home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math:17338 rec.puzzles:8103
- Newsgroups: sci.math,rec.puzzles
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!wupost!mont!mont!stephen
- From: stephen@mont.cs.missouri.edu (Stephen Montgomery-Smith)
- Subject: Re: Integral Puzzle (Cute, not Evil)
- Message-ID: <stephen.725068342@mont>
- Organization: University of Missouri
- References: <1992Dec19.011244.2780@Csli.Stanford.EDU> <1h4hnuINNnec@function.mps.ohio-state.edu> <isopi.724948656@acf9>
- Date: 22 Dec 92 23:52:22 GMT
- Lines: 45
-
- In <isopi.724948656@acf9> isopi@acf9.nyu.edu (Marco Isopi) writes:
-
- >edgar@function.mps.ohio-state.edu (Gerald Edgar) writes:
-
-
- >>In more flowery language: given the L_infinity and L_1 norms of f, estimate
- >>the L_2 norm of f.
-
- >And then Riesz-Thorin theorem shold bring you home.
- >Or there are other and better ways?
-
- >> Gerald A. Edgar Internet: edgar@mps.ohio-state.edu
-
- >Marco Isopi
-
- If you use the real interpolation method instead, you get a stronger result:
-
- || f ||_{2,1} <= c sqrt( ||f||_1 ||f||_infty ).
-
- Here ||f||_{2,1} is a so called Lorentz space. It is defined:
-
- ||f||_{2,1} = int_0^infty sqrt(measure(|f|>t)) dt .
-
- One can give a direct proof of the above inequality to get c = 1:
-
- ||f||_{2,1} = int_0^||f||_infty sqrt(measure(|f|>t)) dt
- <= (int_0^||f||_infty 1 dt)^(1/2)
- (int_0^||f||_infty measure(|f|>t) dt)^(1/2)
- = sqrt( ||f||_1 ||f||_infty ) .
-
- The quantity ||f||_{2,1} is always larger than or equal to ||f||_2, and
- usually is not equal (equal only when |f| takes two values one of which is 0).
-
- If you are interested in finding out more about Lorentz spaces, here are
- some references.
-
- C.~Bennett and R.~Sharpley,\sl\ Interpolation of Operators,\rm\
- Academic Press 1988.\cr
- J.~Bergh and J. L\"ofstr\"om,\sl\ Interpolation Spaces,\rm\
- Springer-Verlag 1976.\cr
- R.A.~Hunt,\rm\ On $L(p,q)$\ spaces,\sl\ L'Enseignement Math.\ (2)
- {\bf 12} (1966), 249--275.\cr
-
- Stephen
-
-