |
2008/12/18-2009/1/2 [Computer/Theory] UID:52270 Activity:nil |
12/17 What is going on here? (graph of money supply) http://tinyurl.com/5on6w9 \_ Fed keeps cutting interest rate, and printing money to give out to banks. Banks don't need to lend out money, because the Fed keeps giving more to them. So, more money supply without more circulation. \_ This is GWB's last chance to give cash rebates to the people! \_ who's people? \_ You meant to write, "whose people?" \_ or "who is people?" "Are we dancer?" \_ 2nd sentence doesn't make any sense. Banks are soaking up loans from the Fed to shore up their balance sheets. Once it doesn't look like they could collapse at any moment, "they will start lending again." Which means lending to more people than just those who are a sure thing (fico 750+ ?), like they're doing now. |
2008/11/25-28 [Computer/Theory] UID:52108 Activity:nil |
11/25 Holy crap! I was the one asking how to open a locked up masterlock a few days ago. I tried the algorithm on this webpage: http://www.wikihow.com/Crack-a-%22Master-Lock%22-Combination-Lock and it worked! Thanks to whoever suggested it. \_ Good to know, thanks for the reply. |
2008/11/25-12/1 [Politics/Foreign, Computer/Theory] UID:52096 Activity:nil |
11/25 "Study: Math teachers 1 chapter ahead of students" http://news.yahoo.com/s/ap/20081125/ap_on_re_us/qualified_teachers Our education system in the world's sole super-power country is so broken. \_ lollllzzzzzzzzz \_ God Bless, and free market works. -conservative \_ Free market gives us private schools \_ nice socialist bailout yer free market is taking \_ Banking and mortgages aren't a free market. They are deeply intertwined with the government (Fed, Fannie/Freddie) |
2008/5/19-23 [Recreation/Humor, Computer/Theory] UID:50000 Activity:nil Cat_by:auto |
5/19 http://nfccomic.com/comics/205.jpg Math comic, geek humor. \_ "humor" \_ That's hilarious. Let's come up with some other hilarious ideas for comics. How about one where pi comes along and the others eat him? HAHAHA! |
2008/3/21-25 [Computer/Theory] UID:49522 Activity:nil |
3/20 Road Coloring Problem solved. http://news.yahoo.com/s/ap/20080320/ap_on_re_mi_ea/israel_math_riddle "Let's say you are lost in a town you have never been in before and you have to get to a friend's house and there are no street signs, the directions will work no matter what." How likely is it that all the intersections (vertices) in a town have the same number of streets leaving them (out-degree)? \_ This is not possible, I don't think, unless there is some kind of landmark you can orient yourself to. \_ How about if you have bendy roads? \_ Yes, I guess I can imagine a town that is all on one circular road, with one way in and out. So your directions could just be "keep going around the circle until you get to my house." |
2008/3/18-21 [Science/Electric, Computer/SW/OS/FreeBSD, Computer/Theory] UID:49488 Activity:low |
3/18 Walking robot video, really cool! http://www.youtube.com/watch?v=W1czBcnX1Ww \_ From MIT. All the good stuff's from MIT & Furd. Where's cool stuff from Cal? \_ BSD? Atomic Bomb? \_ Actually RHex, which was a philosophical precursor to BigDog, came out of motion work done by Bob Full at Cal. -tom \_ Wow, that *is* impressive, especially the recovery after getting kicked. And recovering on ice! Walking across loose stones! Jumping! \_ That is utterly cool! Thank you for posting that! \_ Mount a machine gun on top... John Conner would be proud! \_ They showed that on discovery channel or something months ago. Where are the sexless tv watchers when you need them? \_ obIWelcomeOurNewRobotOverlords |
2008/2/26-3/4 [Politics/Domestic/President/Bush, Politics/Foreign/MiddleEast/Iraq, Computer/Theory] UID:49265 Activity:nil |
2/26 the day the routers died http://www.youtube.com/watch?v=_y36fG2Oba0 |
2007/10/24-26 [Computer/Theory] UID:48433 Activity:moderate |
10/24 20 year old Brit proves Wolfram's 2 state 3 color Turing machine is universal. http://blog.wolfram.com/2007/10/the_prize_is_won_the_simplest.html \_ And Wolfram is still an ass. \_ Why do you hate cellular automata? \_ He always comes off as self-aggrandizing but I don't know anything else about his character. \- at one point all the other names that came up when you started mathematica were suing Wolfram. when Wm Kahan took him down in 10 Evans, that was totally awesome. "you had to be there". \_ What do you mean "took him down"? \_ "verbally abused" |
2007/8/30-9/3 [Computer/Theory] UID:47837 Activity:nil |
8/30 what AI is editing the motd? that's some next level shit. \_ Sorry. -- ilyas \_ ilyas -> i l y a s -> s l y a i -> Sly AI \_ ilya shpitser -> lisp hysteria |
2007/8/20 [Computer/Theory] UID:47669 Activity:nil |
8/20 origami transformer insanity WHAT THE HELL happened since then? Did he get a brain transplant? Arrgh. Pitchforks. Riots. Street. \_ I WHAT THE HELL happened since then? Did he get a brain transplant? Arrgh. Pitchforks. Riots. Street. \_ I met Dick Cheney in 2004 when he gave a speech at my work. He came across as very articulate, experienced, and a natural born leader. So if you feel he has changed in some way then it must have happened since 2004. \_ So we both agree he was smart at some point. What the hell happened between 1994 and up to the Invasion http://wonderdasher.blog.sohu.com/59275208.html met Dick Cheney in 2004 when he gave a speech at my work. He came across as very articulate, experienced, and a natural born leader. So if you feel he has changed in some way then it must have happened since 2004. \_ So we both agree he was smart at some point. What the hell happened between 1994 and up to the Invasion WHAT THE HELL happened since then? Did he get a brain transplant? Arrgh. Pitchforks. Riots. Street. \_ I met Dick Cheney in 2004 when he gave a speech at my work. He came across as very articulate, experienced, and a natural born leader. So if you feel he has changed in some way then it must have happened since 2004. \_ So we both agree he was smart at some point. What the hell happened between 1994 and up to the Invasion WHAT THE HELL happened since then? Did he get a brain transplant? Arrgh. Pitchforks. Riots. Street. \_ I met Dick Cheney in 2004 when he gave a speech at my work. He came across as very articulate, experienced, and a natural born leader. So if you feel he has changed in some way then it must have happened since 2004. \_ So we both agree he was smart at some point. What the hell happened between 1994 and up to the Invasion |
2007/5/7-9 [Computer/SW/OS/Windows, Computer/Theory, Computer/HW/IO] UID:46535 Activity:nil |
5/6 Police Lineups go Virtual: http://www.primidi.com/2007/05/06.html#a1828 |
2007/3/20-22 [Science/Physics, Computer/Theory] UID:46030 Activity:nil |
3/20 E8 Lie group mapped: http://www.aimath.org/E8/E8release.txt http://www.aimath.org/E8 \_ "While many scientific projects involve processing large amounts of data, the E8 calculation is very different, as the size of the input is comparatively small, but the answer itself is enormous, and very dense." How is E8 significant in this sense? I can easily state a problem like "Find the first N prime numbers." and input N=1000000". Then the input is small and the answer is enormous and it involves lots of computation. Is that the same? \_ How about "what is the ratio of the circumference of a circle to it's diameter, to 10^8 places"? |
2007/3/1-3 [Computer/Theory] UID:45845 Activity:nil |
3/1 UW-Madison Number-theorists finally derive Ramanujan mock theta functions: http://www.physorg.com/news91813611.html \_ "What made mock theta functions all the more inscrutable was the fact that the first few pages of Ramanujan's letter were lost. Those pages may have contained more clues, ...... All those secrets died with Ramanujan just two months after he wrote the letter," This sounds so much like Fermat's Last Theorem. Coincidence, or by intention? This sounds so much like Fermat's Last Theorem. Coincidence, or on purpose? |
2007/2/25-3/1 [Computer/Theory] UID:45816 Activity:moderate |
2/24 For those w/ a math interest KCSM (Channel 17/43.1) in the Bay Area is airing a biography of Paul Erdos this Wednesday, 28 Feb. \_ After the program, I dans@soda, will present a comprehensive critique here in the soda motd. \_ Really? I mean, Paul Erdos rocks and all, but, if you're tuning in to the program, you already know that. \_ Really? I mean, Paul Erdos rocks and all, but, if you're tuning in to the program, you already know that. P.S. Don't impersonate people on the motd. It's lame. And if you're going to do it, at least get the details right. -dans \- VFR Jones, of the "Jones Polynomial" fame is giving a talk on March 7 5pm at the Bancroft Hotel. dans will follow it up with a talk on the HOMFLY polynomial. On a different subject, Martin Jay, will be giving a talk on Lying in Politics [he's not going to say what you might expect, unless maybe you know him and read his LRB? essay on this] on Feb 28 5pm Bancroft Hotel. Martin Jay is not a normal person ... I think he has a photographic memory. \_ Heh. I took Math 1B from Jones. Teaching lower-division math classes made him *pissed*. -dans \_ You really enjoy making threads tack so they are about you, dont you? \_ So then don't become a professor. Use your super math skills in industry. People who become profs and then get mad when they have to teach are idiots. I'm sure even his dumb ass had to take 1B or equivalent once upon a time and someone taught it to him. \_ Don't hate the player, hate the game: Berkeley math doesn't even ask about teaching on their faculty job applications. The department doesn't care, so why should individual faculty? \- what is ironic is one of the most notorious math profs, Hurricane Wu, cares deeply about teaching ... and even writes papers on how to teach math. ob: http://soda.berkeley.edu/~echeng/hurricane_wu \_ Dude, were you abused by a theoretical math professor? I mean, Vaughn Jones is a Fields Medalist, somehow I suspect the satisfaction and rewards with being, literally, one of the worlds greatest mathematicians probably outweighs the annoyance of having to teach Math 1B once every few years. -dans \_ And then these sorts typically take out their frustration and annoyance on their undergrads which is no good for anyone involved. They should just hire non-prof. teachers to take care of the lower division stuff. \_ I think you are mis-diagnosing the problem. It's not that 1a/b are elementary, it's that they are not really math classes in the sense that mathematicians practice math. These classes exist to teach scientists and engineers very useful practical skills, none of which are relevant for a working mathematician. The way for them to be taught right is for people who acctually use calc to teach them, i.e. science and engineering professors. The reason this rarely happens is political: math departments get money based on the huge enrollment, and don't want to ever lose that money to physics, EE, or ME or whatever. At a school other than Berkeley, they actually had applied physics profs teaching the 50a/b series, and those sections were waaaay over enrolled, while only people with a schedule conflict would sign up with the math guy. \_ The CS department has introduced CS 70 as a replacement for Math 55 (formerly 50a/b) for basically this reason. The math dept. didn't care much for 55 since math people don't take it, so it was generally not well supported. It's really a class for CS types, so it was moved to CS. -gm \- curious point: among non-medical faculty VFR Jones is one of the highest paid profs in the uc system. it's a strange outlier. \_ He's a Field's Medalist for pete's sake. If that doesn't earn him a ridiculously high salary, what will? -dans \- dood, you have to stop assuming everybody else is stupid. uc has a bunch of other nobels and fields medalists and obviously i was noting VFRJ compared to them, not a lecturer in comp lit. it might be interesting to see TTAO's recent salary curve. in addition to prestige there is mkt price/op cost, and how much $$$ they bring into the university. e.g. kahan has a turing award but i bet patterson brings in more $$$. and i didnt say it was ridiculously high. i'm focusing on it's outliernss, not it's magnitude. \_ But he's a *field medalist*!!!@!1111 -!dans \_ Why do you think I'm assuming everyone else is stupid? Insecure, and maybe a little irrational possibly, but not stupid. -dans \_ See, this, right here, is why no one likes you. \_ No, that right there is what I do to prevent having to put up with fuckheads like you in my social circle. The precise term for this is "standards". -dans \_ Hint: you can't get out of the hole by continuing to dig. \_ Hint: I don't need the affirmation of anonymous motd hosers to feel secure in my sense of self. -dans We're gonna need a bigger shovel... _/ \_ I'll take backhoes for $200, Alex! -dans \_ HOly fucking crap I want to buy a backhoe for 200 dollars. \_ try ofarrell && larkin \_ Could you please provide more information? thanks! -guy who wants backhoe \_ The $200 back hoe's are usually Transvestites. \_ dans, a couple of us got a good laugh out of your use of "theoretical math professor". keep up the inadvertent humor. |
2006/11/11 [Reference/Religion, Computer/Theory, Computer/SW/Unix] UID:45329 Activity:high |
11/11 Curt looked at his erection, amazed. "I thought it had stopped earlier... My God, I'm huge!" Tentatively, he touched his engorged tip and patted it with his finger tips. It bobbed slightly as he stood up. Picking up a tape measure from the hall desk, Christa quickly came over to Curt and measured him. She let out a low whistle of appreciation at his new length as she wrapped the tape measure around his shaft. She looked up at him... "Curt, you've grown to eleven inches long and five-and-a-half inches around." She did some quick mental math. "Jesus, Curt. You're 1.7507 inches thick! Make sure you keep that thing the hell away from me." |
