Stuart Presnell

@logopetria

Teaching Associate in the philosophy department () at the University of Bristol ()

Vrijeme pridruživanja: ožujak 2009.

Tweetovi

Blokirali ste korisnika/cu @logopetria

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @logopetria

  1. proslijedio/la je Tweet
    2. velj

    You can get 50% off Type Driven Development with Idris today at , as well as some nice Haskell books And since it's Groundhog Day you can buy them over and over again! Yay!

    Poništi
  2. 30. sij

    The lettering of the title “the bus” and some aspect of the drawing style is strongly reminding me of some other cartoon. But what is it? This has been bothering me all afternoon!

    Poništi
  3. proslijedio/la je Tweet
    28. sij

    OK. Conference Call Poetry time. Rules: from what people say, write down random phrases or clauses you like. In order. No deleting. No editing. If you have an odd line, you're on the lookout for something that might work next. Give it a go. It is fun.

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    18. sij

    🚫🚫CANCELLATION NOTICE🚫🚫 Unfortunately we have had to cancel this mornings run as the paths are too dangerous to run on! Sorry to everyone but safety must come first!

    Poništi
  5. 17. sij

    Caveat: I have no expertise in any of this, so there may be mistakes in what I've written. I'm looking forward to the article that will explain it all properly!

    Prikaži ovu nit
    Poništi
  6. 17. sij

    Can this result be turned around to give a characterisation of (some aspect of) quantum mechanics in terms of purely computational considerations? That would be very surprising and exciting! 32/32

    Prikaži ovu nit
    Poništi
  7. 17. sij

    PS: An interesting aspect of this (noted by commenter "RandomOracle" on Scott Aaronson's blog post) is that it seems to be very specific to quantum entanglement. If the Provers share more or less correlation than is provided by entanglement, the result breaks down. 31/

    Prikaži ovu nit
    Poništi
  8. 17. sij

    But part of the fun of this result (and of Interactive Proofs more generally) is the lesson that even if you're an incredibly powerful malevolent trickster god, you still can't fool the teacher by pretending you've done your homework! 30/

    Prikaži ovu nit
    Poništi
  9. 17. sij

    This presumably doesn't have practical consequences, because we don't have reliable access to boundlessly super-powerful but untrusted Provers. (If someone could demonstrate that they had such access, that would be an even bigger result!) 29/

    Prikaži ovu nit
    Poništi
  10. 17. sij

    Even such an absurdly difficult problem falls within the scope of MIP*. This means that super-powerful Provers who share entanglement could persuade you they had a solution, but only if they really genuinely had one. If they're bluffing you can catch them! 28/

    Prikaži ovu nit
    Poništi
  11. 17. sij

    Think of any problem you like that in principle could be solved by a computer. Spare no expense! If you need a super-gigantic computer with enormous amounts of memory, that's ok! If the computer would need to run for a million trillion years, that's fine as well. 27/

    Prikaži ovu nit
    Poništi
  12. 17. sij

    As Scott Aaronson puts it: "There is a protocol by which two entangled provers can convince a polynomial-time verifier of the answer to _any computable problem whatsoever_ (!!), or indeed that a given Turing machine halts." 26/

    Prikaži ovu nit
    Poništi
  13. 17. sij

    The newest result, by Natarajan, Vidick, Wright, Yuen, and Zhengfeng Ji takes this much, much further! Which problems can be verified by separated Provers who share entanglement? All of them! 25/

    Prikaži ovu nit
    Poništi
  14. 17. sij

    This short story by Henry Yuen () illustrates the 2019 result: "You feel like the wizard's apprentice, summoning some deep, extraordinary powers beyond your capability to control." 24/

    Prikaži ovu nit
    Poništi
  15. 17. sij

    Then in 2019, Anand Natarajan and John Wright expanded this to a broader class of problems. This is explained very nicely in this article in by Kevin Hartnett ()

    Prikaži ovu nit
    Poništi
  16. 17. sij

    First, in 2012, Thomas Vidick and Tsuyoshi Ito showed that giving entanglement to the Provers doesn't diminish their ability to demonstrate solutions. If a problem's in MIP (where they don't have entanglement) then it's still in MIP* (where they do have this resource). 22/

    Prikaži ovu nit
    Poništi
  17. 17. sij

    The class of problems whose solutions can be demonstrated by multiple Provers who share entanglement is called MIP*. The argument above suggests that MIP* should contain fewer problems than MIP. But this turns out to be wrong -- very wrong! 21/

    Prikaži ovu nit
    Poništi
  18. 17. sij

    So the natural guess would be that adding this resource means that fewer problems can be verified. Demonstrations that would otherwise be convincing might now be no more than the result of quantum-enhanced cheating by the Provers. 20/

    Prikaži ovu nit
    Poništi
  19. 17. sij

    So we might expect that giving the Provers this extra resource would allow them to cheat -- to co-ordinate their answers more effectively when they don't have a solution and are just bluffing. This would make them less trustworthy, and so less useful to the Verifier. 19/

    Prikaži ovu nit
    Poništi
  20. 17. sij

    (This assistance from entanglement is called "Quantum pseudo-telepathy"! It's as if the two players are given a spooky ability to guess what each other is going to do, and make their moves accordingly, in a way that gives them an advantage in the game.) 18/

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·