This is not quite correct: It cannot be proven true. But it could be proven false with a counterexample. The speaker would simply have to point to a concrete algorithm that can be performed by a person with pencil and paper but not by a Turing machine (or vice versa).
I haven't had the time to watch the video, but from the comments here it appears that the speaker did not provide such an example.
One is the real-number GCD algorithm; basically the same as Euclid's standard GCD algorithm, but with real numbers. He seems to be arguing that since Turing machines cannot represent real numbers, they cannot compute this algorithm. On the other hand, it can be easily computed with pen and paper, and he gives an example of a ruler-and-compass algorithm for evaluating it.
To me this seems a little wishy-washy; a Turing machine can approximate the problem to an arbitrary degree of precisions, just as a pen-and-paper approach is only capable to executing to a finite degree of approximation. Still, an "idealized" ruler and compass does provide a framework for describing an algorithm which is not directly Turing computable.
In fact Algebra has very deep connections to pen and paper geometry.
Apart from that... Due to the increased use of statistics, the postulates of CS might shift. One example being CAP theorem.
Every number that has periodicity in its decimals is rational... √2 doesn't have periodic decimals.
Transcendental numbers are non algebraic numbers, ie. they are not the root of any non zero polynomial with rational coefficient. √2 is algebraic
> Still, an "idealized" ruler and compass does provide a framework for describing an algorithm which is not directly Turing computable.
You could describe such an algorithm, but you could not do it in this universe. This is called hypercomputation.
"If non-computable inputs are permitted then non computable outputs are attainable"
What you're describing is the reverse direction, simulating a TM with pencil and paper, and this is doable so long as you are allowed two stacks of paper to work with, and an unlimited supply of pencil/paper. The question of simulating TMs in the real world really doesn't have anything to do with the Church-Turing thesis, though.
>Despite the generality of what I wrote above, the only feasible counterexample would be one that people can do on paper but a Turing machine somehow can't.
If some dimensionless measurable physical constant were a non-computable number, then the Church-Turing thesis would be false. In fact, if we think that certain quantities are in some sense random or arbitrary, rather than something that could be written as an equation, we should expect the Church-Turing thesis to be false with probability 1.
That's what I'm describing, but what is it the reverse of?
You wrote: "If theoretical physics can find a complete model of all particles and spacetime, and we can show that a Turing Machine can emulate it", to which I added "then a person with pencil and paper can simulate its execution", so putting these together a person with pencil and paper can simulate your complete physics model. How does that suffice as a counterexample to the Church-Turing thesis?
Also once we're talking real world, entropy is finite and there's only so many computations you can do before you run out
Church-Turing is an approach, a labeling, a thesis, not a theorem. Not being a theorem, it is not subject to proof.
One could exhibit a plausible argument that Church-Turing is wrong but that argument would not be a proof. It would not involve a sequence of mathematical argumentation that is standard for those artifacts which we call proofs. Just no.
Pointing to a concrete algorithm would not be a "proof". Maybe it would be very persuasive and the world would be persuaded to abandon Church-Turing. Still it would not be a proof.
Edit: I will admit my claim here involves a "mathematicians' thesis" that proofs are mathematical objects rather than just extra-strong arguments. So be it. I believe the whole reason Church-Turing is labeled a "thesis" rather than a theorem was because it's framers were using this distinction. And the distinction appears in the foundation of mathematics. Philosophers still argue for and against the continuum hypothesis even it's well not to be provable or disprovable - the argument is whether assuming it's true or falsehood produces good or bad mathematics. etc.
No, this is a circular definition. We define an "algorithm" as something that can be performed by a Turing machine. So the Church-Turing thesis is a trivial tautology. (Though of course a useless one.)
No, that's a terrible definition, and it isn't what computer scientists typically use, though it might be what one says if they're not careful. If someone says that "algorithm" means "what a Turing machine can compute", it's because they believe the Church-Turing thesis, which says that they're the same. If they're the same, then you can talk about them interchangeably, and "Turing machine computations" have a cleaner definition than "pen and paper algorithms", so it's convenient to talk about "Turing machine computations" as if that's the ground truth.
But if the Church-Turing thesis is false, then they're not the same, and you should be very careful not to mix them up. And we need a word for "the sorts of things you can compute by hand by a set of well-defined steps", and that's what "algorithm" has always meant.
EDIT: Note that I'm making a prediction. If a person told you that "algorithm" means "what a Turing machine can compute", ask them what "algorithm" means if the Church-Turing thesis is false. My prediction is that they'll use a different defintion, based on what you can effectively compute by hand, like the "Effective Method" wikipedia article that my sibling posted.
https://en.wikipedia.org/wiki/Effective_method
"- It consists of a finite number of exact, finite instructions.
