Skip to main content
13 events
when toggle format what by license comment
May 27, 2016 at 18:25 vote accept StevenMurray
May 27, 2016 at 18:23 comment added StevenMurray Thanks everyone. I think I have solved it, and it doesn't require the inverse after all. @Robert 's answer led me to the solution by showing that I could construct the integral from a to b rather than some function of $u$ or its inverse. Once in that space, it is not too hard.
May 22, 2016 at 7:46 comment added Peter Kravchuk One could try introducing the inverse via a Lagrange multiplier, i.e. add a constraint $f(u(x))-x=0$, where $f$ will be the inverse. This is perhaps in the end equivalent to Robert's suggestion
May 21, 2016 at 22:21 answer added Robert Bryant timeline score: 4
May 21, 2016 at 19:31 history edited StevenMurray CC BY-SA 3.0
added 1550 characters in body
May 21, 2016 at 19:17 comment added StevenMurray Yes, a very specific application -- I guess it may turn out that I have constructed the problem incorrectly, but I don't think so. I'll edit the question with more details as to the nature of the problem
May 21, 2016 at 18:30 comment added Andreas Rüdinger For me the question does not look very "natural". The functional inverse of $u: [a,b] \to \mathbb{R}$ is not meant to be integrated over the same domain as $u$. Do you have an application in mind?
May 21, 2016 at 18:29 comment added StevenMurray Yeah, that's why I thought it might actually be impossible in a general sense. But you never know what some clever mathematician has come up with...
May 21, 2016 at 17:42 comment added Siminore It looks hard, since you need to invert $u + \varepsilon \varphi$...
May 21, 2016 at 17:10 comment added StevenMurray Ah, good point. I meant the former.
May 21, 2016 at 12:22 comment added Igor Khavkine By "inverse" $u^{-1}$, do you mean $u(u^{-1}(x)) = x$ or $u(x) u^{-1}(x) = 1$?
May 21, 2016 at 11:14 review First posts
May 21, 2016 at 12:23
May 21, 2016 at 11:13 history asked StevenMurray CC BY-SA 3.0