Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

  1.  

    I would like to argue in favor of reopening this question. It has already generated quite a productive conversation, it was asked honestly, and it seems like it could be of interest to, say, beginning graduate students.

    • CommentAuthorHarry Gindi
    • CommentTimeFeb 15th 2011 edited
     

    I disagree. I think that the question comes off as embarrassingly naïve. Its sheer wrongheadedness makes me question whether or not the OP has ever even read or written a mathematical proof.

  2.  

    Harry, you seem a little belligerent today. Remember: it's the green pills in the morning and the red ones at night.

    That said, I agree with your underlying point. I don't see any evidence that the OP really understands the true purpose of mathematical proofs (to forestall the usual anonymous comments, I'm not convinced that I understand it either. But I know what it isn't when I see it. Kevin's answer comes closest, though I think that I've seen Eugenia Cheng say something similar in a more polite way.).

    If someone thinks that there is a genuine question underneath, but that the OP isn't communicating it well, then either edit the question or re-ask it. However, I suspect that there isn't a genuine question. What there may be is a blog post on the nature (and nurture) of mathematical proofs and that there are many different levels of understanding a proof, and to reach a certain level you (usually) have to do some activity relating to the proof. Off the top of my head, I would class the levels something like:

    1. It looks alright to me.

      Activity: I read the statement of the theorem, and the author once bought me a drink.

    2. The main idea of the proof is ...

      Activity: I read the proof all the way through and my eyes didn't glass over at all (honestly, well, apart from the point where the author came in and offered to buy me a pint if I didn't read the next paragraph too closely).

    3. The proof depends on ... and leads to ...

      Activity: I read the proof twice!

    4. I could write out this proof without looking at it.

      Activity: Make that three times.

    5. I could rewrite this proof.

      Activity: You can't make this claim (honestly) unless you have actually done it.

    6. This proof generalises to the case of the analytic and algebraic topology of locally Euclidean parameterization of infinitely differentiable Riemannian manifolds.

      Activity: Rewrite the proof in popular song form.

  3.  
    Andrew, that's metrization, not parameterization. See http://mathoverflow.net/questions/36139
  4.  
    Hello guys,

    I cannot agree with anyone here. This is not a question of interest only to beginning graduate students. It is a serious question in foundations of mathematics, and I'm afraid that it is just such a case that even to understand that it is a serious question you have be a bit into logic yourself. (I'll elaborate on why I think so below.)

    Now from what you're writing, it seems that none of you here (the first 3 people in this thread) and none of those whose closed the question is a proof theorist nor even is doing any work related to foundations. A quick glance at the user profiles doesn't seem to contradict this impression.

    It further seems to me that many are misreading this question as a "soft-question" or a question in "general mathematics", which it is not. The OP has clearly marked it by the "proof-theory" tag, and none other, which makes sense to me.

    Now let me elaborate. If you're deeply into varieties or manifolds, I can understand why you might view this question as silly: it is definitely not going to help you in understanding proofs about model categories; nor how to solve PDEs. But why do you guys judge for all mathematics? Are you fully satisfied with your understanding of the proof of e.g. Goedel's incompletess theorem? (To the point of understanding also the work of Gentzen, Goodstein, Kirby-Palais and Harvey Friedman, to mention only a few, which may be viewed as trying to explain Goedel's theorem from different perspectives?) Now why do you think that you know it better than the OP (even if he/she is an undergrad student) what aspects of proofs are relevant for proofs of theorems in the area of foundations?

    I think Mathoverflow should have a policy that questions in logic shouldn't be closed by topologists (note: I am a topologist and not a logician). This is not the first time that this comes to my mind, prompted by strangely closed questions. In practice it could be that one cannot vote to close a question whose tags have no intersection with his "interesting tags" (and also one cannot add to somebody else's question a tag that's not among his "interesting tags"). Apparently one would think twice before changing his "interesting tags" just to close somebody's question.

    Added later: the OP also has a point that his question does end with a question mark, which not everybody has noticed. Some commentators seem to assume that the OP just wants to state his position, but at least from the answers you can see that a question exists, whether you dismiss it as impractical and idiosyncratic or not.
    • CommentAuthorHarry Gindi
    • CommentTimeFeb 15th 2011 edited
     

    It is a serious question in foundations of mathematics

    Such questions exist?

