Categories

Tuesday's Class, Preparing for Exam 2

Exam 2 will be handed out Thursday, April 8 and due Tuesday, April 13. It will cover everything through today’s class (Barbara Liskov’s lecture), but emphasize things that have been covered since Exam 1: class 10-18, Problem sets 4 and 5.

If you have any topics you would like me to review in class Tuesday, or questions you want me to go over, please post them here.

As a separate game theory challenge, if exactly ONE student in the class sends me an email before noon Tuesday, April 8 with the subject line “Exempt Me”, that student will be exempt from Exam 1. If exactly two students send me such an email, no one is exempt from the exam but both of those students are “exempt” from any three questions (receive full credit for those questions without any answers). If more than two students send me such emails, no one gets anything. You can coordinate amongst yourselves anyway you want for this challenge (short of breaking any laws or causing or threatening physical harm, of course).

18 comments to Tuesday’s Class, Preparing for Exam 2

  • GameTheoryPlayer

    Dear CS3102 Class,

    I am aware of two people who have already e-mailed the professor for this challenge.

  • akg9r

    Could you spend time reviewing reduction proofs?

  • Robyn Short

    I second going over reduction proofs.

    Your “game theory challenge” reminds me of an aggression study they were doing at Brookhaven National Laboratory on cocaine addicts. It tested their aggression by seeing how often they would “harm” (in terms of the game, not physical harm) the other players when it served no advantage to them.

    • Okay, I’ll go over some example reduction proofs Tuesday.

      As for the cocaine-addicted rats, I think its more like the prisoner’s dilemma, except with many more than two players. Every individual student’s condition is only improved by sending me the email (i.e., it can not hurt your own interests by doing it), but if everyone acts in their own self-interest no one gets anything.

  • Ryan Matt

    April Fools, right?

    In all honesty, I wouldn’t trust “GameTheoryPlayer” up at the top, since he could be self servingly protecting the fact that he knows of no one else who has emailed Evans and is hoping that everyone gives up and doesn’t email them, heck it could even be Evans himself with a fake account just to throw us off. I emailed him for sure, on the narrow hope that he decides that anyone who emailed him actually does get exempted from Exam 1.

    Reduction proofs sound like a good idea too.

    In a Markov machine, output is dependent on the state?

    • Robyn Short

      Interesting thoughts… I assumed GameTheoryPlayer was Evans and he was toying with us to see how selfish (wanting the exemption for ourselves) or malicious (wanting no one else to have the exemption) we were, so I opted not to play because the spirit of the game is negative.

      • Interesting theory. I can (honestly) answer a few questions:

        1. I am not GameTheoryPlayer.
        2. Despite the date, this is not an April Fool’s joke, the promised exemptions are real.

  • moc8f

    I also approve of going over reduction proofs as well as a brief review of undecidability and how logic will lead to eventual madness.

    http://en.wikipedia.org/wiki/Tragedy_of_the_commons

  • jcf4r

    Well…there are 4 possibilities here: no one emails, 1 person emails and gets total exemption, 2 people email and get full credit on 3 problems, and more than that email and no one gets anything.

    The most “three musketeers” choice would be the first, were we to band together, not submit any emails, and ignore the game wholly. But even in ignoring the game, we still play it. However, the most beneficial overall would be the 3rd, since that would help the greatest amount of people. But that puts into question our willingness to help other people through no gain of our own. and further…how much can we trust the others to resist the temptation. since this is a one-off trial…it’s not even testing game theory, but rather our own fallibility.

    There’s a 5th option which is to test the rules of the game…say by instead of saying “exempt me” we say “exempt___” someone else in such a way that each person has a request from someone else to exempt them or better yet just “exempt everyone”.

    Ultimately though…we’re dissenting players in a game we never had a choice in in the first place. And like Neo did in the Matrix…we must realize that there is no spoon.

  • Kevin Leach

    Like the prisoner’s dilemma, the best choice for each individual is to defect on the rest of the class (by sending the email). The penalty for defecting against everyone is the same as what the sucker would get through cooperating.

    That is, the worst thing that could happen by sending the email is that you don’t get any bonus–but that’s the same thing that would happen if you didn’t send an email. So, you risk nothing by sending the email, but you still might get the bonus on the small chance that no one else sends the email. Thus, sending the email can only serve to help you and never hurt you, so why wouldn’t each person send one?

    I submit that no one will win since (most) everyone will send the email. There has to be a penalty for the people that defect (eg, more than 2 emails = -10 points or something) before the game will be reasonable.

    Anyway, can you go over reduction proofs and some more undecidable problems?

  • Renardy

    I think that this problem has an easy enough solution. If the class were to identify the person who would score the highest score on the exam, they should have that person exempt themselves. Due to the generous nature of Professor Evans, there will undoubtedly be extra credit options on the exam. If the highest scoring student is removed from the pool, I’m assuming Prof Evans would just give that person a 100, effectively keeping said person from attaining a 100+ score (which they would have gotten anyway). This would lower the class average, and cause Prof Evans to offer some sort of curve or improvement opportunity, which would raise everyone’s final scores. The person who would have scored over 100 would also benefit from not having to take the exam. Everyone wins. QED

    • tdr3b

      I don’t want everyone to win. I want to win. But since I’d score the highest on the exam for sure, just exempt me. I don’t need more than 100 points anyway. I’ve got an un-countably infinite infinite number, collab would crash if Evans tried to put my grade in. In fact, I don’t think we’d see much of him if he tried, as I don’t suppose he would finish either unless he was trying to do it on his cell phone while lecturing. Then I’d let him stop at 9999999999999999999999.

  • mmk2d

    Could you review the alternate proof for the self-rejecting language presented in class 17 and the table for counting all languages in sigma star from class 15?

    • It seems there is most interest in going over reduction proofs, so I will spend most of today’s class on this. If there is time, I’ll also try to fit in another example of a proof similar to the self-rejecting language. Note that Sipser’s proof that the Halting problem is undecidable uses a similar diagonalization argument to the one we used in class for self-rejecting.

      On counting, the PS5 comments for Problem 1 give a few more examples that should be helpful.

      I would, of course, be happy to go over any of the things I don’t get to in class today during office hours.

  • no1lakersfan2007

    Could you also give a brief overview of what Barbara Liskov talked about on Thursday, so that I can try to understand the important points from her guest lecture. I was at the talk but I want to make sure I understand the important things she brought up.

    • I don’t think I’ll have time to get to this since explaining it well would probably take most of a class. I’m glad most people in this class made it to Prof. Liskov’s talk and hope you got a lot out of the talk, but I won’t be asking any questions about it on the exam since although there are strong connections between any problem an CS theory, anything technical about this is probably too far outside the scope of this class to be a fair exam question.