Not signed in (Sign In)

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

Is this question acceptable?: Question on SIGGRAPH paper on Mass spring model for hair

Bottom of Page

1 to 22 of 22

  1.  
    • CommentAuthorNilima
    • CommentTimeMay 23rd 2011
     
    Dear MO Meta
    As a fairly new user, I'd like to understand the rationale behind closing the question asked in:

    http://mathoverflow.net/questions/65768/a-mass-spring-model-for-hair-simulation-closed

    The paper itself, which appears in a major visualization/graphics conference, concerns questions of algorithms and efficiency for modeling hair. The stated algorithm isn't clear on a key step, and the user 'Sebastian' asks for clarification. It is certainly not obvious to me how to address his question in a manner guaranteeing both overall consistency, order of accuracy and efficiency (each hair is being modeled as a distinct entity). The authors of the paper have made a major claim, and the details of the algorithm are important to be able to prove the claim. I'd put this squarely in the realms of numerical analysis and scientific computing. Some expert out there may have encountered, and solved, a similar problem before.

    In order to understand the etiquette around MO better, I'd be grateful for any clarifications on why this is not an appropriate question, or how it could be better phrased.
    • CommentAuthorgrp
    • CommentTimeMay 23rd 2011
     

    I found the question as being a variation on "I'm stuck on a point on this paper; please read it for me and explain it to me." If the question had more relevent detail from the paper so that I did not have to read the paper to extract it, and it had the results of the questioners brainwork on it, e.g. "I tried this method but blah prevented me from using it; I thought about doing this but that is supposed to work only when conditions are satisfied that I know do not apply here", then I would look at it more closely, and attempt an answer or a redirect. As it is written, it is not clear (without referring to the link, which I have yet to do) how the symbols in the equations are related, or whether one can be used with the other. Since I do not know yet if the subject interests me, I am reluctant to spend more time on it.

    Hopefully the people who voted to close would explain their rationale for closing. As MathOverflow questions go, I've seen better and more defensible ones than this closed, and I am not so sad to see this question get closed. This question, as is the case with many others, can be rewritten and improved to be more suitable for MathOverflow; when that happens, people can flag it for moderator attention or reopen it in other ways.

    Gerhard "Ask Me About System Design" Paseman, 2011.05.23

    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011 edited
     
    Dear Nilima,
    As Gerhard says, questions that say "read the paper at this link" generally are unpopular on MO, not entirely dependent on the subject of the paper. If, as you say, there are questions about the validity of the results, this would be another unpopular feature. I paid no attention to the question myself, I cannot be sure.

    I would say that you have enough experience to design a more acceptable question dealing with this. The usual idea is for you to put a sample question on Meta and ask people if your question would fare better. If so, you then put your question on MO. One additional point is that you have allowed yourself to be identified, you are a real and substantial person. Questions from known people tend to get the benefit of the doubt. See my recent, apparently pretty stupid, question on "When animals attack."

    P.S. Do you have two accounts? The moderators can merge them.
    • CommentAuthorNilima
    • CommentTimeMay 23rd 2011 edited
     
    Gerhard and Will,

    there's enough detail in the stated MO question for a numerical analyst to parse it, and identify a problem in the algorithm (even before reading the paper). The notation is also quite standard in numerical analysis/scientific computing. The person asking the question has precisely nailed the difficulty with implementing or proving things about this algorithm. I'm less familiar with notation and jargon in many fields of mathematics, and there are certainly questions which employ both jargon and notation. I try to use this as an opportunity to get informed.

    Rephrasing it as Gerhard suggested would require someone to implement the method (second-guessing the original intent of the paper's authors), and then report on performance. That's quite onerous. Or (better) one would want to present a proof of what goes wrong on the method - second guessing the intent of the paper's authors. There may, however, be a 'canonical' method for fixing the algorithms, which someone on MO may know. Simulation of materials at this scale is not my area of expertise, so I'd like to hear what the experts say.

    And of course you should not feel obliged to follow the link and read it unless you are curious!

    Will, you make a fair point, it's optimal for people to identify themselves. I also like the 'when animals attack' question.

    I hope to hear from the moderators!
    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011
     
    I think I left this page without clicking on "Add your comments." I put a note in the "merge accounts" question suggesting that they merge the two accounts with your name. That may work. If they do not believe the two people are both you, hilarity ensues.

    Let's see, issues, somewhat worse with an anonymous OP, include: has this person put in sufficient work on this before asking others for their effort, is the person even qualified to discuss this material intelligently? About a disputed result, has enough time passed for experts who do not use MO to have their say? Is the amount of effort required to resolve the question equivalent to a publishable article, in which case it is all just too long for MO anyway?

