Sunday, March 27, 2011
Thursday, March 24, 2011
Since I am in the middle of some very interesting discussions and experiments concerning this topic: it really does look like the difference in speed between the different multi-core/multi-processor Stata/MP versions is considerable. Right now I am performing a comparison exercise* with my friend and co-author Miguel, where, it turns out, a number of certain estimations with a 8-core Stata/MP 11 take
about 10 almost 15 times less real time (I know, I hardly believe it myself) than (almost) the same estimations done with a 2-core Stata/MP 11. Taking the median across all estimations commands in Stata, an 8-core will outperform a 2-core by a factor of 2.28 (NB: this ratio appears to be larger than the price ratio of these two multicore versions!): I computed this figure from other stats available in this 250 pages report on Stata MP's performance.
The next quest should be to assess Stata's bold claim: "From dual-core laptops to the big iron of multiprocessor servers, Stata gets the most out of multicore systems. No other statistical software comes close" (my emphasis in bold). If true, that surely ought to boost Stata's status in the statistics/econometrics community (part of which-- I plead guilty too-- is currently infatuated with using other programming languages like Ox, Fortran, Gauss, Matlab etc, while typically leaving Stata for simple exercises or data manipulation).
*brief update footnote here, for clarification (well, at least part of the clarification..., the huge, factor
10 15, execution time differential remains somewhat puzzling): we cannot really perform this comparative exercise directly, as "(almost) the same" here means that we do have identical specifications (and the same software and operating system), but nonetheless different (very large) panel datasets-- and in this particular case the difference in the structure (specifically, the "connectedness" of the cross-sectional-time series data, but the intention was not to give too many details) of those datasets matters, beyond an eventual difference in the number of observations (for our purpose virtually the same); if we ran identical specifications on exactly the same dataset, with the 8-core and respectively the dual-core Stata/MP, the speed ratio could not be higher than 8/2=4, the theoretical limit. Given the type of exercise performed here, the real time differential due solely to the difference in core numbers is most likely close to this limit.
*brief update footnote here, for clarification (well, at least part of the clarification..., the huge, factor
Tuesday, March 22, 2011
A couple of books I have read within the past few weeks, most of them on my Kindle 3G device(*):
Scott Berkun's "Confessions of a Public Speaker" (get the Kindle edition): the author is a professional speaker, in front of audiences large and small, hence he has got some very helpful tips for anyone who ever needs to engage in public speaking, mostly drawn from his own experiences. These tips range from the a priori preparation for your speech, to engaging your audience, to knowing what to do when you have a too large room and too few people in there etc.What is missing, but understandably so given the author's career, is the treatment of academic public lectures, which can be quite different than any other public types of speech-- nevertheless a series of Berkun's recommendations work perfectly also for academics. The book is at the same time a very fun reading.
Jason Zweig's "The Little Book of Safe Money: How to Conquer Killer Markets, Con Artists, and Yourself" (get the Kindle edition): useful advice for small investors (that can be probably summed up by the following quotes "for most small investors, an indexed bond mutual fond is the best way to go" and "[d]on't invest in leveraged and inverse ETFs. Leave them to professional traders" ...) and adapted particularly for the US audience (you do find for instance very helpful info on the so-called '529 plans,' including some online links to compare various such plans, in case you are interested in that). Definitely not for more sophisticated investors (some of you, readers of this blog) or for people who've read much more detailed stuff on financial investments earlier (although I am myself in this category, sometimes it is useful to consult other perspectives at a more elementary level). A tad too cautious and limited on scope and purpose. Anyhow, if you are a beginner and (plan to) live in the US, by all means do read this book.