2006/11/2 [Computer/SW/Languages/Misc, Computer/Theory] UID:45090 Activity:high |
11/02 What is the best way to implement this? I have a queue where jobs are submitted for processing, and multiple worker threads asynchronously pull the front job from the queue to process it. I also have N (N=3) priority classes. Normally I would just implement this as N FIFO queues and that would be fine. But wait, there's more! Some jobs cannot be processed until other jobs (their dependancies) have been completed. A job and its dependancy may have different priorities. Then, certain jobs may require the exclusive use of a resource being used to process a different job in a different worker thread being fed from the same pool of jobs (think dining philosophers problem). Any ideas? \_ Classic priority inversion problem. \_ Is this your 162 homework assignment? \_ Nah, extending an existing system for work. -OP \_ This sounds amazingly similar to some of the job queueing stuff handled by Sun's N1 Grid Engine. You might be able to adapt it to your use, or if not, grab the source -- Sun publishes this -- and see how they solved the same problems. (We use N1GE and our users are pretty happy with it) -ERic \_ Grid computing? NOOOOOOOO!!!! \_ and in spite of the name, it is not really Grid computing, it is more of a batch job submission / processing engine. A great way to distribute tasks across many systems. SUN Marketing FTL -ERicx across many systems. SUN Marketing FTL -ERic \_ Here is a link to the source: http://gridengine.sunsource.net/servlets/ProjectSource GridEngine is a decent piece of code written by fairly good engineers. I worked on a interdepartment project w/ the N1GE team while at Sun and they were very nice responsive people. I used to work w/ the N1GE group when I was at Sun, if you need a contact, I can find out who is currently working on queue mgmt. \_ and in spite of the name, it is not really Grid computing, it is more of a batch job submission / processing engine. A great way to distribute tasks across many systems. SUN Marketing FTL -ERic \_ Would it be an acceptable solution to have multiple threads processing the jobs that have no dependencies, but have only one thread process jobs with dependencies? That thread could then do a topological sort on the elements in its queue when it goes to read from the queue. I guess after sorting, the job entire dependency graph could be handed to a worker thread... - ciyer \_ There may be a high fraction of jobs with (relatively simple) dependancy graphs. I'm thinking when a job with dependancies gets queued, it's dependancies can get (upwards only) priority inheritance. |
2006/10/26-30 [Computer/Theory] UID:44999 Activity:moderate |
10/26 I'm finding OSC's anti-homework rants fairly amusing http://www.ornery.org/essays/warwatch/2006-09-17-1.html http://www.ornery.org/essays/warwatch/2006-09-24-1.html \_ I'm finding them insightful. Have you read the books he cites? \_ No, I haven't. I think I will, but I'm not in a huge hurry. (As my daughter won't be in school for another 2 years.) \_ Well, as an adult who went through the educational system, what do you think? I think homework has value. I am not sure that a 5 year old needs to do 4 hours per night, but this kook seems to say that it's not even valuable to a high school student and I strongly disagree with that. The other thing this putz needs to do is figure out what a private school is. Don't like the teachers? Switch schools. What the hell does he think a letter-writing campaign will do to a LAUSD teacher? \_ OSC doesn't say anything about letter-writing campaigns in the article, are you refering to something in the books? I haven't read those, and can't comment on them. In the second essay he covers what kind of HW is and isn't reasonable for HS students, and that sounds pretty right on. I really don't remember elementry well enough to say. I don't recall oppressive amounts of HW anyway. That doesn't mean it doesn't happen somewhere. He also points out that his daughter currently is not overburdend, so he doesn't need a private school. Wait, you didn't read the article, did you? -op \_ I read the article and in it he says that if your child's teacher is assigning more homework than you would like then you should start writing letters. It should be every parent's right to raise an idiot child, after all. In the second article, he talks about how people don't need basic math if they are going into the fields of English or PE. (Math is a subject that is homework-intensive.) What does it matter if the Chinese or Russians are better at math, he says. I can't really support that. He thinks it is okay to go gather leaves, but not to do math homework. The truth about most parents who think their kids have too much homework is that their kids just aren't very bright and that's why it takes them so long to do it. I have witnessed this first hand. \_ Yep, a lot of what he says is a little wacky, no argument there. However, he doesn't say non-science people don't need "basic math" he says the don't need 4 years of HS math (up to calculus.) I personally don't really agree, (sounds like the German system, which has well known defects) but you don't need to exaggerate. On the letter writing, he advocates 1 letter to the teacher with a cc to the principal, that's not a "campaign" that's documentation. I can't respond to your last sentence. Again, it differs by location. \_ Note that Math-up-to-Calculus is not required for students who do not intend to attend college. For those who do intend to go to college, I can't recommend math up to Calculus enough because: a) it exposes students to a field that they may find interesting enough to pursue, & b) it teaches the kind of discipline you need to do better than a C average in college (and get anywhere in life). \_ Field? You mean math? I would argue that you don't really get exposed to that "field" until you take something like 104. Maybe kids should take analysis in highschool? I was lucky enough to get exposed to things like analysis and differential geometry in highschool and did get hooked, but I don't think most highschool students who take up through calc get exposed to anything remoteley resembling real math as it's practiced by mathematicians. \_ They should have real analysis in highschool if the goal is to get kids hooked on math. Too bad most highschool math teachers are not qualified to teach that. \- i loved analysis [mostly because i like topology] but i dont agree with this at all. being able to think in terms of marginal rates of change, gradients, minimization and maximizing of functions is useful for many many problems, even if in just a conceptual sense where you arent actually doing the math ... and some things are probably EASIER to explain with elementary calculus without calculus [like a lot of mirco econ]. on the other hand, i dont think most people really need deep understanding of continuity, let alone more complicated stuff like compactness. for science people, understanding more about integration theory and measure isnt bad. but this is a much smaller group than the "practically everyone" who would benefit by learning some calculus. rather than analysis, i think i'd recommend a analysis, i'd recommend a class on "science that everybody should know" [DNA, gravity, a little atomic theory, conservation laws, evolution, some astronomy etc]. \_ Dude, I'm not seriously advocating forcing everyone in highschool to take analysis. I do think that in a perfect world, there would be an *opportunity* for highschool kids to get exposed to it if they're really into math. Even if it's just as simple as having a calc teacher that knows enough to hand their best, most bored students a copy of Rudin to read for fun. \_ Agreed! However, the same could be said of critical analysis of history and current events and the lack of qualified instructors. Cf. "Lies My Teacher Taught Me" \_ Is history addictive like math? I'm not sure. "Don't worry, kid. It's just a harmless little book with some funny symbols...and the first one's free"-- and then you're hooked! Before you know it, you're living in your car and ripping off people's TV sets to get money to buy your Springer-Verlag books. |
2006/10/7-10 [Computer/Theory] UID:44712 Activity:nil |
10/7 Did anyone take Math 128B (Numerical analysis II) last semester with Kahan? Can you comment on your experience, selection of topics, grading, etc? He's teaching it again next semester. \_ I'm sorry I believe you're in the wrong place. This is the Computer Science of the Ultrabored Alumni. We talk about optimal ways of wasting our employees' money on wall, motd, bitchx, and irc. |
2006/9/15-19 [Politics/Domestic/Crime, Computer/Theory] UID:44392 Activity:nil |
9/15 Math professor suspended over allegedly racist test question http://www.thefire.org/index.php/article/7275.html \_ "This punishment is not only unfair and a violation of the First Amendment, but also totally unnecessary." The questions you put on a test that you create as part of your paid employment is "free speech"? \_ the argument has been made that the gubment doesn't have a right to restrict speech, even if they are paying you. (This was a public institution), if it had bee a private school, it would have been different. \_ Aren't these FIRE guys part of that nutter Horowitz's group? |
2006/8/27-29 [Politics/Foreign/Asia/China, Computer/Theory] UID:44165 Activity:nil 80%like:44172 |
8/27 hey math fans read this: http://www.newyorker.com/fact/content/articles/060828fa_fact2 \_ Thanks! great article. -math fan |
2006/8/25-29 [Computer/Theory] UID:44156 Activity:moderate |
8/25 http://www.ucla.edu/about/faculty/tao.html \_ Sigh. Tao >>>>>>>>>> me. \_ "Such is Tao's reputation that mathematicians now compete to interest him in their problems, and he is becoming a kind of Mr. Fix-it for frustrated researchers. 'If you're stuck on a problem, then one way out is to interest Terence Tao'" \_ This is why I got out of math. No amount of hard work or slightly clever tricks or listening to the advice of smart old guys can come close to compensating for not being a genius in math. Why should I spend three years of my life working hard on a problem that some genius can solve in a couple hours? Other sciences are not like that. \- yeah, i know that feeling. years ago a friend of mine asked me about an "extra credit" problem for his 50a class. I couldnt believe I couldnt figure it out ... and started working on it with some other friends for a few days. We then started talking to Robin Hartshorne about it ... it was amazing the stuff he came up with in an hour. BTW, it turned out it was an open problem not an "extra credit" problem. an "extra credit" problem. It is interesting how many of these people are just not normal ... Solovay, for example. Another friend of mine reported a homeless man in the elevator to the math dept ... their reaction was "oh, that is Prof x". \_ Was he bald and reading minds? \_ Is there (and I really am just curious) something to be said about someone this young winning a Fields medal? If this is truly like the Nobel prize of math, shouldn't/aren't awards like this usually given to people who are older (i.e. more well established in the field, more published, etc.)? I'm just wondering if this is a subtle publicity stunt of sorts. But I do realize this kid is a math genius, too. \_ You have to be under 40 to win the Fields Medal. \_ OH, I didn't know that actually. Thanks. -op \_ OH, I didn't know that actually. Thanks. -pp \- note: that is "why" AWILES didnt get a FM for the FLT proof, but got a consolation prize of some lawn furniture. \- No, it's not a "stunt". I'm actually kinda surprised Tao is getting all this special attention. I think th point of the \- It's not a "stunt". I'm actually kinda surprised Tao is getting all this special attention. I think the point of the "under 40" rule is so that the prize doesnt become a "liefetime achievement award". Sort of the opposite of a Life Oscar ... who/how old was the youngest Life Oscar Winner? \_ But what if you do something great after 40? Wiles got screwed Don't the rules imply that if you don't do anything great b4 40, you won't. Wasn't it RIBET's result that turned Wiles onto Fermat? He must have forgotten to check how old he was. \- Wiles hardly got screwed ... and we only think in those terms because he coincidentally missed by only a couple of years [would you have said "he go screwed" if he did the work when he was 50?]. While WILES isnt exactly a \_ Yes. There should be a Nobel or equiv for Math. household name, you have heard of him ... how many Fields Medalists have you heard of? Did HAWKING get screwed because he hasnt won a Physics nobel? It's not like WILES would have been able to get a better academic position if he won a FM. In some cases the person is honored by the prize, in other cases the prize is honored by them ... do yu think VS Naipual readership changed after he won the lit Nobel? \_ I think most people in the physics department believe Erwin Hahn got screwed for not getting it for spin echo. \- i think there are probably more valid cases of screwage when person A B C D are involved in idea X, but only A B C win an award rather than nobdy getting an award for some idea. I guess the famous case of this is RFRANKLIN in the DNA episode [nobel can only be split up to 3ways], \_ also can't be given to dead people. which she was. and I dunno how valid that case is, but I suppose CHIEN-SHIUNG WU [UCB Physics PhD ... in 1940!] has a pretty good case in the YANG-LEE PARITY case. As for UCB Physics, do people think ZUMINO has been screwed? \- BTW, if you are interested in a somewhat interesting take on "math culture", see: http://www.dpmms.cam.ac.uk/~wtg10/2cultures.pdf It's by Timothy GOWERS, who is Yet Another Fields Medalist [and was a student of BELA BOLLOBAS, whom some of you may know, if you have studied GRAPH THEORY]. GOWERS divides mathematicians into "problem solvers" and theory builders ... I think generally people outside the math community -- but who have studied some math -- dont hear hear much about people working on inscrutable, hard problems. [ObTrivia: GOWERS won the Fields medal the same year as CURT MCMULLEN, who used to be at Berkeley but was stolen by Harvard and is kind of a jerk, and BORCHARDS, who is at Berkeley still and works on MOONSHINE, and had brain analyzed by ALI G's cousin, who is a famous research psychologist]. |