    Some of these ideas depend heavily on existing frequenters of MO. In scientific computing or numerical analysis, MO does not have a huge collection of specialists, but some. My feeling is that frequency of questions, and users, is going to continue in a sort of PageRank style distribution, if a question gets no good answers, that person my leave MO. So no topic will ever truly die out on MO, but many will never get very large. In contrast, one may ask some very sophisticated algebraic geometry questions on MO and get quick, insightful answers, so new users in that field may stick around.

    So, my final concern, if you feel dissatisfied with the treatment of the hair question, please stick around anyway. Unless "sebastian" posts on Meta, you are the best bet for coming up with a more successful version of that question. Otherwise, all that is happening is that you have alerted people with at least 3000 points to the idea that casting votes to re-open the question might be a good idea. It takes five votes to re-open, and such is simply not that common an occurrence.
    • CommentAuthorNilima
    • CommentTimeMay 23rd 2011
     
    Thank you, Will, for merging my accounts!

    You raise good questions. Since I'm not an expert in the field of large-scale material simulation, I don't follow that literature very closely. Certainly Fedkiw and his co-authors are very well-known in this field (Fedkiw himself may be the only mathematician to be awarded a Sloan, a PECASE and an Academy Award), so I'm inclined to believe the result is not disputed. I just don't know how the details are filled in, in the algorithm.

    Your observation about the disciplinary distribution of users is valuable. It would indeed be a shame if some sub-disciplines languished for lack of activity, reinforced in the manner you describe. Our discipline is balkanized enough.
    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011
     
    Nilima, I do not think your two accounts are merged yet, but I made the request in the right place, it should not take more than a day or two. Note that at this time of year, moderators may be busy with graduations and so forth.

    I may have misunderstood your initial remarks about the linked article, as I tried to download the first page and lost patience. You were not so much saying that there were doubts about the results among other experts, rather that the OP had some trouble implementing a key step where their exposition was a bit obscure. If so, that seems closer to Gerhard's description of what he found lacking in the question.
    • CommentAuthorgrp
    • CommentTimeMay 23rd 2011
     

    To make things clear, my opinion is my own, and I do not claim to have a good read on the quality of a question as viewed by a specialist in the field. I am, however, the world's foremost authority on my own opinion; I believe I have accurately reported my reaction to the question, which may (or may not) be representative of the MathOverflow community.

    Also, I do not know why the question was closed, but I do know that it is a question that could stand improvement. I had a suggestion for what would make it appeal more to me. Also, a really great question would be accessible to the expert and non-expert, and a good question would (in my opinion) go a little further to meet a general audience than this one has. If you use the meta.mathoverflow search tab, you can find some of the previous discussions about closed questions and why they were closed, as well as discussions on how to make a question better. You may learn more from doing that than from engaging me further in this discussion (although I don't mind the discussion).

    Gerhard "Ask Me About System Design" Paseman, 2011.05.23

    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011
     
    Gerhard, I think i figured out the cuboid animals, why the Music Man reference?
    • CommentAuthorRyan Budney
    • CommentTimeMay 23rd 2011 edited
     
    I voted to close largely because all the onus was put on the reader to read the paper, to figure out what the OP was asking about before anything else. If the author made an attempt to parse the key ideas like the actual differential equation involved, or the ideas behind the original DE that would be a great help -- for example, specifying all the notation. But it reads more like a coding question, which isn't really an MO topic. We've kept similar questions open before (and closed many similar questions) but there's a strong impetus on the OP to make the question as self-contained and compelling as possible. These kinds of questions tend to be kind of borderline. I was maybe a little more trigger-happy than usual because it seemed like there was a glut of not-so-good questions today. Maybe I'm just grumpy.
    • CommentAuthorNilima
    • CommentTimeMay 23rd 2011
     
    Thanks, Ryan.

    This is evidently an issue of sub-disciplines not communicating fluently. To clarify: I believe most numerical analysts would know the equation of a mass spring (which is referred to in the title). Likewise, the notation is considered standard in the field.

    I think that to start this question at "Consider the equation of a mass spring $F=m \frac{d^2 x}{d t^2}$', followed by a discussion of discretization strategies for each variable, the fact that hundreds of thousands of hair are involved, etc. would render the question more accessible to a wider MO readership. It also means a numerical analyst would have to wade through lots of introductory material before getting to the gist of the question. I intend this as a constructive comment. I don't think MO would expect people to define SO(n) or S_n from scratch simply because this is unfamiliar notation to some research communities. This makes questions unwieldy.

    At any rate, given the relative paucity of numerical analysis on MO (192 questions) compared to, say, algebraic geometry or number theory, your point is well taken. The better represented areas share much notation in common. The general readership is perhaps not familiar with the verbiage, models and notation of numerical analysis, and given this, the question indeed may appear to put the onus on the reader.

