Not signed in (Sign In)

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


    I've been stuck on a nasty technical lemma in a book I've been reading for the past week (because the proof seems to just be incorrect). Unfortunately, since it is a technical lemma very dependent on a lot of earlier material, I am pretty sure that I won't get an answer here just by giving the statement and the original proof. It seems like I would have to review a good deal of earlier material in the question here on MO.

    How should I ask this sort of question?


    One idea: If you suspect the proof is wrong, you could try to find the smallest chunk of proof whose validity you're unsure about and formulate it as a question or a request for a counterexample. If the proof shows that X is a Y, but you know for sure that not every X is a Y, determine what additional hypotheses you have on this particular X. Presumably, the hypotheses in the lemma are enough (otherwise you'd say the lemma is wrong, not just the proof). Somewhere in the middle there might be a tractable question of the form, "Is every X with properties P and Q a Y?" Of course, this recipe might need some tweaking for your specific case.

    In any case, you should probably say that you're trying to understand the proof of some result (and include as precise a reference as possible). It may turn out that there are implicit (or well-hidden) hypotheses that you're missing, which would be very hard for someone to guess without knowing the exact reference.

    • CommentAuthorbbischof
    • CommentTimeAug 17th 2010
    Have you tried the obvious? Asking a professor who knows how to fix it? Or is the book/subject too obscure for that?
    • CommentAuthorKevin Lin
    • CommentTimeAug 18th 2010

    Or ask the author.

    • CommentAuthorHarry Gindi
    • CommentTimeAug 18th 2010 edited

    @Kevin: Well, I've asked the author a lot of questions already, and I told him I wasn't going to bother him for a while (since the questions were getting somewhat frequent, I wanted to avoid annoying him), and going back on something I said only a week ago seems to be pretty bad.

    @Bryan: Yep, I asked another professor by e-mail, but he said he won't have time to look at it until next week, and I also asked on the nForum, but Urs seems to be too busy with Chern-Weil theory to work out a correct proof.

    @Anton: Here is the actual question itself. It is very specific indeed. I'm not really sure how to turn it into a good MO question (I have, of course, managed to ask many questions like you suggested above, but this one seems to be particularly resistant.