Steven Levitt and Stephen Dubner's "Superfreakonomics" (get the Kindle edition). I find this better written than their first Freakonomics success (see, e.g., the first bullet point here for some of the huge debate on that earlier book)-- probably experience and learning on the job-- though it contains as little economics as the first one (it is written more in an investigative journalism style: Dubner obviously has the feeling for what sells and what not). That being said, I still think it is eventually a great service even to the economics academic community (at least in drawing attention to the range of situations/phenomena that economics can be applied to; NB: I personally doubt it would give the/a too wrong impression of what economists do and should do-- if anything, it might convey the idea that some economists know very well how to sell their books,), and probably that writing it in more 'econo-prose' would have reduced its popularity. I would particularly recommend it to any non-economists, for sure (e.g. last winter I noticed the Romanian translation of the book is already out, and made it present to a good old Transylvanian friend of mine-- in function of his reaction I might offer it to my parents as well). By the way, related, the Freakonomics documentary is currently available for instant viewing on Netflix (I correct myself: right now their site is down, but it is the first time I do not see it functioning), if you are in US and have access to that service-- I plan to watch it one of the next days.
And now to my two favorites in this set:
John Adams's "Hallelujah Junction. Composing an American Life" (unfortunately not available on Kindle): this is one of the best autobiographies I have ever read, within the musical realm and beyond. Honest, full of wit, style, and extremely informative-- and that to say the least. As somebody who's listened to a large set of Adams's compositions (and working on the remaining ones...) and who regularly follows his blog, I simply had to read his autobiographical piece in order to have access to Adams's detailed perspective on both his own work and his general ideas of music (composition, interpretation, intrigues, all the hoopla), politics, life, everything... Highly recommended: one of those books that you simply cannot stop reading once you have started. I'll most likely refer to passages or ideas from "Hallelujah Junction" in many of my future blogposts.
Steven Landsburg's "The Big Questions" (get it on Kindle): I will only tell you that this book is from my perspective really scary, in that its author appears to think about crucial stuff --i.e. mathematics, economics, physics, philosophy, tax policies, trade, sex, religion, life, Universe etc.-- in an almost identical way to yours truly (ok: I should maybe phrase it as "I think almost exactly like him"-- after all he was there before me...). In any case, even if you've known all the time and agreed all the way with everything in the book (though he does cover an enormous range of topics, requiring that you'd be an extremely well read person), Landsburg's exposition is one of the clearest I've ever encountered (this is true also for his previous books, some of them recommended earlier on this blog; I will only say here that I even used an earlier quote from him as one of the "propositions" --"stellingen", in Dutch-- to go along with my PhD thesis defence-- see under IX in this list). In particular, he's got the rare ability to explain in few paragraphs, mainly by means of easy examples and analogies, some highly complex material (try explaining quantum physics or deontological vs. consequential ethics to the non-initiated!), albeit he needs to strip them down to the bare essence (but then again this is a popularizing book, aimed for a wide audience). Landsburg's writing style is also something I personally like a lot: he's got just the right amounts of humor, irony, and even the tiny bit of arrogance (why not?) mixed in his book. Check out also Landsburg's "The Big Questions" blog, something I've followed and enjoyed since its very start. I end with a larger quote from the introductory chapter of "The Big Questions" that is likely to give you a good feeling of Landsburg's 'creed' and goal in this book (as well as the similarity with my own academic training and beliefs, c.f. the scary/eerie part from above):
"[...] I went on to a career of research in teaching in both mathematics and economics, with a little dabbling in physics along the way, but I never lost sight of my fascination with the big questions of philosophy: Where did the Universe come from? Why is there something instead of nothing? How is knowledge possible? What justifies a belief? How can we tell right from wrong, and good from evil? How should we live our lives?
Philosophers have useful ways of thinking about these questions, but so do people who are not philosophers. Physicists know something about the origins of the Universe; mathematicians know something about the patterns of reality; economists know something about how our choices affect the lives of others, which is not distinct from the problem of distinguishing right from wrong. I've come to believe that these disciplines provide some of the best available tools for chipping away at the problems of philosophy.
When a man with a hammer tells you that everything looks like a nail, you should doubt his objectivity. When a man who knows some math and economics tells you that the problems of philosophy can be solved with math and economics, you're entitled to exactly the same reaction. But in this case I believe the causality runs the other way: I was drawn to math and economics because they illuminate the big questions. I saw the nails and went out to find a hammer. [...]"