2006/8/22-24 [Computer/Theory] UID:44100 Activity:kinda low |
8/22 Someone who looks like ilyas solves historic math problem, shuns prize http://www.cnn.com/2006/TECH/science/08/22/math.genius.ap \- Also headline story on THE REPORT OF COLBERT. Probably on You tube soon. \_ It's on the Comedy Central cite -- brilliant! \- Hello, for those of you into school pride, Berkeley is a big deal in this area of math. PERELMAN was operating out of UCB briefly. The pf POINCARE and THURSTON GEOMETRIZATION is a really big deal ... note also THURSTON also used to operate out of UCB. And SMALE was one of UCB's older Fields Medalists, who earlier had an important result related to POINCARE. Of course Old Professor CHERN is super-important ... who taught here for like 30yrs and started MSRI. [BTW, UCLA owns one of the other Fields Medalists ... and another one used to teach here ... Of course Old Professor CHERN is super-important ... also UCB prof and started MSRI. [BTW, UCLA owns one of the other Fields Medalists ... and another one used to teach here ... but it looks like he was procured by Princeton since, so I guess we dont lose a parking spot next to Evans]. Here are two other really weird math people: http://en.wikipedia.org/wiki/Grothendieck ... stopped doing math and moved to Andorra. http://bookbuzz.com/MBIO_About_Erdos.htm ... who was a borderline homeless speed freak. I once touched PERELMAN. --lewis@soda I once touched PERELMAN. Feel free to send me your questions on low-dimensional topology. --lewis@@soda \_ That guy doesn't look remotely like ilyas. \_ He does however look a lot like Prof. George Bergman. |
2006/8/15-17 [Computer/Theory] UID:44010 Activity:nil |
8/15 Dubya's speaking skills: 10 years ago and now ... If he was a racehorse I think he would have been put out of his misery a long time ago ... http://www.informationclearinghouse.info/video1019.htm \_ A mexican friend of mine claims that Fox's command of Spanish is even worse than Bush's command of English. My friend speaks good enough English, in my opinion, to be able to fairly compare the two(his English is much better than Bush's). Can any Spanish speakers confirm or deny his claim? \_ Maybe it's the overhyping of the video captions, but I just don't see it as that big a deal. \_ Yeah, since when were communication skills and command of the English language a relevant part of the job of being President of the United States? \_ yeah, I caught that was/were eror right after checking \_ yeah, I caught that was/were eror right after checkingu \_ yeah, I caught that was/were error right after checking in the motd \_ Um, not a big deal? That he could speak in complete and COMPOUND sentences and 10 years later can barely utter two words without tripping up? \_ Uh, so? 10 years ago I knew how to do Calculus. Now I don't. Also, running the state of Texas is a much simpler job than running the entire US. \_ If doing Calculus was vital to doing your job, then it would be bad if you no longer knew how to do it. Are you arguing that public speaking is not an important skill to have when you are President of the United States? \_ Yeah, no shit. He has less public speaking ability than the average PTA member. \_ Uh, so? 10 years ago I knew how to do Calculus. Now I don't. Also, running the state of Texas is a much simpler job than running the entire US. \_ The 10-year-old clip looks like he's reading off a teleprompter. -tom \_ I don't think he's reading off a teleprompter. He's looking away from the camera probably because he's addressing the moder- ator. If he were reading off a teleprompter he'd be looking directly at the camera. \_ So what. I've heard worse. Since when was proper syntax and grammar a requirement for holding office? -Arnold Schwarzeneggar \_ Because the Governator needs to be Recalled. |
2006/8/9 [Computer/Theory, Computer/SW/Security] UID:43952 Activity:nil |
8/9 Can someone update soda's ssh host keys on http://www.csua.berkeley.edu/computing/hardware I think the new keys are: RSA - 9c:a4:3a:66:23:22:b0:2f:ba:87:2a:ca:03:c5:24:b6 DSA - 93:1d:30:88:65:a5:fa:38:6f:06:a3:86:12:0d:85:8b \_ That's what you'd like us to believe. |
2006/4/2 [Politics/Foreign/Asia/Korea, Computer/Theory] UID:42614 Activity:nil 80%like:42612 |
4/2 http://www.engadget.com/2006/04/02/south-korea-wants-100-robot-market-penetration-by-2020 South Korea Wants 100% Robot Penetration by 2020 |
2006/4/2-4 [Politics/Foreign/Asia/Korea, Computer/Theory] UID:42612 Activity:nil 80%like:42614 |
4/2 http://tinyurl.com/juoet (engadget.com) South Korea Wants 100% Robot Penetration by 2020 \_ yermom prefers the human touch. |
2006/3/26-27 [Computer/Theory] UID:42439 Activity:nil |
3/25 National Post on Noam Chomsky: http://tinyurl.com/rwteu (canada.com/nationalpost) \_ So he is a smart investor, too. My respect for him grows. |
2006/3/25-27 [Computer/SW/Languages/C_Cplusplus, Computer/Theory, Computer/SW/Unix] UID:42432 Activity:nil |
3/25 The Ultimate Casio Watch: http://www.watchreport.com/2006/03/review_of_the_m.html \- i used to have an older incarnation of the Pathfinder [till it fell off at Mughal Sarai Junction] and a couple of comments: 1. it is good they have changed this to a strap rather than a bracelet design. bracelet not good for these kinds of watches. i nearly lost mine after a b'let failure on a catemaranin 9ft seas, which is not where you want to try and fix something with <1mm springs. 2. i find the themometer of limited use, since while on your wrist you are really measuing the "temperature near you wrist". 3. if you really need a compass, you need to take a real orienteering compass ... these watch compasses are sort of just opportunistic things. 4. some of the featuers like altitude alarm, altitude recording, multiple alarms i dont find that useful [you cant upload the telementry data to a computer can you ... GPS mre useful in this area] and these things tend to make accessing the frequently used function slower and more complicated (like say converting between ft <-> m ... which was my main complaint with my old watch 5. often you can get an older model for vastly cheaper by just sacrificing some function you may not really need [you can get some older p'finder models for around $85 vs $300+ for this one, which is significant if you think of this as a piece of gear rather than your watch]. thanks for posting the link. i am happy to give advice on outdoor gear except for aid climbing and snow and ice. --psb \- i used to have an older incarnation of the Pathfinder [till it fell off at Mughal Sarai Junction] and a couple of comments: 1. it is good they have changed this to a strap and not a bracelet design. bracelet not good for these kinds of watches ... nearly lost mine after a b'let failure on a catemaran ... which is not where you want to try and fix something with <1mm springs. 2. i find the themometer of limited use, since while on your wrist you are really measuing the "temperature near you wrist". 3. if you really need a compass, you need to take a real orienteering compass ... these watch compasses are sort of just opportunistic things. 4. some of the featuers like altitude alarm, altitude recording, multiple alarms i dont find that useful [you cant upload the telementry data to a computer can you ... GPS mre useful in this area] and these things tend to make accessing the frequently used function slower and more complicated (like say converting between ft <-> m ... which was my main complaint with my old watch 5. often you can get an older model for vastly cheaper by just sacrificing some function you may not really need [you can get some older p'finder models for around $85 vs $300+ for this one, which is significant if you think of this as a piece of gear rather than your watch]. thanks for posting the link. i am happy to give advice on outdoor gear except for aid climbing and snow and ice. \_ I wanted to get this watch b/c it was the first casio I've seen that has world atomic timekeeping, is solar powered, water resistant and it has that "indiglo"-like lighting. The compass was an added bonus, but I usually carry my eTrex GPS so its not an essential for me. |
2006/2/22-27 [Computer/Theory] UID:41957 Activity:nil |
2/22 Quantum computer works best switched off -http://csua.org/u/f25 |
2006/1/26-29 [Politics/Domestic/Immigration, Computer/Theory] UID:41560 Activity:nil |
1/26 Hello realkitty: http://news.yahoo.com/s/nm/20060126/od_nm/japan_robot_dc \_ People have been predicting this trend for years. Given Japan's generally anti-immigration stance and its soon-to-be-rapidly-shrinking population, manpower will be quite scarce. That combined with a culture that's accepting of (or even enamored with) robots and seeing them soon in day-to-day jobs seems inevitable. |
2006/1/23-24 [Computer/Theory, Computer/Domains] UID:41481 Activity:nil |
1/21 The authenticity of host 'soda.berkeley.edu (128.32.112.233)' can't be established. RSA key fingerprint is e1:9c:e5:c7:f9:9f:f3:af:04:ef:df:2d:63:b0:84:4a. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added 'soda.berkeley.edu' (RSA) to the list of known hosts. Password: soda [1] Errr, wasn't that domain name supposed to stop working like, a decade ago? \_ Why would it? -tom \_ Because for months/years they've been insisting they're going to stop hosting most top-level berkeley hostnames in DNS. \_ Who has been insisting that? Certainly not CNS. -tom \_ regime change |
2006/1/21 [Computer/Theory, Computer/SW/Unix] UID:41467 Activity:nil |
1/21 Is there a good way to write a log file in a multi-threaded system that doesn't require locks? It seems like there should be some sort of atomic fsync() or something. |
2006/1/4-6 [Computer/Theory, Computer/SW/Security] UID:41226 Activity:nil |
1/4 "Mo. Researchers Find Largest Prime Number" http://news.yahoo.com/s/ap/20060104/ap_on_sc/largest_prime_number Why are people interested in finding large prime numbers? They already know that there are infinte number of primes, so what's the point of finding them? \_ because they are there. finding more may help with proving (or disproving) conjectures about dist. of primes, etc \_ You know that prime numbers have a lot to do with public key cryptography right? \_ Yeah, but with a prime as large as 30 million bits? \_ This is usually tangential to burning in a new supercomputer. They let it sit there and compute prime for a bit. As computers get ever faster, they find new primes and it generates a little PR for the guys running the new computer. At least this is how most of these ginormous primes are discovered. \_ Learning how to work with large primes has value. We used to compute pi to billions of digits. Now we test primes. \_ This particular project is more like SETI-at-home and is validating a s/w concept re: distributed computing. Lots of these primes are incidental discoveries. \_ This is usually tangential to burning in a new supercomputer. They let it sit there and compute prime for a bit. As computers get ever faster, they find new primes and it generates a little PR for the guys running the new computer. At least this is how most of these ginormous primes are discovered. |
2005/11/23-26 [Politics/Foreign, Computer/Theory] UID:40719 Activity:nil |
11/23 http://news.yahoo.com/s/ap/20051123/ap_on_re_us/southern_identity_accents Ooooh I've always had uninflected tone that oozes authority and refinement and I didn't even know it. Maybe I'll impress hot southern belle when I go to the south. \_ why the sarcasm? are you telling me you can take people with southern accents seriously? |
2005/11/7-8 [Computer/Theory] UID:40473 Activity:kinda low |
11/7 Re: "Teaser: write an algorithm for finding prime numbers." on 11/1, I wrote some C code to find all prime numbers between 1 and 4294967295 with 32-bit math (no 64-bit or floating-point math). It took about 84 hours on a 502MHz Sun-Blade-100. \_ Quickly thought through, totally untested alg. Make a list of (n, n^2) tuples for all n < 2^16 and n is prime. For all numbers from 2^16 to 2^32 go through the list if n^2 > number you are prime, break. if number % n = 0 you are not prime. \_ That's more or less what I do in my 13-KByte 83-hour algorithm, except that I only keep a list of n instead of (n, n^2). I'll try the sieve method. --- OP \_ So I ran: % nice +20 primes 1 4294967295 > primes.txt on my 800MHz Athlon. It took 8 minutes on the nose. % wc -l primes.txt 203280221 primes.txt --dbushong \_ Wow! Must be a better algorithm. Where's the source code? Better yet, what's the algorithm? Thx. -- OP \_ Pretty standard sieve of eratosthenes: http://www.freebsd.org/cgi/cvsweb.cgi/src/games/primes --dbushong \_ What's the memory usage? Mine is ~26KB. -- OP \_ What's the memory usage? Mine is ~13KB. -- OP \_ IIRC, it was a few megs --dbushong \_ Did you go from 1 to N or 1 to sqrt(n). That would speed things up. \_ I go from 1 to sqrt(n). -- OP \_ are you avoiding even #'s? how about factors of 3, 5, 7, etc. that's basically what "primes" is doing. \_ but there are easier tests for divisibility for smaller primes than there are for larger ones. also, a test for divisibility by 2, 3, 5, 7 and other smaller primes do more to knock out parts of the field. |
2005/11/1-2 [Computer/Theory, Computer/SW/Languages] UID:40394 Activity:high |
