- Brascamp-Lieb inequality
In
mathematics , the Brascamp-Lieb inequality is a result ingeometry concerningintegrable function s on "n"-dimension alEuclidean space R"n". It generalizes theLoomis-Whitney inequality , thePrékopa-Leindler inequality andHölder's inequality , and is named afterHerm Jan Brascamp andElliott H. Lieb . The original inequality (called the geometric inequality here) is in [H.J. Brascamp and E.H. Lieb, "Best Constants in Young's Inequality, Its""Converse and Its Generalization to More Than Three Functions", Adv. in Math.20, 151-172 (1976).] .Its generalization, stated first, is in [E.H.Lieb, "Gaussian Kernels have only Gaussian Maximizers", Inventiones Mathematicae102, pp. 179-208 (1990).]tatement of the inequality
Fix
natural number s "m" and "n". For 1 ≤ "i" ≤ "m", let "n""i" ∈ N and let "c""i" > 0 so that:
Choose non-negative, integrable functions
:
and
surjective linear map s:
Then the following inequality holds:
:
where "D" is given by
:
Another way to state this is that the constant "D" is what one would obtain byrestricting attention to the case in which each is a centered Gaussianfunction, namely .
Relationships to other inequalities
The geometric Brascamp-Lieb inequality
The geometric Brascamp-Lieb inequality is a special case of the above, and was used by Ball (1989) to provide upper bounds for volumes of central sections of cubes.
For "i" = 1, ..., "m", let "c""i" > 0 and let "u""i" ∈ S"n"−1 be a unit vector; suppose that that "c""i" and "u""i" satisfy
:
for all "x" in R"n". Let "f""i" ∈ "L"1(R; [0, +∞] ) for each "i" = 1, ..., "m". Then
:
The geometric Brascamp-Lieb inequality follows from the Brascamp-Lieb inequality as stated above by taking "n""i" = 1 and "B""i"("x") = "x" · "u""i". Then, for "z""i" ∈ R,
:
It follows that "D" = 1 in this case.
Hölder's inequality
As another special case, take "n""i" = "n", "B""i" = id, the
identity map on R"n", replacing "f""i" by , and let "c""i" = 1 / "p""i" for 1 ≤ "i" ≤ "m". Then:
and the log-concavity of the
determinant of apositive definite matrix implies that "D" = 1. This yields Hölder's inequality in R"n"::
References
* cite book
last = Ball
first = Keith M.
chapter = Volumes of sections of cubes and related problems
title = Geometric aspects of functional analysis (1987--88)
editor = J. Lindenstrauss and V.D. Milman
series = Lecture Notes in Math., Vol. 1376
pages = pp. 251–260
publisher = Springer
location = Berlin
year = 1989
* cite journal
last=Gardner
first=Richard J.
title=The Brunn-Minkowski inequality
journal=Bull. Amer. Math. Soc. (N.S.)
volume=39
issue=3
year=2002
pages= pp. 355–405 (electronic)
issn = 0273-0979
url = http://www.ams.org/bull/2002-39-03/S0273-0979-02-00941-2/S0273-0979-02-00941-2.pdf
doi=10.1090/S0273-0979-02-00941-2
Wikimedia Foundation. 2010.