Friday, May 23, 2014

EU elections

The elections to the European Parliament – one that cannot propose laws, but where some representatives may at least talk to each other – are getting started. Here in Czechia, the polls open at 2 pm today, on Friday, and continue tomorrow when they end at 2 pm, too. After the first day (almost), the Czech turnout is 5% so it is totally plausible that the total turnout will be below 10%. Compare it with over 80% in Crimea, for example, or 75% in Donetsk and Luhansk regions.

As I have already mentioned, I am planning to vote for Petr Mach's Party of Free Citizens for the first time. He got 2.5% in the latest parliamentary elections which was impressive and there are polls that suggest that he has a chance to get 5% this time. The vote isn't necessarily lost. If he gets above the threshold, and I hope it's enough, it will have one MEP, namely Mach himself.

In my opinion, he has thought a lot about the problems and dangers posed by the ongoing processes and trends in the European Union. I expect him to raise relevant ideas. He is a free-market advocate and a teacher of economics. All these things are positives from my viewpoint.

Especially if you are a European reader, I am interested in your decisions whom to vote.


  1. I am ashamed not to know, but what is that island in the heart of darkness?

    By the way, the topo makes clear the beginning of the tank route from the Atlantic to the Pacific.

  2. It looks like Die Schweiz to me.

  3. Nice map. The Cyprus island in the Belarus see is pretty.

  4. It is rather incredible that you could reach such a high rank.However I too would not expect a physicist to win that contest. I wish you the best of luck anyway.

  5. (UK)

    Objective: Elect rabidly anti-EU, rabidly 'racist' anti-immigrunt representative.

    It's a shame there weren't any on the ballot paper. Next best was the BNP and UKIP. Tough choice. Still, best not split the most likely anti-EU vote. Go for maximum damage to the traitor parties. So UKIP, reluctantly.

    It's a pity they are so vacuous when it comes to PLANNING our exit. We need hard-core killers working on that.

  6. Here in Italy we will vote on sunday. I'm planning to vote for the little Scelta Europea. They support ALDE (liberaldemocratic party) and Guy Verhofstadt.
    In Italy the threeshold for getting at least one MEP is 4%, and my party is projected by polls to get ony 2.5%.

    The main anti-EU party is M5S, of the former TV-comic Beppe Grillo. They are projected around 25-30%, more or less the same as the Democratic Party (that supports the socialists in Europe). I think they both are morons. M5S is composed half communist and half fascist (sometimes in the same person, but that's not strange when you look close at these silly ideologies) and really they just say bullishits. The Democratic Party... well they are socialists, enough said!
    I have to say i'm not really anti-EU, i'd just like a different EU, more integrated as a real federal entity and one market (for capital, goods, people) and i also think that a free (from government) market is almost all the time the right solution to almost all problems. So, really i don't think i have other choices to vote for libdem (liberal has, thankfully, a different meaning here in Europe than in USA). You need to consider also the third great party is Berlusconi's one... yes, he is still alive and kicking...

    By the way, we are deservely the second or the first "I" in PIIGS, so i think as italians we really need to shut up and not complain too much about EU 'cause we are part of the problem, not of the solution... but our politicians don't seem to understand that... not a surprise: if free market is almost always the solution, politics is almost always the problem.

  7. My greatest complain about ALDE is that they also buy (not so much as other parties but still buying) the "climate change" bullshits...

  8. Did you see? Nigel Farage says UKIP might find a common field with the French Front National. About time.

  9. Thanks, and LOL, I do - I do plan to be at the top place tomorrow due to an invention that I can no longer verify today due to the 5-submission-per-day limit. But truth to be told, I was already planning to be a leader today! ;-)

  10. Social advocacy educates us that intelligence does not exist, it cannot be measured, it doesn't make a difference. We must then compassionately confiscate Luboš' stolen gains and return them to the deserving, We seek $2.8 million in tax-exempt funding to locate a sexually harassed Black Muslim lesbian single mother of six intellectually challenged intravenous drug addict with AIDS doing the Macarena in a powered wheelchair who rightfully owns title to Luboš' outrageous accomplishment,

    Make it first in the queue, Luboš! Work harder! the congenitally inconsequential always need so much more than they are given. Who hears them when they cry?