(*)which is absolutely fantastic and you must have it, despite 1. the fact that David Letterman cannot find page numbers on the Kindle books -- and hates it because of that; 2. the fact that each Kindle book costs more if you buy it outside US-- function of the IP address from where you log in to your Amazon account when buying it (before investigating and realizing this, I did actually buy quite a few Kindle books during my European winter holidays-- and ended up paying about 50 bucks more than I would have paid, had I bought them here in the US). In any case, I am the owner of more than 100 Kindle books already (yes, Amazon, I expect some loyalty awards!): given my fairly long commuting time between Lakeview, where I live, and Northwestern Univ in Evanston, I can read up to two or three books in good weeks.
Friday, March 18, 2011
Saturday, March 12, 2011
I tend to watch great movies with a considerable delay..., but then again, my feeling is that I get to appreciate them much more in this way, aside all the initial hype and the turmoil associated with a new release (an auxiliary gain is that I can discard-- without having to go through the pain of actually watching them-- a lot of movies which were released with high expectations, only to turn out total flops, hence really what remains is above average or, in the best case, excellent). One of these movies is Tim Burton's Sweeney Todd (IMDb, RottenTomatoes), the movie version of Stephen Sondheim's famous musical that premiered only days before I was born (i.e., an eternity ago)-- and which I still promise myself to see on Broadway at some point (or: in London-- Sweeney Todd the musical was rumored to know a revival in London this year-- perhaps with Alan Rickman in the role of Judge Turpin, as in the movie; but, as it looks, this will hardly happen before next year...). In any case, this movie easily gets very high in my top movies all time, at least top 10 (my 25-movie hierarchy put together a while ago has changed significantly since then, as you no doubt noticed if you followed --particularly recent--blogposts set in my "movies" category).
Briefly, this movie broke several personal records among my earlier movie rankings/assessments: i). it is by far my favorite film version of a musical (Chicago was my top choice so far); ii). it is also my winning choice for a dark humor movie (title claimed earlier by Delicatessen); iii). it is my number one Tim Burton movie so far (though I still have to see some of his movies that might challenge that); iv). it is my best Johnny Depp movie to date (and I can safely say that I have seen most productions in which he acted; however, with Depp one can expect any time a new movie which can claim the first place in this ranking; another interesting observation I would make here is that Johnny Depp has that amazing quality of almost always being outstanding in his role, even if the rest of the movie in which he happens to play is mediocre or worse, which happened in a few cases). These opinions were strengthened after watching 'the make of', i.e. a series of interviews available also on the DVD of the movie, with Tim Burton, Johnny Depp, Helena Bonham Carter, Alan Rickman, and several others among the actors and crew, plus the one and only Stephen Sondheim*.
Anyway, the point of me writing this post was to emphasize my possibly favorite scene of the movie (I say "possibly" because this choice might be mood-dependent to some extent), also my favorite piece of music from the soundtrack, the "Pretty Women" duet scene between Sweeney Todd and Judge Turpin (which remained in my memory and would not go away since). Here's the full scene from the movie. Here's an audio version only, with better sound quality. And here's an audio version with lyrics**.
* I thought of Sondheim as a genius ever since I first saw/listened to West Side Story, the Bernstein-Sondheim masterpiece. Among the interviews available on the DVD of "Sweeney Todd" there is one amusing line of Alan Rickman (who, by the way, played superbly Judge Turpin; more generally, the few movies where I have seen Rickman acting--still have to see some of the older ones-- already place him extremely high in my actors' ranking). He confesses that he thought absolutely crazy the fact that the great Stephen Sondheim himself came to listen to, criticize, and encourage the actors--most of whom never sang publicly before and were thus 100% amateurs when it came to musicals.
** There are other fabulous parts of the soundtrack/scenes (after all, the whole thing is superlative, you should not miss anything!). A personal selection: "No place like London", "Johanna" (Antony's version), "By the sea", "Epiphany", "My friends", "A little priest"-- the latter with a bonus: "A little priest" 2005 live version, with the original Sweeney Todd musical casting of Angela Landsbury as Mrs. Lovett and Len Cariou as Sweeney Todd.