Talk:Time loop logic

From Wikipedia, the free encyclopedia

You have new messages (last change).

What is that? Temporal logic has a very well defined meaning in computer science, which has nothing to do with what is said here (which, by the way, sounds baloney)David.Monniaux 17:48, 18 Sep 2003 (UTC)

Wasn't aware the term was already in use for that. I'll move this over to time loop logic, then; it's the term that's used in this paper by Hans Moravec. Bryan

As a computational model, how is time loop logic different from non-deterministic computation? How does getting an answer from the future differ from guessing it? Gdr 01:09, 2004 Jul 22 (UTC)

If you set up the system so that guessing incorrectly produces a paradox, then you physically cannot guess an incorrect answer. You'll guess a correct answer every time. Note that this depends on a couple of assumptions about how time travel works, as mentioned in the article, and that you've set up the system robustly enough that other non-paradoxical outcomes (such as the computer breaking down before it gets a chance to run the program) have low probabilities compared to guessing correctly. Bryan 03:10, 22 Jul 2004 (UTC)
Interesting. So as a computational model, how is time loop logic different from non-deterministic computation? Consider the class of problems solvable in time O(f(n)) in time loop logic, call it TLLTIME(f(n)). How does TLLTIME(f(n)) differ from NTIME(f(n))? Gdr 15:46, 2004 Jul 22 (UTC)
Hm... I'm not very familiar with non-deterministic computation, so I don't think I can answer this question myself with any confidence. Looking over Moravec's article in the external links section, though, it looks like he proposes using time loop logic elements to actually build nondeterministic computers in the "solving chess" section and the "non computable" section down near the end. Am I interpreting that correctly? Bryan 07:18, 23 Jul 2004 (UTC)
A fine point that must be specified is whether "solvable in time O(f(n))" means that the result is returned within f(n) or that the machine runs only f(n). Due to time-travel there is a difference, e.g. in Moravec's article the machine that solves non-computable problems yields the solution after very short time, but needs to run infinitely long. Let's assume we are talking about the total running time of the machine. Then Moravec's construction of a chess computer seems easily to generalise to a simulator for alternating Turing machines, so TLLTIME(poly) contains PSPACE. Assuming that NP is strictly contained in PSPACE (unproven assumption), it follows that TLLTIME(f(n)) strictly contains NTIME(f(n)). --DniQ 21:12, 7 May 2005 (UTC)
They are two different entyties, time loop logic is a physical model that shows how to build the non deterministic computer. Comparing both is like comparing boolean logic and modern computers.

It could obey Novikov's principle, or maybe it'll just obey Dr. Emmett Brown's principle, and "cause a chain reaction that would unravel the very fabric of the space-time continuum and destroy the entire universe"... can't wait to find out which :) —EatMyShortz 00:37, 17 February 2006 (UTC)


[edit] Super-supercomputers

I was intregued by the concept of using time travel(/communication) for computation.

Wouldn't a supercomputer (able to compute fast) that were to be able to send and recieve itself messages(/calculations if you will) back and forth through time be the worlds fastest supercomputer?.. seen that it knows the answer, even before the problem is sent.

There wouldn't be a problem if the answer to any calculative problem would be corrected and sent back correctly. (paradox or not) There would however be a problem if the incorrect answer to a calculative problem would be assumed as the correct one if the calculative problem hasn't arrived yet, and sent back to an earlier time (to hold up the theory of fastest supercomputer).

A simple solution to that problem would be that the computer would be waiting, perhaps with multiple recieved answers (along with the problem) ready, and sends a correct/incorrect message along with the answer (and the problem) afterwards, trusting itselfs answer, but still with the "waiting" filter. Taken that in consideration, the computer will again be the fastest, seen that it knows the correct answer (with the "correct" tag) before the problem has even been put in, it only has to compute, wich answer is correct, if none of them are, it corrects.

Say there are three solutions sent back, it's a far less computing to say 1, 2 or 3 are true or false, then to calculate.

Paradox or not, the system works (even better then normal supercomputers). Right?

I'm not a rocket scientist, I just like theorizing.. plz comment

Misterbean2000 06:49, 6 May 2006 (UTC)

Well, the problem is that there is a paradox, and that this idea rather depends on it. If I understand you correctly, you're taking results from the future, verifying them, then sending them back in time in a different format than you received them. Sending something back that's different than what you received for yourself is possible if and only if paradoxes are possible.
Luckily, we don't really know whether paradoxes are possible, since, as far as I know, nobody has proven the existence or nonexistence of time travel, so it'd be very difficult to experimentally determine whether they are :)


[edit] causality

Perhaps I'm an idiot, but doesn't the example in the article defy causality? Step 1 depends on step 2.1, and step 2.1 can only happen 'after' step 1. It seems to me that even if the computer in question were able to send data back and forth through time, the only possible non-paradoxical outcome of that algorithm is that the computer sits in a wait loop until someone turns it off.

Shouldn't it be something like this?

  1. Wait for the result to be transmitted from the future for at most n seconds.
    1. If n seconds have passed, stop waiting and send a random number back in time.
  2. Upon receiving the result, test whether it is a factor by dividing the input number by it.
    1. If the received result is indeed a correct factor of the number, send the result back in time.
    2. Else, if the received result is not a correct factor of the number (or no result is received at all within the desired timeframe), generate a number different from the received result and send it back in time. Note that this results in a paradox, since the result sent back is not the same as the one that was received.

—The preceding unsigned comment was added by 69.133.103.18 (talk • contribs) on 14:08, 23 June 2006.

The whole point of a time loop is that the concept of causality may break down, such that you can't pinpoint to a specific event as being the "cause", since tracing back further and further just gets you round and round the loop. You might want to see articles Predestination paradox and Ontological paradox for details. 131.107.0.73 02:38, 19 July 2006 (UTC)

I understand Novikov's principle, but I don't see how it applies here. With these ideas, you can make a computer that always makes a paradox, giving every answer, or lack thereof, a zero probability. Somebody help!

Retrieved from "http://en.wikipedia.org../../../t/i/m/Talk%7ETime_loop_logic_a849.html"