  11. Thanks, Umesh. A meta question: Would you bet that I will be (safely) at the top sometime tomorrow? ;-)

  12. Looking at the way you surged up in the past 48 hrs, I would put my money on it; also that I think you posses fantastic 'fluid intelligence' to solve such problems - keep going and all the best.

  13. Also, I must confess I don't totally understand the 'Score' column on the Leaderboard. How is it computed?

  14. Thanks, Bill! It is very surprising and sort of humiliating to learn but I think that when one looks at the great ideas to solve the problem and evaluates them which of them made a difference, it's the non-physics ideas that did! The clever physics doesn't seem to matter much.

    In some sense, the machines with the state-of-the-art algorithms are able to learn whatever physical patterns are in the data by themselves. So when we-humans actually use some variables, invariants, more natural coordinates, and so on, it's always just for us, stupid humans, to see it more quickly. Computers sort of don't need such crutches.

  15. The map has Cyprus in the middle of Russia!

    EU elections in Greece are skewed because the opposition party ( a leftist party whose leader is Tsipars and a candidate of the left for European "president" ) has made it a referendum against the strict economic measures taken following the memorandum betwee EU, EU central bank and IMF, the troika.

    The conservatives have just managed to balance the budget with a bit of surplus ( ignoring the service of the debt, but it means we will not be increasing it at least) and the opposition wants us to go back to square number one. Tsipras is a clone of the old man Antreas Papantreou who put us on the road of socialism and a top heavy civil service. He promises everything to everybody if the vote is such that they can claim need for general elections ( normally planned two years hence).

    So I will be conservative and vote conservative, New Democracy. This is not the time to show free thinking.

  16. Very incisive..thanks!

  17. Dear Umesh, it can be interpreted as a "number of standard deviations" (in the sense of 5 being the discovery) that you would get if you employed your technique to distinguish the background and signal.

    It's given by their complicated AMS formula involving a logarithm in a square root, and it's all complicated.

    But it's very useful to understand the simplified picture. The score is essentially the ratio

    s / sqrt(b)

    where "s" is the weighted sum of your "true positives" (you guess "s" as signal, and it indeed is), and "b" is the weighted sum of your "false positives" - you guess "s" but it is "b".

    At the end, "b" ends up being substantially larger than "s" because the weights of "s" are approximately 1/1,000 per correctly guess event. So roughly speaking, the ratio (score, AMS) is, with some normalization of order one,

    1/1000 times correct hits divided by the square root of wrong hits.

    Note that "s" scaling like "sqrt(b)" is natural if most of the stuff is "background" which may "random walk". After "b" events of the background, the natural fluctuation is "sqrt(b)", OK? So if you can show that the actual number is greater by something substantially higher than sqrt(b), like 5*sqrt(b), then you have a discovery.

    For various reasons, it doesn't really matter much how many events you guess to be "s", what's the percentage. (Except that their formula is modified to discourage guessing "very special" ensembles.) There are always errors and all algorithms produce something like s^2*fixednumber wrong hits for s correct hits.

    I found some intuition to think about these numbers of errors etc. but some of it may be what may help me tomorrow to take the lead so I will only publish it if I learn that it's wrong! ;-)

  18. Ah! I see..thanks for the reply. As I already said..keep going and hope to see you on top!!

  19. Dear Bob, as John wrote, it's called Switzerland. They have been neutral for almost 1,000 years and non-membership in the EU is a part of it.

    They like to organize referenda about everything, are German-Italian-French ethnically mixed, and it's fair to say that the EU greatly affects them, anyway.

    It's also true for Norway which is also absent from the EU map - Norway is the missing foreskin on the Sweden penis at the top. Look at the complete map of Europe to see what I mean. In this sense, the EU is a circumcised male European continent.

  20. Good. I can understand him being a little sniffy about the socialism but not the 'racism' which he should 'embrace' with gusto. His prime objective should be to get us out of the EU while inflicting maximum damage on it at the same time. It's possible the two might conflict at some point but not for now at least. Besides, socialism in itself isn't terminal — recovery is possible.

    By the way, here's a quote from your link. Dumbarse labour-voting whitey is finally getting the message.

    oppons, 22 May 2014 9:20am

    I'm a Labour voter, a liberal really. But I will vote UKIP, why?