- When it is applied to a problem from its class:
- It always finishes (terminates) after a finite number of steps.
- It always produces a correct answer.
- In principle, it can be done by a human without any aids except writing materials.- Its instructions need only to be followed rigorously to succeed. In other words, it requires no ingenuity to succeed."
"Several independent efforts to give a formal characterization of effective calculability led to a variety of proposed definitions (general recursion, Turing machines, λ-calculus) that later were shown to be equivalent."
Concretely, that problem might look something like:
"Given N non-communicating humans, each of whom see the same 1000x1000 b/w grid. Emit the same answer as 99% of the humans to the question of whether the grid contains a Q."
That would be a problem where, if you could prove that humans can do it, but TMs can't, you've refuted the CTT.
Do I have that right?
Humans do all sorts of things that Turing machines don't (e.g. a Turing machine can't hold a pencil), so the question there is what constitutes an "algorithm" which comes right back to the definition of computation. Even an example that was obviously "computation" in the intuitive sense wouldn't make the Church-Turing definition "false", it would just mean that it defines something different from what we thought it defines.
So if you toss out the idea of computation involving digital computers, you can exhibit something not (exactly) Turing computable. If you have a chaotically behaving dynamic system and call it a computer, then you have something where computation becomes something in the eye of the beholder.
I'm not sure what questions this answers but there you are.
[1] https://www.researchgate.net/publication/221512843_What_Is_a...
In general, a chaotic system can only be said to "solve" a problem if it can be configured to reliably arrive at (an encoding of) the solution for a given instance, despite any chaotic behaviour exhibited by the system. It wouldn't be reasonable to, for example, claim that a Turing machine cannot solve the problem of predicting the behaviour of a chaotic system with some particular initial conditions, if you couldn't reasonably conifgure the chaotic system itself to do that either.
What a standard Turing machine cannot do is entangle and superpose states. (It can only simulate it at high and variable cost.)
Remember that a Turing machine is an abstract thing. Building an ideal one is not possible. You can try though.
One might say that a "classical analogue" computer that "outputs" line segments could output the square root of two but an digital computer could merely "represent" it.
That could be pure sophistry, especially since such classical analogue never have real infinite precision. But it seems like that's the sort of thing being claimed here.
- You can't output a voltage of sqrt(2) because you output a finite amount of electrons in each timeframe, so your resolution is limited.
- We don't know if you can output a signal of length of sqrt(2) because we aren't sure yet whether time is infinitely divisible [1]
- You might be able to position gears or other physical objects with infinite precision, but reading the value with a higher precision than one plank length seems impossible [2]
Computers can also work symbolically rather than based on some direct binary representation. I think equality and especially comparison become more difficult to determine, but it remains possible.
Here is a fully expanded representation: 1.
Of course my choice of base is a bit exotic, but the point is that no matter what, we're working on symbols. We're just used to seeing some as more direct analogues of numbers than others.
For a proof, consider the fact that every representable number, by definition (I know I haven't given a formal definition), is represented by some finite string. But the number of finite strings is countable. So it must be a measure zero subset of [0, 1], so a random sample is representable with probability zero.
How long is that piece of string?
a) Here, have a look at it. It's exactly that long.
b) 4.51 inches
both answers are correct. "a" is correct by definition. "b" may be totally correct if we add a precision to the answer, like +- 0.01 or whatever we find true.
With how much precision can you measure it?
With how much precision can you describe it?
The answers are pretty similar for humans or turing machines IMO.
There is a debate about that in the philosophy of physics and mathematics. AFAIK, according to the current state of physics real numbers cannot be realized within a finite amount of space, at least Gisin [1] recently argued that way and I've seen other arguments for this view before [2,3,4,5:167-170]. I also remember some argument based on thermodynamics but cannot find the reference.
[1] https://arxiv.org/pdf/1803.06824.pdf [2] https://arxiv.org/abs/math/0411418 [3] https://news.ycombinator.com/item?id=14080024 [4] https://ptolemy.berkeley.edu/~eal/pnerd/blog/are-real-number... [5] https://www.sciencedirect.com/science/article/pii/S030439750...
Ruler and compass are made of atoms though. Or, if not atoms, planck units or something.
What, that's cheating you say? You wanted an exact decimal expansion? Well, you're right, a Turing machine can't produce that. But neither can a ruler and compass.
Same goes for printing 0 to an infinite number of decimal places.
I suspect that the proof is that geometry is uncomputable.
You can "compute" sqrt(2) by the geometric construction of a unit square and "drawing" the diagonal.