    This is natural, but demoralizing for the numerical analyst. After all, one should be able to post quick questions aimed at numerical-PDE experts without going through a a lengthy introduction. But if without this introductory material the question is likely to get closed, one encounters a circular problem, since the numerical-PDE folks are less likely to use MO as a forum. This would be a shame in a philosophical sense, but not a disaster. I personally have no energy to address this.
    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011 edited
     
    Nilima, you have a point. There are three votes to reopen at this instant, after mine. One substitute for introductory material in each question may be a backlog of questions. I don't much care for the question, it is one of many styles I see that tell me the OP is never going to answer anybody else's question. But I can also see this as contributing to the presence of an underrepresented discipline on MO.

    Make that four votes, 21:38 Pacific time.
    • CommentAuthorRyan Budney
    • CommentTimeMay 23rd 2011 edited
     
    Hi Nilima,

    It's true some people would use S_n or SO(n) without comment but I think many posters try to say at least "consider the symmetric group on n letters S(n)" before the first usage of the symbol in an actual formula unless the context was made clear in a previous post.

    If you look at the really good question-askers, they do provide loads of detail. IMO this is a good thing. A well-written question will get a lot of attention by a wide-variety of people, and poorly-written questions tend to be at risk of closure. Poorly-written questions in over-represented areas tend to be cut more slack, if only because the people reading the forum are looking for math to talk about.

    In that regard I think it's not so much a prejudice against numerical analysis you're seeing, as an over-exuberance for the various highly-represented subject areas here. Really they're different!

    The more borderline a question or the less expertise we have here, generally the more demanding we are of the question-asker. I'm not saying this is how it should be, but that's how it works in practice. Perhaps after reading a few more well-written numerical analysis questions more of the audience will be familiar with the terminology and this prejudice will slowly evaporate.

    -ryan
    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011
     
    Open now, 21:58 Pacific.

    It did not move to the front of the Active ordering, I was not sure.
  15.  

    I've reopened the question. I think if you're reading a paper (whether in algebraic geometry or numerical analysis), and one or two steps are unclear, it may be reasonable to ask MathOverflow, just in case someone with more expertise knows how it is done. Assuming the person asking has nontrivial background and has thought about the problem without resolution, this is perhaps an ideal MO use case. I suppose closure could have been prevented if the person asking had said more forcefully or explicitly that those particular steps were unclear in the paper.

  16.  

    (Nilima, you didn't mention Fedkiw's strength-related achievements, like bench pressing 250 kilos and squatting 350.)

    • CommentAuthorNilima
    • CommentTimeMay 23rd 2011 edited
     
    Ryan,
    Thanks again for responding!

    I certainly don't believe there's an intentional prejudice against numerical analysis. I *do* think that, given the relative representations of sub-disciplines, that there is a difference between what is considered an appropriate amount of background and detail, or how exacting the readership may be. This is simply an artifact of how
    MO is or has evolved.

    I'm personally a tad disappointed, mostly because it appears that silos have a way of persisting across media. Someone with more energy than me will, I hope, come up with well-written NA questions. As Izabella Laba memorably posted on her blog, (and I'm citing in entirely a different context),

    'I can’t afford to divert much time and energy to establishing my credentials in yet another community that might or might not turn out well for me, and in a Catch-22 situation, I can’t expect to get established there if I don’t divert enough time and energy.'

    I don't even see why any of this needs to change. MO is what it is, and individuals should decide whether their needs are met by participating in it! I personally find it a useful forum to sneak a peek at what's going on in other sub-disciplines, and am most certainly not attempting to become Queen Defender of Numerical Analysis.
    • CommentAuthorNilima
    • CommentTimeMay 23rd 2011
     
    Whoops - my apologies, I did not realize the question had been reopened. Thanks to all. Thanks also, Scott- I had no idea Fedkiw lifted weights. One more reason for me to be intimidated.
    • CommentAuthorWill Jagy
    • CommentTimeMay 23rd 2011
     
    Nilima, just read stuff I post. Nobody else is important. I also edited a couple of words in the question so it moved to the front of Active.
  20.  
    Hi Nilima. I just notice your posting on Laba's blog. So then I Googled you. I see we're neighbours! Hello from Victoria and welcome to MO, although I see you've been here for some time. :) It's nice to see more people from western Canadian universities about.

    -ryan
    • CommentAuthorgrp
    • CommentTimeMay 24th 2011
     

    Will, it seemed an appropriate double entendre. Also I showed the Robert Preston film to my son recently; we both have bits of it running through our heads. Also, I was not literate enough to make a more closely coupled reference. (And I really was expecting some of O'Rourke's animals to appear.)

    Gerhard "Got Your Daily Pop-Culture Reference?" Paseman, 2011.05.24

    • CommentAuthorWill Jagy
    • CommentTimeMay 24th 2011
     
    The OP has commented on his post, also identifying himself a bit better. He is an undergraduate computer science major and seems to have been reading this Meta thread.

1 to 22 of 22

  1.  
Back to Discussions Top of Page