    My Town is totally overrun with Eastern Europeans and there are no jobs because they took them all and do them for min wage and send it back to Poland/Serbia/Russia.They are no integration at all, they never smile, never say hello, just drink alcohol at 6am and drop their kids off just after at school.

    Good for the Economy? That is rubbish, I don't give a flying crap fuck about the economy right now. If I was a loaded farmer who can buy this slave labour for peanuts then its good, its good for the mega rich who will exploit them, crap for everyone else.

    [From your link to that loss-making peecee scum's house magazine. Comment amended to correct vernacular.]

  21. "The map has Cyprus in the middle of Russia!"

    Shhh! It's OK. I can just about make out that there's a that conventional little square box painted around it —something which apparently gives cartographers a free hand to place things anywhere they like. Just don't tell Mikey Mannic.

  22. "They like to organize referenda about everything..."

    These referenda are a superb restraint on the Swiss political class, whom I understand would have plumped for EU membership if they'd had their own way. Even so, Switzerland is inundated with foreigners (the free-movement shit they have to swallow as part of dealing with the EU) which the Swiss people are not exactly ecstatic about and have recently voted against. I understand that this will cause some difficulty.

    Yet another reason to kill the EU stone dead.

  23. Congratulations, Dr. Motl!

  24. Congratulations on your score!

    "Of course that for some hour during this angry period, I was thinking about plans how to legally exterminate everyone who has ever been connected with this Linux crap."

    Haha, that's funny because, while I sympathize with you, I think you have a few common points with Linus Torvald.

  25. Lol!
    It is so true that these Eastern European people just never smile.

  26. I think he means Cyprus in Russia on your map actually! Look again.

  27. LOL, Anna, I think that Putin et al. would like your comment that the place where Cyprus was placed is in the middle of Russia! The reason is that up to today, the region was actually mostly on the border between Ukraine and Belarus - but if that will be the center of Russia, good for the Kremlin! ;-)

  28. "strict economic measures" -- please tell me what is strict about > 10% annual budget deficit.

    "ignoring the service of the debt" -- right, and ignoring expenses, Greece is running a big surplus. In fact, ignoring reality, Greece is the most productive country on earth.

    "but it means we will not be increasing it at least" -- how so? you will have to pay the interest which *will* increase your debt (by more the 10% per year, as it stands)

    It's said to see that even intelligent Greeks still live in economic fantasy land full of lies and political slogans that do not align with reality.

  29. You are misreading and are being unnecessarily offensive.

    It is not an economic fantasy. It is what eurostat also accepted.

    If I owe the bank, unless I manage to have even a small surplus on my monthly expense budget ( without serving the debt ) I will never be able to repay the bank and will either be increasing the debt if the bank is foolish enough to accept this, or sell the family silver.

    This year with the firings and budget cuts and economy measures the present coalition government managed to stay within the budget and have a small excess which can go for the debt interests. If we were as provident twenty years ago we would not have ended in this mess.

    I hope that that this will be sustained although a restructuring of the huge debt will have to be arranged so that it can be re-payed sometime in the future. This will certainly not happen if there are elections again and the left takes over.

  30. What exactly do you find unnecessarily offensive? That I called you intelligent, or that I state the obvious?

    Interest is in no way optional or discretionary. It is a firm contractual obligation, as binding as any. You are confused about this trivial point. A bank would always assess your liquidity *after* interest. Only principal repayment (prior to maturity) can be optional, depending on the terms of the loan agreement.

    Even the surplus before interest is most likely bogus, a result of accounting tricks such as deferred payments. It is a one-time trick serving a political purpose.

    Eurostat's accepting this hocus pocus does nothing to validate the purported "facts" but only exposes it as an agency promoting a political agenda.

    I agree that the left taking over would be detrimental. But let's face it: Greece cannot and will not pay its debt. Ever.

  31. Dear Comrade YouDon'tSay, like every person who is not a parasite, I am immediately being paid by a corporation - money that the corporation is being paid by others for services they find desirable or necessary.

    Even when I was employed in a public-like sector, I was being paid by a corporation, well, e.g. the Harvard Corporation, that was getting money mostly from resources earned by some real corporations, anyway.

  32. I was wondering if you a job or something.

    What you don't use Fortran???
    Time to go back to school.

    I started with f then b then p then c .....