11/1 Teaser: write an algorithm for finding prime numbers. \_ boolean isPrime(int x) { return true; }. It correctly identifies any prime number as being prime. And boy is it fast. For non-prime number use boolean isPrime(int x) { return false; } which works JUST AS FAST! Note that sometimes users may have problems with the package because they use the wrong function for a given piece of data. Morons. \_ That doesn't answer the question. Try this instead: int findPrime() { return 37;}; \_ How about an algorithm for getting other people to do your homework for you? \_ Here's a homework hint: go look up the strict definition of "algorithm". \_ Al Gore does more of that bizarre swaying thing. I don't think he has much rhythm at all. \_ Are you new to math or new to programming? \_ Are you new to asking homework questions on the motd? |
2005/11/1-3 [Computer/Theory, Politics/Domestic/President] UID:40385 Activity:nil |
11/1 Anyone watch Prison Break? I caught a few episodes but I have no idea what this 'fibonacci' thing is people are looking for. \_ According to a quick google search, Fibonacci is an informant. ... a person, by the way... \_ http://www.entertainyourbrain.com/prisonbreakrev.htm \_ F(0) = F(1) = 1, F(i) = F(i-1) + F(i-2) for i > 1. |
2005/10/18-19 [Science/GlobalWarming, Computer/Theory] UID:40153 Activity:nil |
10/18 http://www.livejournal.com/users/dpodbori/1369.html ON DANGERS OF BEING AN INSECT WITH WINGS AND A MYSTERIOUS INSTANCE OF MASS MAILING \_ Why was this posted? Is it supposed to be funny? \_ No it's not funny. |
2005/10/14 [Computer/Theory, Computer/SW/Unix] UID:40092 Activity:nil |
10/14 If only we had some sort of forum to discuss having some sort of forum so we could disc---... heh sorry. :-) |
2005/9/20-21 [Computer/Theory] UID:39764 Activity:nil |
9/20 UCB professors win 3 (of 25) MacArthur prizes: http://tinyurl.com/bgtq4 \_ I've always been impressed by Berkeley'd research and the brilliant people behind the research. Too bad most of them don't seem to give a damn about teaching or helping undergrads. Funny that great researchers aren't always great teachers. My experience shows quite the opposite. -grad student \_ This is not a coincidence. Tenure decisions are based mostly on research, and at some schools they're based 100% on research. Doing the quality of research that gets you tenure at a place like Berkeley simply does not leave any time to be a decent teacher or to care about undergrads. I know that my advisor is a great teacher from one-on-one situations where he is explaining stuff to grad students, but all the undergrads I've talked to say he totally sucks as a teacher. He shows up for his classes sleep-deprived and barely prepared, and copies his notes onto the board, with errors, then gives a couple really hard exams and calls it teaching. But he got tenure, and does outstanding research. Don't hate the player, hate the game. outstanding research. Don't hate the player, hate the gnome. \_ I agree. I had a professor in the Math Dept. who sucked when I took Math 50B. When I had him later in Math 185 the class was small and I was often the only one in office hours. He was much better then, although still a bit impatient. \_ My advisor tells me not to worry about TAing or teaching. Just concentrate on my research. Research gets you everything, teaching gets you nothing. This is coming out from a well respected guy in Berkeley. -anon grad \_ MacArthurs analyzed. http://www.slate.com/id/85794 \_ This is hilarious, THANK YOU. \_ BRILLANT! |
2005/9/15 [Computer/SW/Apps, Computer/Theory, Computer/SW/Unix] UID:39696 Activity:nil |
9/15 I have a stupid Excel question. It's easy to obtain the product of any two cells, but what is the syntax to specify that I want the product of, say, the previous two cells? That is, if I am in row A I want the product of A1 and A2 entered into A3, in row B I want the product of B1 and B2 entered into B3, and so on - sort of "relative to my position I want to do this". The rows can change lettering as things are inserted or deleted, but the idea is that I always want, say, column 1 and 2 multiplied into column 3 of the same row. How do I do this? I've RTFM. |
2005/9/14-17 [Science/GlobalWarming, Computer/Theory] UID:39673 Activity:nil |
9/14 Cognitive science searches for a common morality http://bostonreview.net/BR30.5/saxe.html |
2005/8/28-29 [Computer/Theory] UID:39317 Activity:moderate |
8/28 Did any of you take Stat 205A? Do you have to know measure theory in advance? \_ I have taken a UCLA equivalent. Usually people recognize a lot of engineering folks want to take such a class, so the usually teach it in a way that doesn't require you attenting a measure theory class (they review the measure theory that's needed). I wouldn't be surprised if they do this at Berkeley also. -- ilyas \- "know measure theory" can "cover a lot of area" ... this may be easier to answer if you mention something about your back- ground. "the only upper div math class i have taken is 104" vs. "i have done lots of math, just not that much analysis/ integration theory etc." i had a good background in measure theory but not much background in stat so i had to relearn a lot of vocab [e.g. the "mapping" from the lebesgue density lot of vocab [like the "mapping" from the lebesgue density ` thm to the martingale density thm]. BTW, if you dont have a Giant Math Brain, one thing about grad math/physics classes: often people sort of work sort of collaboratively so it is helpful if your friends or officemate is in the class ... having to do 20hr problem set by yourself can really suck ass. --psb@stat.berkeley.edu \_ I did applied math as undergrad and took many upper div stat classes but haven't done much real analysis beyond math 104. -op \-i'd say go for it. a lot of applied stat people dont have the greatest background in the deep math of sigma algebras. \_ I will. I am still worried though that prof. Pitman tried to discourage me by saying that ppl who haven't had measure theory usually struggly in this course. had measure theory usually strugle in this course. \- a friend of mine [ucb stat phd, now teaches biostat] came in to stat from EE so she took 104 and 105. if you have taken a lot of math classes i think that more than offsets doing just 105 ... what class is pitman recommending anyway? 105 is the only one that seems to fit the niche. he isnt saying you need 202ab is he. of course i used to get phone calls at 3am from this friend asking "what is a banach space" ... need 202ab is he. of course i used to get 3am phone calls from this friend asking "what is a banach space" ... actually come to think of it i also got a 3am phone call from her because she "heard a strange sound" and was hiding in her closet, but that is another matter. \_ He says it's ok to take 205A now but ideally everyone should take 105 before 205A. Another possibility is to take the new Stat 204 which sounds like Stat 205A 'lite'. \- i knew almost nobody who took 105 ... it isnt always offered even is it? i think you figure it all out in 202ab ... doesnt the ROYDEN book cost like $5000 now? BTW, i think GREEN/PAPA RUDIN is a useful book to have around. The POWER of the RIESZ REPRESENTATION THEOREM. \_ Sounds like she was trying to pick up on you. I hope you banged her. |
2005/8/24-25 [Computer/Theory] UID:39243 Activity:high |
8/24 Computational Geometry question: I have a bunch of points which I use to generate a tiling of unit sqaure by Delaunay triangulation. Given an arbitrary point in the unit square, I want to determine which triangle the point is in. Does this problem have a name? I'm trying to find see what algorithms are out there for doing this, but my Google fu is weak. I'd appreciate any tips/links. By the way, go Republicans! We kick ass! -jblack \_ Yes, the problem is called 'point location.' You can adopt binary search for this problem without much difficulty. You can also use fast randomized algorithms for this problem, see: http://citeseer.ist.psu.edu/cke96fast.html -- ilyas \_ Yeah, binary search was my intuition, I just wanted to confirm that. Thanks for the info / link. - op \_ Without actually computing the triangulation? \_ jblack, you're still on probation. I'm letting you post this time, since you signed your name. \_ Let me get this straight. First you harrass us by constantly posting freeper crap, hiding URL by using IPs, mass posting, insulting and provoking liberals, hiding your identity, so on and so forth even though we asked you to stop, and then you have the nerve to come back to us and ask for help? HELLO??? Either you're an autistic socially inept retard, or you simply need help. SERIOUS help, from a good psychiatric institution. \_ While I can certainly appreciate your zeal, I'd rather not be included in your inclusive 'we' and 'us'. While I don't agree with jblack's politics, he has a right to speak his agree with danh's politics, he has a right to speak his peace. On the other hand, I'm absolutely not going to back up anyone that espouses a philosophy of censorship, and would rather not be implicitly or explicitly represented by anyone that does. -mice \_ I find it amusing that you're not signing your name. -emarkp \_ also, you've been trolled. \_ Dude, you're a damn retard. Do you seriously not see the irony here? You're the one looking unstable. \_ Pot. Kettle. Etc. \_ Haha. This response is truly stupid. \_ This whole thread is stupid. \_ This whole motd is stupid. \_ Stupid is as stupid does. |
2005/7/21-23 [Computer/Theory] UID:38752 Activity:moderate |
7/21 Translated version of Aristotle's Physics: http://etext.library.adelaide.edu.au/a/aristotle/a8ph \_ This is what the kids should be learnen' in school! \_ I know you are making a joke, but in many ways Aristotle's ideas re Physics formed the basis for Newtonian Physics (iirc Aristotle came up w/ the idea of inertia, &c.) so teaching children where modern science came from would be a good thing. \- i think teaching science is more important than where science came from. i thought the st. john's program where you read principia and such was sort of weird. the ancient notations are not reasonable. \_ I agree that teaching science is the most important thing, but many students learn things better if they can follow the historical development of an idea. I also think that the LACK of historical knowledge is one of the biggest problems in this country. \_ One of the huge problems with presenting modern mathematics is a culturally ingrained insistence, among mathematicians, to present the completed edifice of knowledge. Personally, I find math a lot easier to understand if one follows the history of the field rather than the conceptual map of the modern state of the field. -- ilyas modern state of the field. Calculus, for instance, makes a lot more sense when put in context of the physics people were trying to solve by developing infinitesimals. Similarly, group theory was developed to find polynomial roots originally, etc. -- ilyas \_ I think the claim that that will help the average grade school student is ridiculous. The point of a math class at that level is to teach a *skill*, not a set of facts. I think a good math class at the gradeschool level should be closer to a sports class than a history class. Should baseball players be forced to learn cricket before they play real baseball? Besides, neither the math nor the physics taught in school is a "conceptual map of the modern state of the field". Math stops at Newton and physics never goes past the 1930's. I'm writing this rant because I think fucking with the math curriculum is very dangerous. I know people who were in school in the "New Math" era, and most of them didn't learn *anything*. And now there are a bunch of idiots trying to get physics turned into some math-free freshman fluff class so that highschool students can follow the "logical" progression from physics to chemistry to biology. If you haven't heard of it, look up the "physics first" movement. Idiots. \_ Look, I'm not saying that we replace phyiscs/math w/ history lectures. Rather what I'm saying is that when you teach a particular theory, just give 5 mins about where it came from. Some kids will just get it better that way and the ones that really care (yes, some of us NERDS actually did outside unassigned reading for FUN) will go out and read more aobut it. more about it. \_ Fine. I have no problem with history of science as supplementary material, and yes, I spent quite a bit of time reading that sort of thing when I was in school as well. I was addressing the issues of re-aranging the order in which subjects are taught, and of teaching obsolete or otherwise irrelevent technical material. Also, the mainstream curriculum should not be primarily aimed at math nerds. It should be set up so a typical student can get competancy. \_ You are barking up the wrong tree. I am not a proponent of 'touchy-feely' education in the hard sciences. I think to understand what I am saying, you should show up for an upper division or graduate math class sometime. For the record, I think kids should learn what a proof is (and start doing proofs) at the age of 10. It boggles the mind that countries let citizens vote while not explaining to them what a coherent logical argument is. -- ilyas \_ Your patronizing idiocy is impressive, yet typical. I was a math major, have done math research, and tought myself calculus in 7th grade, so you can shove the attitude straight up your ass. What is good for a math major is *not* the same as what is good for insuring basic competence in math for the typical grade school student. \_ Do you even know what you are ranting about anymore? Did I insult your mathematical EPEEN, lafe? -- ilyas \_ I am confused. I was a math major and mathematicians (and physicists, chemists, psychologists, economists, and others) are fascinated with the history and usually make it part of their courses. Is your argument about grade school level classes, because college classes seem to already do what you are suggesting. --dim \_ If you mean a historical presentation of material, only 1 math teacher I had ever did this (my graduate logic instructor at UCLA). Most berkeley math professors sucked as teachers, and presented the material poorly. -- ilyas presented the material poorly. Btw, the UCLA guy's name is Itay Neeman. He is awesome. I recommend taking anything by him if you are ever at UCLA. -- ilyas \_ Perhaps your last statement is true, but it is irrelevant. A lot of classes in different subjects I had at Cal talked about Archimedes, Newton, Riemann, Euler, Descartes, Bohr, and so on. Since science is built on the foundations laid by others it is hard to present the material w/o mentioning these guys and their contributions. Perhaps I misunderstand what you mean exactly. --dim \_ My experience was very different. The only two teachers before Cal who talked about the history of science/math were my Jr HS Bio teacher and my Chem AP teacher. And they only briefly mentioned it. In Cal, only my 7c teacher talked about history and that was only bits about the A-Bomb. I had to take Math 160 (History of Math) to get any exposure to math/science history. My argument is that a historical perspective should be included at all levels. It shouldn't be at the expense of the material but rather as a way of supporting the material and perhaps making it a bit eaiser for students to understand. |