    But in all seriousness, I think that you're reading far too much into the question. Let me schematically split your question into two parts:

    1.) Questions the OP could have asked but didn't.
    2.) Reprimanding those of us who closed it, because we're not logicians.

    I'm sorry, I can't say that I agree with any part of that argument. Maybe if François or Joel were to come on here and voice a similar opinion, I would take notice, but your argument doesn't make any sense to me, and I don't think you have some kind of special insight into the OP's mind or the minds of logicians in general.

    Now why do you think that you know it better than the OP (even if he/she is an undergrad student) what aspects of proofs are relevant for proofs of theorems in the area of foundations?

    It seems like you're asserting that proofs in some areas of math are necessarily more rigorous (in theory) than proofs in others. I don't buy it. Even though the average paper in homological algebra may not be as rigorous as the average paper in a more foundational area, that doesn't mean that homological algebraists and mathematical logicians have a necessarily different conception of what constitutes a rigorous proof when push comes to shove.

    I think Mathoverflow should have a policy that questions in logic shouldn't be closed by topologists (note: I am a topologist and not a logician). This is not the first time that this comes to my mind, prompted by strangely closed questions. In practice it could be that one cannot vote to close a question whose tags have no intersection with his "interesting tags" (and also one cannot add to somebody else's question a tag that's not among his "interesting tags"). Apparently one would think twice before changing his "interesting tags" just to close somebody's question.

    I think that this is a dangerous precedent to set, and I get a visceral reaction just reading it. It kind of goes against everything that mathematicians hold dear and everything that mathematics is about.

  5.  

    To Gerry: That'll teach me to believe google ... (for the record, I put a similar faith in random MO questions so I've just checked and you're correct, at least as far as the 1960 version is concerned)

    To Sergey: I did not vote to close the question, I certainly do not judge myself knowledgeable to decide whether or not a question in proof theory or logic is of a sufficient level to be on MO or not. My remarks, facetious as they were, had very little to do with the content of the question and everything to do with the perceived level of the original questioner. In a situation such as this, I wait until the experts have commented on a question. If it seems that the question is vague, I look at the response of the original questioner to the comments. That is something that I do feel qualified to judge. As yet, there is no real response so I have no inclination to vote to re-open the question. If you (or anyone else) feels that there is a real question underneath this one, then by all means edit or re-ask it!

    I think that the suggestion in your last paragraph is off-beam. Certainly, in those areas where I do not feel that I am an expert then I am much more cautious about voting to close. But I still do so, and to reopen on occasion. I do so when it is clear that, however much there may be a good question underlying the one asked, the questioner themselves does not (appear to) appreciate that fact. This seems perfectly legitimate to me. To be a good MO question requires something of the questioner, not merely the question itself.

    I do not want to see the fragmentation of MO into little ghettos according to some bizarre classification such as "interesting tags" (for one, I don't use those except to make the screen a little more interesting with random colours).

    • CommentAuthoran_mo_user
    • CommentTimeFeb 15th 2011
     
    Just an obersvation as in the amount of comments to the question, it might be overlooked.

    Joel David Hamkins commented that he votes to re-open; since he was directly (and indirectly) mentioned in this discussion I thought this might be relevant.

    [This is a purely technical observation; I refrain from voicing my personal opinion.]
  6.  

    @Sergey: With all due respect, I think you are reading too much into the question. I don't see the question as being about foundations at all, and the OP doesn't say anything to indicate he/she is thinking along these lines.

  7.  
    "Such questions exist?" - Harry, that's exactly the problem I tried to point out: you're assuming too much of a field that's not yours. Would you next suggest to save on logicians' salaries since what they're doing is not serious enough for you (assuming that you know what they're doing)?

    1) He has asked it clearly enough for people to answer. If you're not satisfied with his wording, I think it's more of your problem than his, and you could have edited it. To me his wording is OK, and surely my comment contained nothing about your (1) which you're trying to ascribe to me.

    2) If you read my comment carefully, I'm giving reasons why I think the question should be reopened (and should not have been closed) and I'm suggesting a way to prevent such problems in the future. This is not called "reprimanding". By voting to close the question you asserted that you do know it better than the OP whether his question is relevant or not; now when I'm asking you why is it that you think that you know it better, your answer is to brand this as "reprimanding". Did I get it right?
  8.  
    Harry, thanks that you did elaborate on your answer:

    "It seems like you're asserting that proofs in some areas of math are necessarily more rigorous (in theory) than proofs in others. I don't buy it. Even though the average paper in homological algebra may not be as rigorous as the average paper in a more foundational area, that doesn't mean that homological algebraists and mathematical logicians have a necessarily different conception of what constitutes a rigorous proof when push comes to shove."