    Actually you should be able to just dump the numbers into a stats package that find groups of data.
    No brains needed.
    Even ups uses this for their performance and veichle monitoring.

    I just started a new job and its using red hat.
    Nice change from windows. And the other programmer is from a scientific super computing background too.

    Bye for now.

    Next time in waterloo call me for free dinner.

  33. Tx for the invitation!

    I was taught that Fortran was a piece of post-war history, something that the ancestors would be using as a "computer language" shortly after the war. And that's how I still view Fortran! ;-)

  34. I find offensive the way you are not listening. "dont bother me with facts, I have made up my mind that everything is hocus pocus", and the conspiracy theories about eurostat.

  35. And the imbecile did not allow the students to directly ask questions.

  36. :) Psychologically interesting! You could well be right.

    "Heart of darkness": years ago I would have automatically taken that as the USSR/Russia.

    Today it's the EU. For good reason.

  37. I hope Lubos isn't offended by these racist Eastern European comments.

  38. Spain votes on Sunday, probably with a high abstention rate (above 50%), and a vote with clear internal motivations, a vote to punish/reward to national policies. Only a libertarian party is presented, P-Lib, and is unlikely to get one single MEP (among the fifty-four MEPs Spain contributes to European show). My impression is that people in general have no idea how the European Parliament works, and does not even know what the European parties are.

    The worst, probably one (or even two) representatives of the unicornial far left will jump the queue and get a seat. The best, not a single minute devoted to climate change during the campaign.

  39. Lubos always smile. Look at his picture on his blog! Maybe his long stay in the US?

  40. Hello! I am landing here a couple months after making this G+ post that would almost read like a precursor to this blog post of yours where it not some six years late:) -- while looking for a worked out kinematic description of orbital mechanics and changes of coordinate systems in terms of Fourier decompositions or epicycles, not the pre-Copernican historical stuff which is interesting for itself, but what would be the most extensive scope to which, in hindsight, the epicycles ideas could be expanded to capture more recent knowledge/understanding. Did you by any chance stumble upon something like that?

  41. "Actually you should be able to just dump the numbers into a stats package that find groups of data. No brains needed."

    I was thinking the same thing, why no one has mentioned R? It seems to me that R can be a powerful and versatile tool for this job.

  42. The EU is now demanding that countries count the illegal economy (proxenetism, drug dealing, contrebande) into the GDP !

  43. Don't listen to a word he's saying, Casper. He's a filthy frog. Ya just can't trust 'em. Perfidy, that's them. History says so.

    Still, at least he's not an eastern yurpian.


    Hello Shannon, I hope you are keeping well. :)

    Now, there's a good fellow — butter me one of those croissants and pour me a glass of that wine. Let's enjoy the peace together while we can, before our petit interbellum ends. Oui, mon ami?

    Brek brek! :)

  44. Fortran was designed by a woman. Maybe that is why you don't use it!!!! ;-)

    After the war the computing dept were all rooms full of women doing spreadsheets. Maybe that is why the woman invented fortran to liberate them from the computing sweat shops.

    It was a woman who had the spark idea that solved the German code machine. A 1 to 1 too obvious mapping.

    Last I heard the Fortran woman was making nuclear missiles to prevent more war. She was keeping the men in line to make sure they work.

    Most of the stats packages like R r still in F.

  45. Holland already voted, not with much success. Only 35% voted. The anti-EU party got about 20% of the votes, not that great.

    I'd say vote anti-Brussels. They are too anti-democratic anyway, since the parliament doesn't have any power.

  46. They have much more bullshit. It was Verhofstadt who went to Kiev to support the new neo-nazi government. Verhofstadt is dangerous. He is only going for his own benefit, is part of very large enterprises. Very pro EU. I don't understand that anybody could vote for such a guy.

  47. Thanks. That project R looks interesting. I could use it for work to help pick optimal design configs.

    Will try the quick R to start. May have to compile some Fortran to get it to work on Unix.

    Currently we use cost as the main factor.
    However there are many solutions near the min cost. So we need other factors like temperature margin and simplicity to choose between the options.

    In fact machine learning could also be used.

    See there are more applications than just finding the higgs and other obscure events.

  48. I always smile but the other thing that I didn't even have a slightest idea that it could be about me because I have surely never considered myself an Eastern European.

  49. Your posts seem CS ignorant. Fortran was not designed by a woman. But you can read something about history of COBOL instead...