2005/7/19 [Computer/Theory, Computer/SW/Unix] UID:38706 Activity:nil |
7/19 For the betting types below: You could always use some sort of formulation of the bet that would result in a yes/no: eg: Rove is indicted by 12/31/2005, or Grand Jury evidence shows 1 witness reporting Rove "handled" the memo. \_ You mean something like this? http://csua.org/u/crr \_ error page \_ Fixed |
2005/7/16-19 [Computer/Theory, Computer/HW/CPU, Computer/SW/Apps/Media] UID:38658 Activity:nil |
7/16 Cellular Squirrel: http://web.media.mit.edu/~stefanm/phd/cellularsquirrel \_ What a waste of funding. It's no wonder some people are are skeptical of the "hi-tech" work that comes out of media labs. \_ C'mon, this is cool. I'd much rather have some dancing stuffed rodent than most of the obnoxious headwound-worthy ringtones you get on cell phones. -John \_ the japanese are way ahead of us on this. |
2005/7/12-13 [Consumer/Camera, Reference/Military, Computer/Theory] UID:38548 Activity:nil |
7/11 Just wanted to thank the person who posted about the PBS special Guns, Germs and Steel. Tonight's eps. (the 1st?) was pretty good. I'm not sure I completely agree w/ the premise that geography dictated "success", but I'm pretty sure I can go along w/ the fact that it predisposed certain people to create civilization and culture. \_ I can't believe I missed this book when I came out in 1999. I'm going to have to get it from the library after finals. BTW, he seems to have written another book last year called Collapse. Have you read that one? \_ I forgot to watch. The newest hardback edition of the book out now has a new chapter on the author's theory of the origin of the Japanese language and people. - danh \_ I can't believe I missed this book when I came out in 1999. I'm going to have to get it from the library after finals. BTW, he seems to have written another book last year called Collapse. Have you read that one? \_ I read the stuff at http://pbs.com, which seemed to contain the entire show content, and you can read it all in 5 minutes or whatever. I guess seeing the landscapes and people on camera would be fun. Oh well. \_ I watched it in HD and it looked great. \_ http://www.pbs.org not http://pbs.com. |
2005/7/7-9 [Computer/SW/Languages/C_Cplusplus, Computer/Theory] UID:38465 Activity:nil |
7/7 I never took AI but I'd like to learn something about Bayesian belief, inference, etc. Something like a 10 page intro PDF or a URL would be useful. Recommendation? \_ http://www.cs.ubc.ca/~murphyk/Bayes/bayes.html Kevin is a former Cal grad student, btw. Very smart guy. -- ilyas \- how do you know kevin? http://home.lbl.gov:8080/~psb/ASSOCIATES/Mike+Kevin81.jpg He actually fits into this story: *Boredcast Message from 'psb': Mon Aug 4 15:22:19 2003 ||let's see: A is hosting B for a few days ... B ||who used to date C who is marrying D who was ||interested in A and was told by B to go out with ||A. D was the roomate of the office mate of E who ||used to date X and later dated F who got ||together with B. ||A = psb ||X = rob pike I note in passing B also works on graphical models and was formerly trafficking with various UCLA and Harvard and Berkeley Bayesians. \_ I worked on Kevin's toolbox a long long time ago. We took some classes together also. Kevin interviewed at UCLA for a faculty position, but didn't take it. -- ilyas \_ spasibo ilya! That is exactly what I was looking for. You are a very cool guy. Why do people make fun of you? \_ Probably because I am considered the motd retard. -- ilyas |
2005/7/5 [Computer/Theory, Computer/SW] UID:38412 Activity:low |
7/5 Time to kill your TV: http://news.yahoo.com/s/nm/health_television_dc \_ TV makes you stupid? Who woulda thought? \_ I thought watching Barnie on Ch 9 help toddlers develop, so I regularly let my son watch it. No? \_ No. \_ His future is now ruined. \_ Good grief, Barney is one of the worst things out there. Talk about brain rot. PBS has pretty much lost all good children programming (even Sesame Street is terrible these days). I'd recomment the Noggin channel instead. \_ You're that old and you still believe in purple dinosaurs? |
2005/6/20 [Politics/Foreign/MiddleEast/Iraq, Computer/Theory] UID:38210 Activity:nil |
6/20 In the encyclopedia under "cognitive dissonance" it says "see emarkp and the below Lancet threads." \_ d00d, cognitive dissonance is "no doubt" there were WMDs to: uh, why can't we find any? oh yeah, because they're in syria or buried in the desert. |
2005/5/25-26 [Computer/Theory] UID:37833 Activity:nil |
5/25 The latest on efforts to decipher the Archimedes palimpsest: http://tinyurl.com/9feoa \_ So if we discover that Archimedes discovers something before someone else, are we obligated to rename theorems? Like instead of the Turing Theorem (or stick in whatever name), we'll have to call it the Archimedes Theorem? \_ Probably not. However, Archimedes seems to have alternate (some would say better) derivations of many modern theorems so maybe these will be taught in schools. It is also possible that the palimpsest contains theorems that are completely unknown to modern mathematics (considering Archimedes seems to have known about integral calculus ~ 2000 yrs before Newton and Liebniz). \_ Probably not. We still celebrate Columbus Day after we discover that Columbus wasn't the first (nor the second) to discover the New World. |
2005/5/20-23 [Computer/Theory] UID:37793 Activity:nil |
5/20 I flunked calculus and now I'm struggling in my network class. In the lecture, the professor says p is the probability that a node in ethernet sends, N is the number of nodes, and the following is the slotted Aloha efficiency: N*p*(1-p)^(N-1) = ~1/e = 0.37 I completely understand how he setup the equation, but got lost trying to derive it. He said it's junior high math and hand waived it, but I'd like to know how you do this. Any pointer to old calculus stuff would be appreciated, thanks. \_ http://robotics.eecs.berkeley.edu/~wlr/12203/ethernet%201.pdf Ok fine, the slide is a mistake and he forgot lim \_ Also, it's (1-p)^(N-1) in the slide. And he didn't forget lim. It's an approximation. As N increases, S approaches 1/e. His notation's a little strange, but taking the lim would be even stranger. \_ As stated, this equation is simply untrue (what if N = 100?). -- ilyas \_ http://robotics.eecs.berkeley.edu/~wlr/12203/ethernet%201.pdf Ok fine, the slide is a mistake and he forgot lim \_ Also, it's (1-p)^(N-1) in the slide. \_ ... yeah. Kind of important. -- ilyas And he didn't forget lim. It's an approximation. As N increases, S approaches 1/e. His notation's a little strange, but taking the lim would be even stranger. |
2005/5/15-16 [Computer/Theory] UID:37696 Activity:moderate |
5/14 I just flunked my theory of computation midterm. I'm a bit baffled at one of the questions that I got wrong. According to the answer key, L is finite state, but I'm pretty sure it's not. L={uvu^R | u,v in {0,1}*, |v| is even} So I see that u and u^R are reverse, which already tells me that this is like Leq (need counting), hence not finite state. Is the answer key wrong, or I'm missing something obvious and stupid? \_ u^R means u backwards, right? If so, any even-length string is in the language (let u be empty, and put the whole string in v). No odd-length strings are in the language, because |v| would have to be odd. So L is just the language of even-length strings. --mconst \_ Thanks, yes L must be even. However, FS means that I'm able to build a machine with finite states, which means I should be able to represent it using regular expression. However, I can't possibly represent uvu^R because I need a push-down automaton to count for the reverse, so the answer should be be that L is NOT finite state, right? \_ /csua/tmp/regular-language --mconst \_ How much space do you need to tell if a string has an even number of characters? -- ilyas \_ Yes you're right, it is L={v | |v| is even}. I get confused by that u^R thinking the string MUST conform to that, but when in fact, it doesn't have to. \_ If your getting confused by uvu^R thinking you have to check to make sure u is reversed with u^R, think of this: Let v' = uvu^R. Then consider u' = {} (and u^R = {}). Now u'v'u'^R is in the language because it was a string originally in the language (its just uvu^R). Maybe you can start to see that the language is all the even strings, because just consider both u and u^R to be the emptystring. So you just need to insure v is even (and hence can be recognized with a DFA.) -mrauser \_ Who takes midterms on May 14th? \_ Many people do. I had two on May 14th. \_ What kind of whacked quarter/semester system in that?! |
2005/4/28-30 [Science/Physics, Computer/Theory] UID:37415 Activity:nil |
04/28 Quantum Crypto for Video Conferencing: http://www.nature.com/news/2005/050425/full/050425-9.html \_ I think you don't need to crack every frame to steal secrets. If you can crack the audio stream and you can crack one video frame per two or three seconds, that's good enough. |
2005/2/23-24 [Computer/Theory] UID:36384 Activity:nil |
2/23 In the CLR book, chap 4.1-4.3, there are 3 basic ways of solving recurrence. The first way, 4.1, is the substitution method, which I think is straightforward and simple (albeit sometimes time consuming). The second way, 4.2, is the tree method, which I think is great, but a bit too informal as a proof. The last one, 4.3, master's theorem, is GREAT. Now suppose you're taking a test and forgot your master's theorem, can you actually get prove everything using 4.1? In another word, are there recurrences that can't be derived using 4.1, but can using 4.2 or 4.3? ok thx \_ Hey are you the 'contiguous sequence sum' guy? Why do you feel the need to get the motd to solve your 170 for you? \_ this is kind of embarrassing but it's for the quals and since I haven't done this in ages (several years) I've pretty much forgotten everything. \_ Isn't this the case with math in general? I am having a hard time remembering some things if I haven't used them for about year, although the knowledge seems to come back quicker than when learning things for the first time. \_ Just curious, which class is this for? |
2005/2/21-22 [Recreation/Humor, Computer/Theory] UID:36343 Activity:nil Edit_by:auto |
2/21 Robot Chicken was one of the funniest shows I've seen on TV in a long, long time. Was that really ilyas in the third X-files screen test scene?? -meyers \_ can I download this somewhere? I don't have a TV... \_ http://www.mininova.org/tor/9884 |
2004/12/22 [Computer/Theory] UID:35392 Activity:nil |
12/22 Man builds Mech in backyard: http://tinyurl.com/4w6tq |
2004/11/6-7 [Computer/Theory] UID:34720 Activity:high |
11/5 So, I'm reading "Godel, Escher, Bach" and I came to a part I think I should be able to understand, but I can't. On page 137 Hofstadter introduces a recursive function G, such that G(n)=n-G(G(n-1)) for n > 0, and G(0) = 0; He then shows a graph that is supposed to corrospond to G, figure 30, on page 136. However, I just can't figure out how the graph is related to G. Can someone explain? BTW For G(1)-G(8) I get the values 1, 1, 2, 3, 3, 4, 4, 5. ok thx. \_ The graph shows the tree of recursive values. G(4) and G(5) = 3; G(6) and G(7) = 4, G(9) and G(10) = 6, G(14) and G(15) = 9, etc. -tom \_ Thanks tom. That's pretty neat. \_ Awwww, Tom answers a math and science question. |
2004/11/2 [Computer/Theory] UID:34533 Activity:nil |
11/2 Which is hotter? http://www.japantoday.com/e/?content=picture&id=1469 Girl/Child molester on left: .. Girl on right: \_ What? I see a picture of Ai Tominaga on our right and Hello Kitty on our left. \_ Improve brain, read more motd. \_ who? what? -brain \_ Sorry, please enlighten me. I still don't get it. \_ Degrade brain, read more motd. \_ *bzzzt* Illogical! Humor is Illogical! Positronic brain overload! WARNING! WARNI-- *KA-BLAM* |
2004/11/2 [Computer/Theory] UID:34525 Activity:low |