You cannot do it physically, because marks on paper are not geometric primitives: points are not 0-dimensional; when drawn with a pencil, they have a physical size. A line on paper is not 1-dimensional, etc. But on the other hand, all physical computers are actually equivalent to finite state machines: they don't have infinite (or, unbounded) storage.
I'm halfway through the video, and they are discussing this: the fundamental limit of Turing machines, etc., is that they cannot do an infinite (or unbounded) amount of work in a finite number of steps. As a result, geometric constructions a la Euclid are not computable.
In practice, anything you do with a ruler and compass has an equivalent representation in polynomials that is exactly computable.
¹ http://mathworld.wolfram.com/ConstructibleNumber.html ² http://mathworld.wolfram.com/AlgebraicNumber.html ³ http://mathworld.wolfram.com/ComputableNumber.html ⁴ http://mathworld.wolfram.com/MeasureZero.html
If the problem is "compute and output the decimal representation of the square root of two", then neither the human nor the Turing machine can do it in finite time, because that decimal representation is infinite. In fact, as far as I know this falls out of the scope of the Church-Turing thesis because most definitions of algorithms include that they must run in a finite number of steps, so there is no algorithm to do that.
If the problem is "compute and output a symbolic representation of the square root of two" (in some symbol system where it is finite), then the human and the machine can both do it.
If the problem is "compute and output a segment whose length is sqrt(2) times the length of this segment"... I don't think that's what computation is about. You might as well claim the Church-Turing thesis is false because a Turing machine cannot wave its hand.
I think that qualitatively we can agree that the ruler-and-compass construction of the square root of two is an algorithm. It's a repeatable procedure that can be described in a finite set of steps, and will always yield a provably correct answer. Whereas "waving a hand" is not an algorithm, but the same qualitative definition.
The speaker wrote a paper that proved the Church-Turing thesis for sequential algorithms, primarily by defining in strict terms what a sequential algorithm is. Ruler-and-compass constructions are not sequential algorithms by this definition, and the difficulty in specifying the parameters for a generalized "algorithm" is apparently what leads him to the conclusion that the Church-Turing thesis cannot be true in general.
newtype Real = Real { isGreaterThan :: Rational -> Bool }
sqrt2 :: Real sqrt2 = Real (\r -> r * r < 2)
Doing arithmetic on these numbers is a more complicated matter. But this representation is exact.
newtype Real = Supremum {
isGreaterThan :: Rational -> Bool
}
sqrt2 :: Real
sqrt2 = Supremum (\r -> r * r < 2)My representation is based on dedekind cuts, implemented in Haskell, which is based on the lambda calculus. There are countably many lamdda terms. Thus, I can only represent countably many real numbers. My representation can store all the computable numbers, but there are computable numbers that are not algebraic.
For example, e is not algebraic, but it is computable. Here is a representation of e
factorials :: Rational
factorials = 1:zipWith (*) [1..] factorials
eApproxes :: [Rational]
eApproxes = scanl (+) 0 (map recip factorials)
e :: Rational
e = Rational (\r -> any (r <) eApproxes)Note that there is a commonly accepted extension that is more powerful than a Turing machine: Probabilistic Turing machines. I'm guessing the proof of equivalence generalizes to partial recursive functions and the lambda calculus if you add a 'random oracle'.
An interesting counter-argument is that real random oracles don't exist, and that in reality, we can simulate probabilistic computation using Pseudo Random Number Generators. So:
* True probabilistic computation isn't possible * And anyway, deterministic computation gets close enough to probabilistic computation through PRNGs
Alternatively, one might argue one could build a 'real' random oracle by reading from a 'true' random source. E.g. Cosmic microwave background, atmospheric noise, quantum fluctuations or a video of lava lamps. There is a difference between a random source and a random oracle, but with sufficient storage, that can be patched.
It's one of those talks that seems to talk about things I know but then immediately jumps to theories and definitions I have never heard of. "Sequential algorithms were axiomatized in 2000" - This "axiomization" was apparently performed by the author of the talk, my Googling says. All links about things talked in this talk just seem to lead to other texts and such by this author. Might not be a good sign but maybe it's just a very hard to grasp realm.
Anyway, broadly, the Church-Turing thesis is not a theorem and cannot be proven true or false (so said my professors and Wikidepia). Effectively, it is a definition of computation that has so far been accepted because nothing fundamentally different and more powerful than a Turing machine has been exhibited (there is a theorem that Turing machines, partial recursive functions and the lambda calculus are equivalent - and then there's the jump to "these are really general and I think that about covers it" but still, not a theorem, not provable).
My hunch is the author is mixing in "higher-order" logic "version of Church-Turing" into their argument. Which is basically "cheating", switching to an effectively different question without making things clear. However, I simply don't know. This is basically opaque to my mere MA-level maths.