  50. After installing R, you can install the RStudio IDE, it works in Windows, Linux and Mac and is more friendly that the crude R environment. Unfortunately right now they have the web broken (www.­rstudio.­com), but it works very well and can significantly reduce the learning curve.

    About Machine Learning there are tons of packages (free and propietary), try google "R machine learning".

  51. I strongly recommend "Die Partei", unfortunately only eligible in germany.
    Maybe you negotiate with them and start a bohemian offshoot, then you will have the chance to be a member of the EP for at least one month (they plan to share the won seat(s) by monthly rotation).
    Their party platform is both conservative and progressive, as you may read here:
    (unfortunately again only in german)
    Or just watch one of their official election ads:

    Cheers and happy voting

  52. It was Mrs BB who was ignored with the a-a z-z mapping. So the poles has to tell the English in 1939 July.

  53. "...since the parliament doesn't have any power."

    It has some power, but only to REJECT legislation (proposed by the unaccountable kommission, whose sole prerogative that initiative is), and not always then (but I forget the circumstances for the creepy exceptions). It has NO POWER to INITIATE legislation.

    But then—and this is the really important thing here—it would be as EQUALLY BAD if it did. That's because it has NO DEMOCRATIC LEGITIMACY. None of these shit EU institutions does. And that's VERY SIMPLY because there is NO EU DEMOS.

    And that makes the EVERYTHING to do with the EU a COMPLETE AND UTTER FUCKING LIE.

    Yet no one ever mentions this most fundamental and blindingly obvious DEFECT.

    Geez, just how dumbfuck stupid do people have to be not to see this, that this whole business is a vast CRIMINAL ENTERPRISE — on a CONTINENTAL SCALE already. Oy!

    Just remember that the next time you hear some shit-faced EU apparatchik utter the words 'democracy' or 'democratic'.


    Hey, I guess some folks don't understand what a demos—or even consent—is. Maybe they should take the time to find out. Yes, that and how to wipe their own arses and tie their own shoe laces. TOP TIP/HINT for them: consider the fundamental driver of any kind of independence movement.

  54. I'm no history buff, so this might be a silly question.

    Was there ever a distinction between eastern and western Europe before WWII, apart from the trivial one embodied in references solely for geographical purposes? Didn't the distinction (cultural or whatever) arise only after, and because, the Iron Curtain descended?

  55. Ribbit ribbit... ;-)

  56. I'm under the impression that Lois Haibt was a female member of the team lead by John Backus, that created the Fortran programming language.

  57. Sounds good. Stewed? Yes please. And some of your beret-onion-et-bicyclette soup to go with it would be nice. :)

  58. Soup on a bicyclette? Mmm... I've never tried that one. And you've forgotten the baguette!! ;-)

  59. Grace hopper developed the first working language called A. A-0 was the first version. But the men did not want to use it. As you know real men program on binary!!! Just ask a IBM 360 expert.

    You are correct that Fortran came a little later.
    But Mrs amazing grace still was first.

  60. A0 came out in 52. A2 in 53. B-0 later became COBOL.

    Fortran was started in 1953 by IBM.

  61. Yes you are correct. IBM did Fortran well after 1953. But Grace Hopper beat them with A-0 and A-2 in 1953. The A2 was open source too.

    Rand should have patented the compiler.

  62. My congratulations! After reading your two articles about this challenge, I really want to sign up .... and win, of course ;-)
    I have some doubts: why the training material (with known answers) is supposed to be enough to determine the winning algorithm ? (Why) Do you think that programming skills could be more useful than knowledge of particle physics to win this contest?
    I'm not fully convinced that machine learning is the right approach, even if they claim to have found a strong connection between Deep Learning and High-Energy Physics that would improve the Search for Exotic Particles:

  63. "changes of coordinate systems" read "change of reference frames" of course:) BTW, I found something closer to what I was looking for here:

  64. Lubos if are down to 14th place. Better throw some more coal on the fire and start up another computer.

    The top 59 entries are from 3.6 to 3.69.
    Only a 3 percent variation. Not different using stats significance.

    You are at 98.1% from the top.


  65. Re antisemitism:

    Brussels shooting: Gunman opens fire outside Jewish centre.