11/1 So for the hallway problem, I get: sqrt((x^(1/3)y^(2/3) + x)^2 + (x^(2/3)y^(1/3) + y)^2) I might have made a mistake in my calculations somewhere, but the calculus looks so ugly that I doubt it's anything simple like (x + y)sqrt(2) like someone else guessed. \_ Now that I think about it, I'm more confident in this answer. If x = y, it simplifies down to 2x*sqrt(2), which is what one would expect. \- that is correct, but my formula is nicer. the calculus looks like high school calculus ... although i used a computer to do some of the grunt math. but it's not like you need to use the airy function or anything like that. \_ The calculus is easy; it's just taking a derivative to find extrema. The algebra just makes it ugly, though. \- i used maple to do some of the math quickly ... max len is [X^(2/3) + Y(2^3)]^3/2, with reasonable assumptions. Solution avail at: link:csua.org/u/9rg --danh Harder problems is: unit board/plank vertically flush against a veritcal wall. bottom begins to slide away ... at what point does the tip lose contact with the wall. [from klepner and kolenkow]. \_ the problem is ill-stated. we need to know the height of the halls as well (as anyone who has actually moved a large object up an interior stairway and landing would tell you). \- obviously the question is being asked in 2-d with a zero width rigid board. [ MARS, bitch. ] \_ I'm Rick James, bitch. \_ I'm rich, bitch. \_ I'm dead, bitch. \_ What, me dead? -Alfred E. Neuman \_ That's not obvious at all. Anyone care to solve the 3d case? I'm reminded of the sofa simulation in Dirk Gently's Holistic Detective Agency. \_ figure out how far you can tilt the board, finds its dot product with the floor, and then solve the 2D case for the effective length. assuming the height is not changing around the corner and the floor/wall/ ceiling interfaces are all right angles. |
2004/10/27-28 [Computer/Theory] UID:34393 Activity:nil |
10/27 http:/http://www.csua.berkeley.edu/~ilyas/problems/buchi |
2004/10/27 [Computer/Theory] UID:34382 Activity:nil |
10/27 http://www.csua.berkeley.edu/~ilyas/problems/buchi |
2004/10/21 [Politics/Foreign/Europe, Computer/Theory] UID:34263 Activity:very high |
10/21 What Americans truly loathe about dealing with Europeans: Alan Turing on American efforts to break Enigma: "Generally speaking, their attitude is so purely mechanical and mathematical that they often fail to see the wood for the trees and do not like to admit that experience and a knowledge of immediately prior developments, combined with a little manual work, may often produce the answer more quickly than machinery." I say, old chap, could you possibly be a bit more condescending? http://news.bbc.co.uk/2/hi/uk_news/3758276.stm \_ Yes, but you wouldn't understand it. -John \_ Here's my perspective as a scientist who's dealt with western europeans: they'd rather spend a day blathering and writing on the chalkboard than an hour in the machine shop to just build the damn experiment. I'm excluding the English and the Swiss from That's because all the good scientists from here _/ have moved to the US. \_ Interesting. Why do you think that is? \_ Because govt. & corporate grant money here is about nonexistent, universities/colleges aren't as prevalent as in the US (hence fewer research jobs), US salaries are higher, innovation generally isn't as culturally ingrained, and the academic community here seems pretty small and inbred, wasting a lot of energy going at each others' throats, to name a few reasons. \_ </troll> \_ Every foreign scientists I work with is either here already or trying desperately to come here (USA). The few who were here and had to leave are miserable. They do say that Japan is not too bad. A small handful of foreign scientists have a monopoly on the research in Europe and so for those few they are better off there. \_ Uh, so you're saying that you work in the U.S., and the people you've met while working in the U.S. all work in the U.S.? wow. \_ I work in the US with many foreign scientists and agencies. Some are here for a short time only and some are in Europe and Japan. Just because I work with them doesn't mean they are in the USA. \_ Japan's great for research so long as you're completely willing to give up all rights to the fruit of your research. Cf.: http://optics.org/articles/news/10/2/1/1 \_ This link doesn't work for me. this, but in my opinion western European physicists are all talk and no action, and I would avoid working with them in the future if at all possible. \_ The ones I've met pride themselves on their theoretical work and describe Americans as merely applying the practical to the more deeply intellectual concepts they've originated. They meant it in the same way we used to talk about the Japanese when they were doing nothing but cloning American technology and making shittier cheaper versions of it. \_ WARNING! None of the above posters has a clue. \_ Said warning applies to any motd post, so it is redundant. \_ the tone is condescending, self-righteous, and feeling of superiority... ilyas, is that you? \_ No, see the above sounds like something tom would say -- insulting but uninformative. If I wanted to be condescending to people in this thread I would wonder outloud how many posters actually worked with European scientists, and whether they are aware of differences between europeans working here in the US vs europeans working in europe, etc. Actually, what they said doesn't sound wrong to me. In conclusion, you are an anonymous twit, and if you think feelings of superiority are at my personal core you REALLY REALLY don't know me. -- ilyas \_ If you don't know me by now You will never never ever know me -- h melvin and the bluetones |
2004/10/16-18 [Computer/Theory] UID:34169 Activity:moderate |
10/16 Does anyone here play freeciv? I keep getting my butt kicked by the AI players. I'm playing 1.14.1. I've read the online tutorials, but they all seem to be for previous versions or something. The AI ALWAYS advances in technology much faster than me. Any ideas? \_ I can't speak for FreeCiv, but on non-free-Civ the AI gets innate advantages at the higher difficulty levels. Even if that's not the case in FreeCiv, in non-free-Civ, the AI also likes to trade tech amongst itself a lot. \_ In commercial civ, the AI doesn't have to buy advances. They are randomly assigned for free and just magically pop up in their cities. I've never played freeciv but I'm guessing it does the same since I believe they were trying to duplicate commercial civ. Read the source if you really care. |
2004/10/6 [Computer/HW/Memory, Computer/HW/IO, Computer/Theory] UID:33958 Activity:very high |
10/6 If the technology for borgification was available, what would you do? I think it would be pretty cool to have some sort of 'mind's eye' display and an always-on network connection. The other stuff like collective conciousness and major implantation I'm not so hot on. (I thought of this when seeing someone on their cell phone using a headset) \_ Depends what you mean by 'borgification.' I think it's already happening with cell phones, and the Internet, and it will happen more and more. As long as communicating with others is not intrusive (which are the connotations when you hear about the Borg), it seems like a good thing. -- ilyas \_ If a 'direct neural interface' was available, would you get it? \_ Spam and viruses, directly to your brain, yay! \_ Well, presumably it would handle only media and not 'code' (thoughts). You could get spam, but you'd have to decide to check your mail, which would be handled by a remote server made of silicon. \_ Depends on what I would interface with, I guess. I wouldn't mind things like the ability to instantly do arithmetic, or do google searches. Of course those things don't need a neural interface -- only wearable computers. Can you give me an example of a useful functionality which needs a neural interface? -- ilyas \_ I'd like the equivalent of monitor, keyboard, microphone, and speakers wired into my head. You can have any computer you want on the backend. \_ Why do you need it wired in your head? \_ direct neural interface + Windows 2050 = new meaning for 'blue screen of death'! \_ Which includes Buffer Overrun Suite 2050, the most direct way to overwrite your memory. \_ Be the first to beta-test Microsoft Bloodstream 1.0! \_ Apple iBorg. Think Different. No, Really. (TM) \_ Beware The Phone Company! \_ I just want AR (Augmented Reality) sunglasses. They'd give me heads up GPS stuff, overlay building addresses and street names, use facial recognition to bring up info I have tagged to people, etc. I think that would totally kick ass. --dbushong \_ not sure about borgification, but if I had the holodeck to myself, I'd endulge in my wildest dream. ****DROOOOLLLLL***** \_ 7 of 9 + 69? \_ 69.777... ? |
2004/9/30-10/1 [Computer/Theory] UID:33850 Activity:kinda low |
9/30 http://www.netflix.com/MovieDisplay?trkid=73&movieid=60034810 I haven't seen this yet but what does Noam Chomsky say in the movie? Is he the same Chomsky the brilliant computer scientist, and is he a communist? \_ Noam Chomsky is a linguist (I refrained from making a pun), not a computer person. He make incisive radical-left critiques of American politicians and foreign policy, and the way language is manipulated. \_ Incisive? The man is a freak. \_ Chomsky, champion of the Khmer Rouge. \- Chomsky also invented the PEZ dispenser. \_ Noam Chomsky is a such a big dick, and tax-free! \_ No, not tax-free, you have to find another phrase for Chomsky. How about common use property? \_ How about Nim Chimpsky? (They actually called some sign language chimp that). -- ilyas |
2004/9/20-21 [Recreation/Food, Computer/Theory] UID:33639 Activity:low |
9/20 You overhear a conversation. You know the answer to their question. What do you do? \_ What if you're wrong? \_ If you show uncertainty and weakness, the world will drift toward tragedy. \_ I would probably say something, but it depends on context. E.g. if it is someone sitting next to me at a bar I would be inclined to butt in, but if it someone at a different table at a nice restaurant I would not. \_ If I don't know them, I'd keep my mouth shut. Then I can either listen to an answer from someone more knowledgeable, or quietly laugh at someone bluffing who's trying to impress some chicks. \_ I was in the grocery store and this really fat couple were talking about breakfast cereal and something about calories. Like why raisin bran had a lot of calories etc. and I just proferred that the raisins are sugarcoated. A very awkward silence ensued after which I scooted out of there real quick-like. \_ All cereals have essentially the same calories/ounce, regardless of how much sugar they have. \_ Totally depends of the type of question. If it's something like Math or CS, yeah, butt in. If it's what this guy should do about his cheating girlfriend, stay far away. \_ BZT! You are supposed to ask the guy for his gf's phone #. \_ If it's some guy at a store asking "What do I need to buy to do X" and the salesman's saying "Hmmm...." I'll butt-in. If their friend seems to have some sort of clue, I'll let them make due without my input. \-for sales people, if it is just anecdotal or preferences -- "sure that is a comfortable camping chair" -- i dont comment but i have heard salesmen at places like REI give defintitely bad advice w.r.t. things like water filtration ... so sometimes i will say something after the saleperson goes away. if it is something like "what was the movie with X where he Ys ..." then obviously that is sort of situational. |
2004/9/7-8 [Computer/SW/Languages/C_Cplusplus, Computer/Theory] UID:33385 Activity:kinda low |
9/7 what is the answer to a negative number modulo some positive number? is it that same negative number? \_ -1 % 4 = 3, -2 % 4 = 2, -5 % 4 = 3. I think. \_ That's not what bc or mscalc.exe say, but MS Excel agrees w/you! \_ Ask Dr. Math: http://mathforum.org/library/drmath/view/52343.html \_ In mathematics, a mod b where b is positive always returns positive. In C and C++, the '%' operator is pronounced 'remainder' and the results of a % b if a is negative is implementation defined. \_ Also Wikipedia: http://en.wikipedia.org/wiki/Modular_arithmetic |
2004/8/28-29 [Computer/Theory] UID:33200 Activity:high |
8/28 Hey, someone posted a vaguely interesting scifi story that some socialist wrote. It started with the computers taking over management at Mickey Ds and then the world. Anyone have the URL? tnx. \_ I think the author was Cory Doctorow. \_ That guy had a breathtaking understanding of economics. -- ilyas \_ It was Marshall Brain. Here's the URL: http://marshallbrain.com/robotic-nation.htm I'm interested to hear rebuttals of his economic ideas, if only for the sake of having an argument. This is the motd, right? As I understand it, the problem with his argument is that if robots caused 50% unemployment, the economy would evaporate and no one would have any money to buy robots. \_ In a nutshell: any increase in efficiency is a good thing, not a bad thing. That's why the industrial revolution and the computer revolution made us better off. -- ilyas \_ Actually, I think this is the URL requested: http://marshallbrain.com/manna1.htm \_ It depends on how smart the robots are and how easy they are to make. If you could create a system where robots make more robots from harvesting the raw resources through final construction then why would it be necessary to charge money for a robot? Anyone who didn't want to work wouldn't have to. Farm robots, digging robots, maid robots, builder robots. This is one version of social utopia. Everyone would have all the time they wanted to pursue art & leisure instead of slaving away a large fraction of their life doing a job they don't like. I see nothing wrong with that. I'm for 100% unemployment. \_ I think we're already at the point where people can spend as much time as they want on art. What we get for this is a whole lot of really crappy art. I think 100% unemployment would just lead to even more rampant hiedoism. And more motd trolling. \_ And poor spelling. |
2004/7/28-29 [Computer/SW/Graphics, Computer/Theory] UID:32556 Activity:very high |