    No, that's not what I asserted. You need to replace "more rigorous" by "subject to more proof-theoretic scrutiny" and then that will be right.

    Added later: For instance, have you heard of such a thing as "Reverse Mathematics"? This is all about analyzing proofs of various theorems, mostly in foundations, and I don't think they've got to look at any results of homology algebra yet. But theirs is a very low-level analysis: they just want to see exactly which axiom-like statements were used and things like how many quantifiers and what type of quantifiers are involved. If you care if Riemann Hypothesis might be independent of ZFC, ask these people: they might be able to prove that it's not independent just because its statement is sufficiently simple.

    Proof theory is something of a higher-level analysis, and just to start doing such analysis on some specific proof (rather than a theoretical analysis of an unspecified proof) the first thing you need is to have your proof written up in a way that the OP speaks about. (Only he doesn't want a fully-fledged computer language, which could also work.)

    To summarize: you won't need your homology algebra proofs expanded in the way the OP talks about any time soon just because proof theorists are not that interested (yet?) to understand homology algebra proofs. But once they do want to work on some specific proofs, they do need such expansion. Take any textbook of proof theory, and quite likely in the opening pages you'll find a discussion of the difficulty that such expansions tend to be long and barely readable in practice, and still one has to work with them - usually with aid of the computer. So what's wrong with the OP wondering just how far he can go without a computer?

    I might be missing something (then hopefully proof theorists would correct me) but it seems to me that your voting to close the question as "not a real question" is almost equivalent to a suggestion to stop discussing proof theory at Mathoverflow because it's "not a real theory"...
    • CommentAuthorHarry Gindi
    • CommentTimeFeb 15th 2011 edited
     

    "Such questions exist?" - Harry, that's exactly the problem I tried to point out: you're assuming too much of a field that's not yours. Would you next suggest to save on logicians' salaries since what they're doing is not serious enough for you (assuming that you know what they're doing)?

    Yes, that was a joke. I was parodying your caricature of non-logicians.

    No, that's not what I asserted. You need to replace "more rigorous" by "subject to more proof-theoretic scrutiny" and then that will be right.

    Could you explain what such scrutiny would entail? It sounds like word-games to me. If you're not a logician, how could you assert such a thing anyhow?

    • CommentAuthorHarry Gindi
    • CommentTimeFeb 15th 2011 edited
     

    @an_mo_user: Joel's comment addresses only the existence of software that does this. It doesn't address the main focus of the question, which is philosophical rather than technical (whence came the soft-question tag.)

    If the question were posed in a way that made it evident that it was a software request, I wouldn't have voted to close it.

    • CommentAuthoran_mo_user
    • CommentTimeFeb 15th 2011
     
    @Harry Gindi: Thank you for the response.
    While I am tempted to participate in the discussion, as said, I won't.
    My remark was really meant purely as a pointing out a fact
    (somebody might or might not have overlooked); not as an indirect voicing of
    an opinion.
  9.  
    "Could you explain what such scrutiny would entail? It sounds like word-games to me"

    See above. Once again, that's the problem: something sounds to you, but then other people can't discuss their "word-games" on MathOverflow. Are you aware that a chasing in a diagram of acyclic fibrations and cofibrations would be a very good example of what should sound like a word-game to most non-mathematicians?

    Added later: more about "what such scrutiny would entail". (Probably, nothing for the purposes of homology algebra, or safer to say, about as much as homology algebra itself would entail for the purposes of farming.)

    Basically you want me to explain what proof theory is about. At this level of generality, my answer is not going to give you more than if you ask google, Wikipedia, Mathscinet and finally Mathoverflow. One of a few aspects that I'm most interested in is its relations to higher categories and homotopy theory, more about which can be found here: http://mathoverflow.net/questions/3776
    In fact, if you scroll down to my answer and follow my link to Voevodsky's page, you'll see that he's in fact doing his *homotopy* computations in Coq, one of those full-fledged programming languages that the OP wants to avoid for some reason.
    • CommentAuthorHarry Gindi
    • CommentTimeFeb 15th 2011 edited
     

    @Sergey:

    Added later: For instance, have you heard of such a thing as "Reverse Mathematics"? This is all about analyzing proofs of various theorems, mostly in foundations, and I don't think they've got to look at any results of homology algebra yet. But theirs is a very low-level analysis: they just want to see exactly which axiom-like statements were used and things like how many quantifiers and what type of quantifiers are involved. If you care if Riemann Hypothesis might be independent of ZFC, ask these people: they might be able to prove that it's not independent just because its statement is sufficiently simple.