    "The La Libre newspaper said on its website that an Audi had driven up and parked outside the museum, and that both a passenger and the driver had got out. It said the driver placed two bags on the ground and then opened fire on bystanders before driving off."

    That's it. The only description of the killer was that implied in the title, namely 'gunman' (but that could just be generic*), yet they managed to get a description of the car!

    How queer. I wonder if there's something they're leaving out.

    In fact, I'm sure there is. Any bets on what that might be?

    Yes, 100 to 1 says the terrorist scum were those grunting simian adherents of the religion of piss.

    But shhhh! Don't say anything lest the revelation inflame 'hate' against these splendid diversoids, and some thereby become inclined to return the 'compliment'.
    [Preferably on a suitably bloody massive scale when they do. Ed]

    * The occupation of gunman will remain with its non sex-neutral appellation since it wouldn't be peecee to let it be thought that split-arses aspire to the calling: 'gunperson' is unfair to wimmin.

  66. Well, my friend... as prime minister in Belgium Verhofstadt lowered taxes and reduced the public debt. I have very few expectations from politicians in general, so that's all I ask from them, 'cause any fool is able to overtaxing and overdebting people.

  67. Dear Svik, down to about 17 now. I hope that some of those who jumped above me are TRF readers.

    Yes, this whole top is arguably indistinguishable and may reshuffle in the final results.

    It's not a matter of computer power. My best model is generated in 5 minutes or so and then it deteriorates, probably to overfitting. About 3 truly conceptual ideas of mine to radically improve the score failed in recent days. If I won't get another radical one soon, I will stop working on it.

  68. "The fact that the full theory (e.g. quantum field theory of a Dirac field) respects causality exactly is a derived fact based on various cancellations."- could you please explain which calculations/cancellations are you reffering? Isn't it enough to ensure field commutators vanish for spacelike separations for causality to follow?

  69. Yes, it is enough. However, the field commutators are not directly linked to the observables we are computing, like scattering amplitudes, so one runs the risk that the argument using the vanishing commutators could be formal. It needs to be verified.

    In that sentence, I was perhaps unfortunately assuming that you treat quantum field theory as a relativistic extension of quantum mechanics where particles may propagate in some way - some over trajectories - and merge or split - interactions. It's possible to rewrite the Green's functions in that way. However, the correct causality properties of the Green's functions (which are otherwise implied by the vanishing commutator of yours), if you calculate them in this framework, require you to sum over particles *and* antiparticles propagating in between, so it's "hard" to add pieces to construct a relativistic theory of particles.

    The language of the field operators makes these things pretty obvious. You know the vanishing-commutator argument (you need to verify that these vanishing commutators will be preserved after evolution in time, however). And the field operators may be Fourier expanded into creation and annihilation operators that inevitably contain the antiparticles as well (unless the field is completely real in some way). So in the language of the fields, one may also understand why it's necessary to have antiparticles for the Lorentz symmetry.

  70. thanks! Yes, for me vanishing commutator argument has been confusing and thats why I asked the question. So if I want to understand it more concretely, say someone told me that he found neutrinos travel between source A to detector B faster than speed of light. What expectation value should I calculate in QFT(free or interacting) to show him that his experimental results are at odds with theoritical prediction? (say, not involving any slick vanishing commutator argument) If I understand your above answer the vanishing commutator garuntees that the correct expectation value in this case is going to be zero. Also pardon me (and my silliness) if this is not the right place to ask pedagogical questions.

  71. Hi AAPtard, it's very good you're interested in these matters, and I hope that you will reach a (an even) clearer image than I have. ;-)

    My main warning was that the normal Feynman propagator F(x,y) in the spacetime representation does *not* vanish at spacelike separation. See

    So this could look like QFT itself is producing faster-than-light responses. But in reality, it doesn't, because the actual response is dictated by the commutator and that vanishes at spacelike separations.

    So you want to know how a neutrino created in x point of spacetime affects something measured at y, where x,y may be spacelike separated. The neutrino field does or feel something at "x" that can be represented by the commutator of operators with the neutrino field and one ultimately gets the influence of the neutrino field as proportional to [j(x), j(y)] where j(x) is the "current" coupled to the neutrino field. And that vanishes at spacelike separation.

  72. I’m proud to be European. Europe as a *whole* is the cradle of civilization. The European civilization defined humanity.