7/28 Dear CS PhDs, can someone please enlighten me on the different conferences out there, what they're good for, how prestigious they are, etc? -cs dumb \_ I'll start: // prestigious: ISCA: Computer Architecture \_ HPCA and Micro are also first-rate; ASPLOS is superb as well. Hot Chips: Chip stuff, who cares about prestigious? \_ hot chips are hot \_ this is more of an industry work-in-progress, i-built-this-chip conference rather than an academic one. There aren't even really papers published at this conference. SOSP: OS stuff. Pretty prestigious. DAC: CAD stuff, very prestigious. UAI: bayes net AI stuff, prestigious. AAAI: general AI stuff, prestigious. ISSCC: circuit design, way way prestigious SIAM: Industrial and applied math, new algorithms // somewhat prestigious: USENIX: general shit. Not very elite. Siggraph: graphics stuff \_ Um, as a graphics guy, Siggraph is The Bomb. ~10-15% acceptance rate, extremely prestigious. Trumps all other graphics conferences. \_ Seriously. Is there anything more prestigious than SIGGRAPH in the graphics field? \_ Siggraph is Top Dog. Next tier is stuff like Visualization, Eurographics and Eurographics Rendering, Graphics Hardware, etc. \_ It was a rhetorical question, but thanks for the info. ICSE: software engineering OOPSLA: // not at all prestigious: UIST: at least 5-7 years ago. User Interfaces \_ prestigious UI is an oxymoronx \_ While good UI is very important, UI design is the ghetto of CS. GDC: game designers conference ??? \_ It would help if you narrowed down the field a little more. \_ I think that any of the conferances on this list are gonna be fairly prestigious. The not very prestigious ones are the millions of rinky-dink shows all over the world most of us have hundreds of rinky-dink shows all over the world most of us have never heard of. |
2004/7/16 [Science/Electric, Computer/Theory] UID:32321 Activity:nil |
7/16 Haha: Ebert mentions Wikipedia and criticizes MS Word in the same story! (review of "I, Robot") http://www.suntimes.com/output/ebert1/wkp-news-robot16f.html \_ It's wikipedia #1 fan! We love you! |
2004/7/16 [Computer/Theory] UID:32312 Activity:high |
7/16 What did bh call it when you, say, calculate the factorization of 714 and store it somewhere, rather than recompute it when you try to calculate the prime factorization of 3x714? It wasn't caching, it was something quirky. tia. \_ Memoization? \_ bang! thanks. \_ Incidentally, a good way of understanding what dynamic programming is all about is to look at it as "dumb recursive algorithm with memoization." That's much easier for me to think about than the arrays and loops and so on. -- ilyas \_ somewhat intelligent and helpful posts are not tolerated on the motd. You have been warned. \_ Pot stickers? Glomminess? "I'm sorry I can't help you with that homework problem, we're all too busy here fighting against the war!"? Must've been one of those. It's all he ever talked about. \_ Bah. Harvey is a great teacher. \_ I didn't learn a lot of Computer Science but I sure learned a lot about life! \_ Bahx2. Your ideal teacher is probably Emacs running an Eliza based algorithm. \_ Were you one of the BH following pot sticker eating crowd that hovered around him like moths? \_ Have you stopped beating your wife lately? \_ It wasn't a trick question. The above is either a BH #1 Fan or he isn't. |
2004/7/11-12 [Computer/Theory] UID:32213 Activity:high |
7/10 Was Alan Turing queer? \_ isn't this very well established? \_ You underestimate ignorance. \_ that's what they say, yes. \_ He was outed after his lover beat him and he called the cops. They tried to "cure" him by feeding him estrogen; he went crazy and killed himself by eating a cyanide-laced poison apple a la' Snow White, his favorite book. \_ http://www.theregister.co.uk/2004/06/07/manchester_honours_turing \_ there was a poster here who said he wouldn't use programs written by gay people. Does this mean he should stop using computers altogether? \_ I highly doubt you're using any programs written by Alan Turing. Anyway, FYI Blaise Pascal was also supposedly gay. \_ you've got to be kidding... I used to love Pascal the programming language, eeeck... \_ yeah, and i guess you conservative shit heads will also have to find a new way to remember binomial expansion coefficients since you can't use the Gay Triangle. \_ well I have to say, of all the brilliant yet nerdy geeky smelling scientists I've personally met on seen on literature, he's one of the best looking guy... and Feinman. \_ do you mean feynman? \_ not looking so good dead. \_ or alive either; kind of reminds me of hilfinger |
2004/6/17 [Computer/SW/Languages/Java, Academia/GradSchool, Computer/Theory] UID:30880 Activity:nil |
6/17 http://www.csua.berkeley.edu/~nivra/Day_life.html Good stuff, read it before he deletes it! |
2004/6/14-15 [Science/Electric, Computer/Theory] UID:30784 Activity:very high |
6/14 I finally read my first Asimov book: I, Robot. What's the next book I should read? I should mention that I liked the peaceful, not much happening storyline of I, Robot. \_ Asimov is a complete hack. You should read some good science fiction instead. I suggest Dan Simmons, "Hyperion." -tom \_ I liked Hyperion. The next book was such a slog I lost interest in reading anything further (in the Hyperion series). -- ulysses \_ "Pebble in the Sky," "The Stars, Like Dust" and "Currents of Space" Stars like Dust is really good. - pst \_ Looks like they butchered it for the Will Smith movie, but ah well. Try the first three Foundation books - they are as classic as I, Robot but an exploration of very different topics. Asimov later tried to link up his Foundation and Robot series with follow ons, but its still those first three Foundation books that really do it for me. \_ Yeah, about half-way through the book I realized that the movie will be nothing like the book. I've been meaning to read his books for a while anyway, and the movie was just enough of a catalyst to get me to pick up a book. The back of the book had some summaries for his other books, and it sounds to me that the movie might be closer to the other robot series with the detective. Anyway, the descriptions of the Foundation series doesn't grab me much. Any recommendation for another of the Robot series? \_ Well, there's The Robot Series: Caves of Steel, Naked Sun, Robots and Dawn, Robots and Empire, Robots and Earth Then technically the later Foundation books continue it, but Asimov tried to put all of his fiction in one timeline, so everything's somewhat interconnected. \_ So you've read the movie script? \_ I respectfully disagree, but I have no replacement suggestion. The Foundation series did absolutely nothing for me. Since it's some of his earliest stuff, I chalked that up to youthful rough edges. If you do read the Foundation series, please tell us how you think it compares to I, Robot. I'll wait for your review before reading I, Robot myself. \_ I liked the individual Foundation books but the links between them were a bit forced and weak. \_ Is it true that Asimov had no idea how to resolve the Foundation series? ie, it goes off into a different direction at the end of Foundation & Earth, but he had no idea how to get it back on course (if you read the book, you'll know what i mean) \_ "Never let your sense of morals prevent you in doing what is right." \_ I think it should explicitly be mentioned that _I, Robot_ is a collection of short stories and not a novel unto itself. \_ Yeah, but they fit the timelines of development. I was just at B&N browsing for my next book and the first story from another book _Robot_Dreams_ was a chapter verbatim from _I,_Robot_. But my next book to read is the Hitchhiker's Guide. |
2004/6/9-10 [Computer/Theory] UID:30706 Activity:insanely high |
6/9 A claimed proof of the Riemann Hypothesis is now a top story on /. -- ilyas \_ This isn't really news. De Branges has been claiming this for a long time. He's not a kook, but most of the experts don't seem to believe him. (In particular, note that the paper referred to in this recent press release is actually dated March 2003.) my name is lewis. \_ awww. Well, it would be neat if RH were true. -- ilyas \- It is true. We just have not proven it. \- It is true. We just have not proven it. --psb \_ Yeah, I agree with Feynman's intuition in these things. It would be too ugly for RH to be false. -- ilyas \- an interesitng contrast is the continuum hypothesis. i remember being in a room with large math brains and all of a sudden they started saying things like "i think the continuum hypothesis is true" "i dont think it is true!". although on another occasion the largest brain of all in this area i believe came down on "i think there is some weird set that can be constructed that would show it to be false". --psb \_ CH is independent of ZFC. We finished proving this just today in class. The problem with discussing the truth or falsity of CH or any other statement involving infinite quantities is that you ultimately have to make the case that our world is a model of ZFC. That claim seems very dubious to me -- I am not even sure our world contains aleph_0. For me, the truth or falsity of such statements can only be established w.r.t. some model. -- ilyas \- well the famous indep proof gives people a little bit of an out not avail in the case of RH, where so many papers have to issue the "assumes RH is true" disclaimer. this is not really my area and i neither know much about it nor have good intuition ... i was never lost as fast in a math talk as one by a german fellow on model theory. however the chiatin's omega stuff sound kind of conceptually interesting. worth a look if you are not familiar with it. back to CH/ZFC: some time back P COHEN made the comment "i am an analyst ... but i did some good work in set theory" in 10 Evans. "you had to be there". i could not follow WOODIN's work in this area. --psb \_ I think Omega exists like Pi exists. We have slightly better algorithms for approximating Pi, of course. I think I ll need another year to follow Cohen's forcing stuff. -- ilyas \_ I saw a talk by Woodin on this topic a couple of years ago. He thinks the CH is false. If I remember correctly, his argument is that there are structures that can be built up in ZFC + Projective Determinacy that have nice properties and the potential to yield very interesting new fields of mathematics. One can build similar structures in a different level, but to do so requires that the CH is false. \_ That's not an argument for the falsity of CH, that's an argument that the falsity of CH is useful to assume to get interesting math done. -- ilyas \_ And when talking about questions that are independent of the standard axioms what exactly is the difference? \_ 'False' without qualification means 'false in the real world.' -- ilyas \_ He's still been trying to convince people of this. The proof is pretty painful, and you can't really claim this without getting other people (a lot of other people) to look at the proof. It still might be true, I haven't looked at the proof (and probably couldn't understand it if I did) -chialea \_ Does your motd entry actually say anything? \_ Does she ever? |
2004/6/8 [Computer/Theory] UID:30664 Activity:high |
6/8 50th anniversary of the death of Alan Turing: http://news.bbc.co.uk/1/hi/england/manchester/3781481.stm http://theregister.co.uk/2004/06/07/manchester_honours_turing \_ He's gay. WHY DO YOU HATE AMERICA? \_ He was gay? I am disgusted. I am going to change major. \_ I heard there are even homosexuals in the Bay Area. You might want to transfer to another school. \_ consider the csua culture of ridiculing girlz to keep it an all-male institution. \_ Uh... ridiculing girls? what did I miss? |
2004/5/13-15 [Computer/Theory] UID:30216 Activity:high |
5/12 can someone reccomend a numerical analysis text? i never took numerical analysis as an undergrad, and i need to solve some heat transfer equations numerically. something that is readable and also a good reference would be great. thanks. \_ A common one is "Burden and Faires". There is another equally common one I have forgotten but will remember. \_ Burden and Faires is the one we used when I TA'd 128a. It may be too basic for your needs, and I thought it wasn't a very good book in any case. Have you checked Numerical Recipes in C? Again, I don't know if it will have anything that applies to your specific problem, but is a good reference for lots of different things: http://lib-www.lanl.gov/numerical/bookcpdf.html \_ There are some errors in NRi* books. Caveat Emptor. \_ Are you sure? I was under the impression that they had improved in the 2nd editions. Granted, much of the research in them is behind the times. \_ that's very helpful. thanks! -op \_ Be careful. The license for the NR* books is incredibly restrictive. \_ afaik, there are no particularly great num analysis texts out there, or not in english anyway. burden & faires is okay, but has quite a few errors, some in algorithm pseudocode, which can be particularly annoying. just something to keep in mind. \_ Numerical Recipes in Fortran, Pascal, C, C++, Cobol^H^H^H^H^H \_ Fire up Mathematica and plug it in. |
2004/5/7 [Computer/Theory, Computer/SW/Unix] UID:30093 Activity:high |