    Yeah, but the problem posed by the OP has nothing to do with this sort of thing.

    Proof theory is something of a higher-level analysis, and just to start doing such analysis on some specific proof (rather than a theoretical analysis of an unspecified proof) the first thing you need is to have your proof written up in a way that the OP speaks about. (Only he doesn't want a fully-fledged computer language, which could also work.)

    That sounds like the opposite of proof theory to me. Proof theory takes the syntactic data of a proof and turns it into a combinatorial problem. What you're talking about is some kind of semantic analysis.

    To summarize: you won't need your homology algebra proofs expanded in the way the OP talks about any time soon just because proof theorists are not that interested (yet?) to understand homology algebra proofs. But once they do want to work on some specific proofs, they do need such expansion. Take any textbook of proof theory, and quite likely in the opening pages you'll find a discussion of the difficulty that such expansions tend to be long and barely readable in practice, and still one has to work with them - usually with aid of the computer. So what's wrong with the OP wondering just how far he can go without a computer?

    It's not an interesting mathematical question. We're human beings, not computers, and we can infer meaning from context. Computers must check that every statement follows from the previous one by means of a previously specified transformation.

    I might be missing something (then hopefully proof theorists would correct me) but it seems to me that your voting to close the question as "not a real question" is almost equivalent to a suggestion to stop discussing proof theory at Mathoverflow because it's "not a real theory"...

    My impression is that logicians don't get all that caught up in producing fully formal computer-verifiable proofs of their results.

    • CommentAuthorHarry Gindi
    • CommentTimeFeb 15th 2011 edited
     

    See above. Once again, that's the problem: something sounds to you, but then other people can't discuss their "word-games" on MathOverflow. Are you aware that a chasing in a diagram of acyclic fibrations and cofibrations would be a very good example of what should sound like a word-game to most non-mathematicians?

    Again, you misunderstood. I was saying that you were playing word games by trying to make a distinction between "logical rigour" and "proof-theoretic scrutiny", which I don't think means anything. If a proof is rigorous, it should survive any scrutiny, proof-theoretic or not.

  10.  

    Since there were already three votes to reopen, I used my moderator powers to supply the remaining two votes. In view of this discussion, I feel that I need to justify my decision.

    It seems to me that most of the controversy is based on subjective readings of the question. Even I was guilty of that when I first read the question. However, after careful scrutiny, I came to the realization that this is actually a good question. Indeed, the question asks whether there is a bridge between standard mathematical proofs and formal proofs as defined in logic. The answer to that question is a non-obvious 'yes'.

  11.  
    "I was saying that you were playing word games by trying to make a distinction between "logical rigour" and "proof-theoretic scrutiny", which I don't think means anything. If a proof is rigorous, it should survive any scrutiny, proof-theoretic or not."

    Harry: the purpose of the scrutiny is NOT to check that the proof is correct. That might be the purpose of scrutiny of proofs by computer science people who write those proof verifiers, but on reading books and papers in proof theory (which was a pretty specific selection, not representing all of proof theory) I never encountered this idea. The purpose is to understand proofs (which are already long known to be correct), understand relations between different proofs of the same theorem, and many other things I guess. Serious proof theory started with Hilbert and Gentzen (proof theorists would correct me if forgot someone); you can refer to the Hilbert-Bernays book and to Gentzen's papers to see that proof theory has very little, if anything, to do with verifying validity of proofs.
  12.  
    "That sounds like the opposite of proof theory to me. Proof theory takes the syntactic data of a proof and turns it into a combinatorial problem. What you're talking about is some kind of semantic analysis."

    In devising various formal deduction systems and proving their soundness, equivalence, etc. one is obviously doing a semantic analysis.

    What you're saying I don't get. Is your combinatorial problem something that is supposed to be equivalent to the validity of a proof? Note that I'm giving references and you're not.