  73. Right, Giotis, I share your viewpoint although many South African, Middle East, North African, Chinese, Mexican, and Indian people would disagree about the cradles of civilization.

    But even if one adopts the perspective of glorious Europe - the perspective of 1500 AD, sort of - I feel that this pride is sort of vacuous because the civilization spread almost everywhere, including the U.S. or Canada or Australia etc. And on the other side, one has future and euphemistically speaking, I am much less sure whether the future is what Europe may be proud about. ;-)

  74. Prostitution and drugs are two areas that flourish as economies tank and people become desperate. I look forward to CNBS headlines that gloss over negative GDP by playing up the growth sectors of whoring and crack-peddling. Pimps and dealers can be interviewed for their expert opinions.

    Italy is a special case, though. When 40% of GDP comes from organized crime, you really don't have a lot of choice but to declare it. Shopping for Italian labels supports organized crime in a way that buying a bag of weed could never hope to.

    Institutionalized criminality, we should be so surprised.

  75. There is no such thing as a “free market,” Lubos. Either a market is regulated or it is dominated by its largest players. Either way, the market is not “free.” What Adam Smith referred to as “free markets” were those freed from monopolistic or oligarchic control, usually by aristocrats and sometimes by the Crown, but rigged in any case to establish insurmountable barriers to entry to entrepreneurs like himself. Our contemporary corporatists have naturally corrupted this understanding to mean markets that are freely available for them to rig, which is the opposite of what Smith meant. Of course, the “state” conventionally refers to the federal government.

  76. On the contrary, Cynthia. There's nothing such as a "regulated market". If something is regulated, it's not a market. The very point of the market is that it regulates itself. If it doesn't, it's the kind of communist misery we were forced to "enjoy" for 40+ years. No, thank you, I prefer to contribute donations to any assassin who will kindly regulate away those who want the regulate our free markets. I've been fighting for that for years and I don't want to surrender any easily.

    It's completely normal and unavoidable that there are largest players in markets, science, any contest of any sort and by definition, it means that they have a larger influence than others. If you are fighting against these things, you are fighting against tautologies and you can't really win such a struggle. But for me, it's not just a rational issue. I have a positive emotional attachment to many corporations! They're making our lives, at least my, much better. They represent most of what I consider the human and civilizational progress.

  77. Mind you, it will free up prisons, tribunals and the police. Lol, imagine that :-)

  78. I see you are still sending max entries per day with a total of 30.

    Buy there seems to be a limit at 3.699999. Gabor is closest. Maybe there is a flaw in the math and only nobody can tunnel over 3.7.

    There is a long way to go to teach sigma 5.0

    Nobody needs to find a radical new approach.
    May try quantum tunelling with a resonant frequency.
    Or call john Baez.

  79. I depleted the limit 30 minutes after I woke up today. It isn't too time consuming. The best score was improved by 0.003 or so haha.

    The apparent convergence below 3.7 is intriguing, I noticed that as well. but it must be a coincidence. If you happened to steal their solution file, you could get arbitrarily high.

    The difference between the current leader's score and mine is possibly within noise. But he is really good - this Hungarian chap has won some previous AI contests.

    The failure of about 5 ingenious methods of mine to improve the score suggests that the limit of legitimate methods is really in the ballpark of 3.6 and 3.7 may already be due to good luck - which may evaporate when the remaining 82% of the solution is used instead of the 18% now.

  80. That limit is 1 / (10 * alpha) - 10.

    Using the fine structured constant as alpha.
    That fine higgs has some structure after all.

    Try to break 3.7035999


  81. Good morning. Hope you can solve it today.

    If not I may have to dust off my Fortran compiler.

    I would scale the data to show the small changes at a 0 to 10 scale and use clustering stats packages. Use linear log or energy based scaling.

    Look at the distribution of each param to determine the best scale to make the higgs data show it self.

  82. LOL, amazing numerology especially because it works and it isn't too far from how it works.

    The AMS score actually has a regulating term +10 in the denominator - did you see that equation?

    So in some sense, one is given "10 false positives" (or 5, because the weights of "b" events may be 2 in average) even if he scores everything right. It's there to prevent people from guessing very small collections of "s" events, to disfavor them from transcripts with very few "s" events.