5/7 How do I make ls sort files by size like -t for time? thx \_ From man ls: The following is how to do an ls listing sorted by size (and shows why ls does not need a separate option for this): ls -l | sort -n +4 \_ use Linux, it's -S \_ I want to do this on soda! \_ download gnu fileutils, compile, presto. \- ls -ls | sort -n[r] --psb \_ ls -ls | sort -n[r] sort: No match. \- dont actually type the [], you doofus --psb \_ ls -l | sort +4 is more general. \- it's more fragile. ls -s always makes the sort field first. otherwise the field can depend on whether the group owner is listed etc ... varies by ls flavor, ls aliases etc --psb \_ well, I meant since it depends on block size it would fail by default for tiny files. \- fair enough, but really arnet you just looking for the biggest file 95% of the time. --psb \_ Add a -n for good measure. Not all platforms will detect an all numeric column, so "sort -n +4" \_ ok, but he said "want to do this on soda!" :) frankly though I think the man page is stupid and it should be an ls option for convenience. they have the other sorting functionality. \_ They have sorting for time becuase sorting human readable dates is difficult with other default command line tools. Sorting file size is easy. This is the UNIX way. many modular pieces you can piece together to do what you want. Linux makes you weak. --scotsman \_ it's a bunch of extra characters to type for a basic, common desire that could very easily just be in ls where normal people expect. (hey, why should sort have a field option anyway? you can just manipulate lines using 'cut' etc. It's the way real men do it.) \_ don't be a doofus--it's not like FreeBSD "ls" is "pure" (color "ls" doesn't make you weak, but sorting on size does?) -tom \_ I wasn't talking about purity. And I guess it wasn't clear that I was being facetious. But the point stands that to pull sort into ls with yet another command line option makes ls more tedious to remember how to use and postpones the newer user's learning how to use sort. And who uses color ls? --scotsman \_ how is it more tedious? If you don't want to use -S, pipe it through sort instead. -tom \_ I guess I'm just sick of interviewing IT drones who can't write a decent command line.. --scotsman \- how about adding a du flag that just means du all the directories under here but not the files. how about another du flag that is directories only in size order. and another that sorts based on number of files rather than cumul size.. --psb \_ Ah yes, I forgot; the only "official" command-line options are the ones that FreeBSD has adopted. Therefore it's OK to use "find -delete", but not "ls -S" -tom \- pure is BSD 4.3 --psb \_ color ls is the best thing since linux! just like color monitor is the best thing since monochrome green terminals. \_ Once you begin to appreciate color ls, you will never go back, like a lot of things. \- dired >> ls --psb \_ using sort -n +5 gets rid of the color, oh fuck, fuck!! \_ If you are using a terminal with black background, try LSCOLORS=BxDxFxdxCxdxdxBxBxBxBx and do ls -G \_ Cool, ls -lG / looks nice! |
2004/4/21-22 [Computer/Theory] UID:13323 Activity:high |
4/21 Cute problem: You are given n bits which resulted from flipping a biased coin n times (you don't know the bias). The problem is to come up with an algorithm which will produce as many unbiased independent random bits from this string as possible (on average). You have no other sources of randomness except the input bits. One possible algorithm would be to pair up the bits, and produce a random bit 1 if the pair is 01, 0 if the pair is 10, and ignore the pair otherwise. This isn't optimal on average, however. -- ilyas \_ assume n is sufficiently large, first calculate the (biased) expected value over the n bits. then instead of pairing up bits, sample m bits at a time, and if there are more than (expected value * m) bits with 1, output 1 or 0 otherwise. slide the sample "window" down and continue. \_ How do you sample without any additional randomness? -- ilyas \_ a simple example of the sampling logic is to take bits 0~m, then 1~m+1, 2~m+2, etc. this does not require added randomness. a more sophisticated variation would vary the next sample depending on the previous 0 or 1 output (unbiased). \_ The problem is if you use more than $n/m$ windows the resulting bits will not be independent anymore, yes? -- ilyas \_ yes for the simple logic. what do i win for spelling out a more sophisticated logic? \_ 1 mid sized cupie doll. \_ eh? expected value? \_ I am interpreting "on average" here to be, "it doesn't have to be perfect, but come up with something good". \_ I think "optimal on average" has to mean something like it produces more bits (on average) than any other such algorithm that also works over the full range of possible biases. I don't see how it could be optimal for each possible bias (e.g. to be optimal in the case where there is no bias, it would have to output n bits for any input, but this would obviously not work if there is bias). - Johnny Von Neumann \_ With all due respect Mr. von Neumann, if the bias is 1, no algorithm will output any random bits at all. You simply have no randomness to work with in that case. To put it another way, the 'optimal' number of bits you should get on average isn't fixed, but depends on the bias. The metaphor here is 'refining oil'. The 'tainted' randomness you get as input is like crude oil. The 'unbiased' randomness you get at the output is like refined oil you get after processing the crude. The amount of stuff that comes out depends on how much crude you have to work with, and how 'crude' it is. -- ilyas \_ With all due respect, you don't seem to have understood what I wrote. An algorithm can't be optimal for the _unbiased_ case (in which case it would have to have no waste product) and also work in other cases, unless you could distinguish this case in advance. \_ I can see why you would think so, but surprisingly, it can. -- ilyas \_ I think ilyas means "optimal" in the sense that a competing algorithm would produce less random bits over the entire space of possible input strings. The "optimal" algorithm is the one most successful at producing the most random bits over the entire space of input strings, relative to all other lame algorithms that produce not as many random bits. Algorithms which produce ANY non-random bits are super-lame and do not qualify for consideration at all. \_ Yes, this is what I think he means, too, and is pretty much what I said above. He seems to be claiming something stronger, which I still think is impossible, but perhaps I have a blind spot. My intuition is that you can do about twice as well as the simple algorithm above. \_ Interpret it as "the expected number of random bits produced by the algorithm, given the input distribution determined by the unknown bias must be as high as possible." I mean "on average" precisely. It's good to think about why is it that the algorithm isn't caring what the input distribution is, essentially. -- ilyas \_ How about something like this: Suppose you saw k heads out of n flips. Lexicographically order all n-choose-k sequences of n flips that have k heads, find the index i of the actual sequence of observed flips amongst these n-choose-k possibilities, and take the binary expansion of i. This requires some fiddling when n-choose-k isn't a power of two, though --- maybe leaving out the last bit in i in this case. \_ This is one correct algorithm, well done. This is called a 'codebook based on the method of types', I think. Another algorithm: run the string through gzip. This algorithm and yours work for the same reason. In both cases you get n*H(p) unbiased random bits out, (which is the optimal number), where p is the unknown bias, and H is the entropy. The reason we don't care about the actual value of p is because the algorithms in question are 'universal source coding algorithms'. This is why people use gzip on just any old file, and expect it to work well. Mr. von Neumann's contemporary by name of Claude Shannon worked all this out in a famous paper. -- ilyas \_ What does "n-choose-k sequences" mean? \_ It's a function often used in combinatorics. You can google for its definition, and you can ask google to compute it for you (e.g. "10 choose 5"). -- ilyas \_ I'd get a better coin. It sounds like someone shaved yours. |
2004/4/6-7 [Computer/Companies/Google, Computer/Theory] UID:13045 Activity:nil |
4/6 Looking for introductory material on SVM, C4., Ripper, and Naive Bayes. Most of the junk on google are too deep for comprehension. I just need introductory material. ok thx \_ http://paulgraham.com / google "A Plan For Spam" \_ where did you hear about all of these terms? did you learn about them with regards to spam filtering? naive bayes is a specific text classification strategy. SVM and C4 are generic classifiers. \_ which one is most powerful? \_ dude, you're asking a hardcore computer science theory question. Most of the people here are sysadms (they don't even know how to do FFT in their sleep). Wrong place. \_ Naive Bayes is hardcore if you call counting hardcore. SVM stuff isn't really CS, although AI people use it (they are big on separating things with planes). Also, the poster above is wrong, Naive Bayes is also a general classifier. An interesting thing about classifiers is that you don't want the one that's 'most powerful' because it will overfit. You want one that's barely smart enough to work. Statisticians call this 'bias/variance tradeoff'. -- ilyas \_ I'm a sysadmin. I graduated with a ~3.5 in CS. I sysadmin because the options are 1) coding which I hate and it's all going to India anyway, 2) I don't want a MS/PhD because I like to eat and pay my bills, 3) I work for about 1.5 hours a day and get paid more than almost every coder I've ever met. Instead of being bitter that you work harder for less while always having to worry about when your job is going offshore, you should learn some industry best practices and get paid to surf like me. What good is FFT in your sleep if in your waking hours you're looking for a non-existent job in your field? None of this applies to the PhDs and most MS people who can always teach if they must. \_ FFT? \_ Final Fantasy Tactics! \_ Fast-fourier transform. \_ this is very sad. |
2004/3/31-4/2 [Computer/Theory] UID:12953 Activity:high |
3/31 CSUA math majors, I need to take Math 104 next semester. Since there are several sections, I want to pick the one taught by someone who's a decent lecturer and a fair grader. I can take it with either of these profs: Mints, Yong, Sarason, Silver. Any good or bad experiences with these? \_ I've taken stuff with Sarason and Silver years ago. Frankly, Silver is horrible. Brilliant, nice guy but is really introverted to an extreme degree. He's really hard to communicate with. Sarason is a "no frills" guy - tough, but a straight shooter. Perhaps one of the best analysis teachers Berkeley has. He is also a solid graudate analysis instructor as well. I have no experience w/Young or Mints. -fab \_ I had a very good experience with Sarason for 104. If I remember \_ I had a very good experience with Saruman for 104. If I remember correctly, Silver is the old guy with the alarming permanent sore on his head who refuses to look at the class. His lectures are worthless. Avoid. are worthless. Avoid. \_ I took H104 with Saruman many many years ago. He was good then. \_ I took H104 with Sarason many many years ago. He was good then. \_ So... You have chosen... the way of pain! \_ 104 entails a little bit of real work, but if you find it demanding, you shoiuld probably change field. \_ Silver. Real analysis is a lot of work, don't take a lot of other hard classes next semester, but I am sure you know all this already. -- ilyas \_ 104 entails a little bit of real work, but if you find it demanding, you shoiuld probably change field. demanding, you should probably change field. \_ Hahahahahaha! \_ Are you sure you have to take 104? Do you want to take 104? Is it a prereq for something else? Or do you just need any upper division math? \_ It's a prereq for complex analysis, probability theory, measure \- strange as this may seem, i am not sure you really need 104 for 185. 185 and 110 are the only ud math classes i can think of where you actually solve problems rather than just doing pfs [like do some integrals] i note in passing, some of the comments on AMAZONG for Baby Rudin are pretty funny. (in a math gnurd sort of way, not like story of ping). btw, is berkeley still using Rudin or has C. Pugh's book replaced it? if your brain is not extremely large, the bartle analysis book might be a good crutch. do not get the horrible Dover book even if it is still $7. --psb \_ What does "needed" mean in math? All you really "need" is enough 'math maturity' to understand the proofs and do problems. 185 "needs" 104 because a lot of 185 proofs assume things you proved in 104, much like an operating system assumes some things ee people do. On the practical side, the kinds of problems and proofs one does in 104 come up in 185. 185 is an easier class. One wonders if a prereq for something else? Or do you just need any upper division math? \_ It's a prereq for complex analysis, probability theory, measure that's because it always comes after 104. \_ Some of 110 sections this semester are using Sheldon Axler's liner algebra texbook which is mostly concept oriented rather than computational in nature. So far I enjoy learning from it. theory, point set topology, and pretty much any other 'continuous math' class you might want to take. -- ilyas \_ I need and want to take it. I am aspiring to math minor. Also, I am hoping to take advanced probability course later. \_ Only three more classes for a major! -- ilyas \_ And you can have classes overlap between majors, so there really is no good reason not to get a major. --darin btw-104 isn't as hard as is rumored, but IS at least as important as is rumored. \_ I never took 104 but lots of geeky friends did. It \_ I need and want to take it. I am aspiring to math minor. Also, I am hoping to take advanced probability course later. \_ Only three more classes for a major! -- ilyas really is no good reason not to get a major. --darin btw-104 isn't as hard as is rumored, but IS at least as important as is rumored. \_ I never took 104 but lots of geeky friends did. It seemed to me that the difficulty level comes from the prof teaching that semester, not the material. seemed to me that the difficulty level comes from the prof teaching that semester, not the material. \- helo. i thought (h)113 was a *lot* harder than (h)104. material in 104 was tought but exams were pretty easy. is this going to be your first class with 10hr problem sets? oh if you want to do probability might take 105 too. measure theory is quite exciting. \_ Ditto, algebra was much harder than analysis for me. -- ilyas \_ That's because you two are both insane. |
2004/3/29-30 [Computer/Theory] UID:12909 Activity:kinda low 60%like:29886 |
3/29 What's the typical way to represent a DFA (or any other FSM) in text? \_ +-------+ +-----+ | start |---------0------> | end | +-------+ +-----+ \_ state transition table... |
12/24 |