Talk:Superconvergence

I'm concerned about the accuracy (or at least: the universality) of this definition. My doctoral thesis was on the subject of Superconvergence in the Finite Element Method; the definition of Superconvergence appropriate to that field is "convergence to a higher order of accuracy than would otherwise be expected" (for example, effected by some combination of choice of triangulation and recovery method). Nick Levine 13:21, 24 December 2005 (UTC)

My experience is the same, "superconvergence" means "converges faster than expected". Please be bold and change the article. Don't worry too much about making mistakes, I will check what you are doing. I'm not sure what to do with the meaning of "superconvergence" described in the article (basically quadratic convergence). I've never come across it, but perhaps it's best to leave it in. -- Jitse Niesen (talk) 14:28, 24 December 2005 (UTC)

Done (to an extent). I think it would be best of the quadratic convergence definition were moved to a page on its own and referred to briefly here. I don't think we need a disambiguation page, because IMO the quadratic convergence definition is not in common use (whatever PlanetMath may have to say on the matter). However I'm open to argument, and don't intend to act on this until (a) some time has passed and (b) I've got around to fleshing out the new material. Nick Levine 18:19, 13 January 2006 (UTC)

Also, I think the existing wording of the quadratical meaning of the concept is very clumsy and inaccurate. It is much better to formulate it in terms of the numbers themselves than obscurely referring to their binary expansions. And the whole "has $2^{i}-1$ zeros" part doesn't sound right, as it does not have to have exactly this number of zeroes. Wherever the quadratic definition is decided to be placed, it should be reworded. -- Meni Rosenfeld (talk) 19:31, 19 January 2006 (UTC)