  83. My personal best jumped by 0.01 now, to 3.63552. It's the first modest but clear success of the heterotic imperial nose method - the heterosis added 0.015 relatively to the naive average of the scores. ;-)

  84. What do you mean by clustering? What do you want to cluster? The features? To find out which features are more discriminative there are various methods to do it. Just google "feature selection". You will find a huge literature about this topic

    The best way to go is to use crossvalidation on the training data set, ie. train on 4/5ths of the data and use 1/5th of the data for validation, and then rotate. Average the perforfomance over all rotations.

    The important thing is what to use as a metric for the validation. Classification accuracy is not very good, especially for datasets with class imbalance (one class more frequent than the other).

    The best thing is to compute a confusion matrix for each classification result.

    and then to compute different metrics from the confusion matrix. I use the Matthews correlation coefficient, which is similar to Pearsons correlation coefficient but for binary data. The advantage is that all - TP, TN, FP, FN - is evaluated. I use the confusion matrix to plot an ROC curve for every classification and compute AUC ROC

    Why is it important that also the FP, FN get evaluated? Because sometimes a FN is more dangerous than FP, i.e. missing one false negative cancer diagnosis is more dangerous than 10 false positives.

    So when submitting do kaggle, evaluate your models with the Matthews coefficient and not with classification accuracy. Because the MC contains also the FP, FN etc.

  85. Yes, of course, I encountered it. It was central to the point of your message. I don't see that Jews "push" their intelligence "to the point of morbidity." I don't even know what pushing one's intelligence to the point of morbidity would consist of. It sounds like some rare pathology. Can you name a few Jews who have pushed their intelligence to the point of morbidity, and tell me just what it is that they did, even if they are people you know (in which case I don't want their names.)

  86. If I may add a few more words to this thread, lemme first congratulate you on this absolutely magnificent display of brain power! (Assuming you've stopped trying, of course.) Even if my money can be argued to be not on the dot as far as the ranks are concerned, it still is amazing that you've come to within 10 of possibly the most serious programming folks on the planet. Check out the CV of the top guy:

    In my opinion, it's a true testament to the level of a top string theory professional with very high standards. It's only fearsome to imagine how Edward Witten, or other luminaries famed for their 'sheer intelligence', like Feynman, von Neumann would have been in their heyday. SCARRYY!!!

  87. The hibertating bears woke up with a 3.70685.

    Well there goes my theory with alpha. And no I did not see the 10 in the formula.

    I suspect that the signal need to be amplified or focused to see it at all. Looks like they are picking off the easy high amplitude events only.

    So the real contest has not even started.

    So I see there are 5 risers and 2 new commers.
    So about 10 people working hard on this problem.

  88. Hi Svik, I have already made a $100 bet that 3.8 won't be surpassed in the final score (from the 82%). If you want, I can explain you how the "focusing" actually works or doesn't work.

    It's really the property of known algorithms that even though the mysterious file contains 30% of signal events, it's optimal to guess about 15% of your transcript to be "s".

    If this fluke is neglected, it's a general rule that the number of false negatives - your guessed as "s" that are "b" - is actually scaling as the number of true positives "s=s" squared, so s/sqrt(b) stays fixed.

    You optimally localize many boxes, and the more uniformly you do it through all kinds of events where the signal may reside, the more correctly you reproduce the ideal distribution that exists and is nonzero everywhere, the higher score you get.

  89. Hope you win the 3.8 limit bet.
    What are you going to spend the winnings on?

    Better yet go for the $7000.

  90. Pretty fancy, back in the top 10 as of June 4.

  91. You're number 5 on the leader board, putting you in the top 1%. Awesome Lubos!

  92. Thanks, John. It was #4 when I posted the new record submission today. ;-)

    Chances are 30% that I will be in top 2 tomorrow. That's the probability that a boost I have already tried and previously and prepared in the new parameter & code conditions improves the score by more than 0.015 which I need for top 2.

  93. Haha, this serves TD right :-D!

    So maybe he should put more effort into his own business he is paid for, than into being a sourball who tries to imitate his friend the Trollking by spreading a bad anti-(theoretical) physics mood on his blog ...?

    And of course congratulations to you Lumo, looking forward to see you at 1 ;-)

  94. Impressive pickup in your ranking!
    It would be great if you could beat them all.

  95. Still gaining, up to #3 as of June 7.