• 検索結果がありません。

, II{ Vk }llr(E)=

N/A
N/A
Protected

Academic year: 2021

シェア ", II{ Vk }llr(E)="

Copied!
5
0
0

読み込み中.... (全文を見る)

全文

(1)

THE VON NEUMANN-JORDAN CONSTANT

FOR lpn(Lp)-SPACES

By

Mikio KATo and Ken-ichi MiyAzAKi*)

(Received November 30, 1992)

Introduction

In our recent work [9] the authors gave an interpolation theoretical proof of the generalized Clarkson inequalities ([6] ; see also [7], [10], [8]). The von Neumann-- Jordan constant (for a Banach space) ([5]) is closely related with these inequalities.

It is well-known that Hilbert spaces are characterized as those Banach spaces with this constant 1. 'Besides his celebrated "Clarkson's inequalities" ([2]), J. A. Clark- son showed in [3] that this constant for the Lebesgue space Lp(O, 1) has the value 22max(i/P•i/P')-i. The aim of this paper is, applying our argument in [9] , to show that for general Lp =Lp(X, .X, pt) the von Neumann-Jordan constant for lp"(Lp) (Lp- valued lp"-space) has the above value. The same are obtained for Lp and tp(Lp) as its corollaries by letting n = 1 and n . oo respectively.

Preliminaries

Let Lp==Lp(X, X, pt), 1 ;S P gES oo, be the usual Lp-space on an arbitrary but fix- ed measure space (X, .X, pt). For a Banach space E, let lr"(E), 1$r S oo,be the space of E-valued sequences {xk} of length n with the norm

, II{ Vk }llr(E)=

(,Z"=,llx,Llr)'ir if 1 ,:s .Åq ..,

max 11 xk ll if r= oo.

ISkSn

, The space of all sequences {xk} in E with Il{xk}11r(E)Åqoo is denoted by lr(E), where the norm is as above with n= oo. For E=Lp, we write simply 11 ' llr(p) for ll ' 11r(Lp)

in each case.

The von Neumann-Jordan constant for a Banach space E is defined to be the smallest constant C =C(E) satisfying

*) The authors are panially supported by the Grant-in-Aid for Scientific Research from the

Ministry of Education, Science and Culture (04640172, 1992) .

(2)

(i) -23-s llXf("Y.llft,'.11Li!l,ii/,)Y ll2 ;s c

for all xand y in E with llxll2+11yj12;O. As is easily seen, if C is best possible in the right-hand side inequality of (1), then so is 1/C in the left For any Banach space E, 1SC(E) S2; and it is a Hilbert space if and only if C(E)= 1 (Jordan and von Neumann [5]). For E =Lp(O, 1), C(Lp(O, 1))= 22MaX`'iP•'!P""i (Clarkson

[3] ; see also [4], 5. 15).

The von Neumann-Jordan constant for the space lp"(Lp)

We first prepare our tools concerning the complex method of interpolation.

LEMMA 1 ([1], Theorm 5.1.1). Let 1 SPo, PiÅq oo and 1 /P =(1-e) /Po+0/Pi

?vith OÅqeÅq1. Then,

(L p, , Lp ,) [e] = Lp (equal norm s).

LEMMA 2 ([1], Theorems 5.1.2, 4.1.2 and 4.2.1). Let (Xo, Xi) be a compatible co ap le of Ba nach spa ces. Let 1 S Po, Pi S oo (Pi Åq oo ) and 1 /P = (1 - e) /Po + e/Pi

with OÅqeÅq1. Then,

( lp", (Xo), lp", (Xi))[ei=lp"((Xo, Xi) [e]) (egttal norms?.

Let further (Yo, Yi) be another compatible couple of Banach spaces and 1/q=

(1-e)/qo+e/qi (1 $ qo, qi S oo, not qo=qi == oo). Let lpn,(x,) - l,n,(y,)

T:

tp",(Xi) -. I,",(]Yi)

with the norms Mo andMi respectively. Then, T: l7((Xo, Xi)[e]) ' l:((Yo, Yi)[e]) with the norm M satisforing

M$ M,i-eMe.

in the following, iet A = (l -l )

LEMMA 3. Let 1 ;ll P $ 2 and 1/P +1/P' = 1. Then,

(2) llA : lp2(lpn (Lp)). I;t(l,n (L,))11 =2iip'.

(3)

PRooF. For the cases where P==1 or 2, (2) is easily calculated as follows:

Letting F, GE lp"(Lp) (P = 1, 2), we have (3) M, = llA : t,2 (l,n (L,)) ---År lk(l,n (L,))ll max (Il F + G lliq), IlF ' G lli(i))

11Fll,,.S +"

IIGP

II.. 7Eo 11Flli(i)+11GIIia)

:1

(4) M, == llA : tg (l,n (L,))- lg(tg (L,))ll

(llF + G 11Z,,, + llF - G II3,,,)ii2 = llFll3,,,S+"11GPIIi,,, io (llFl13,,, + IIGII3,,,)ii2

= 2112

by the parallelogram law.

Let now 1 ÅqP Åq 2. Put e= 2/P' (O ÅqeÅq 1). Then, since (Li,L2)[e]=Lp (equal norms)

by Lemma 1, we have by Lemma 2 with Xo= Yo=l?(Li) and Xi= Yi=lg(L2) (li(l?(Li)),l22(IZ(L2))[ei=lp2(lp"(Lp)) (equalnorms)

and

(lk(l?(Li)), l22(l2" (L2))[e] = t3r(lp" (Lp)) (equal norms).

By Lemma2 again with (3) and (4) we obtain

n

llA : t;(lpn(Lp)) - l;r(lpn(Lp))ll s Ml-eMq, = 2iiP'.

. THEoREM. Let1$PSoo. Then, thevonNeumann-tJordanconstantforlp"(Lp)

and lp (Lp) is 22 M"X (i/P' i/P') -i, where 1/p + 1/p' = 1.

PRooF. Wefirstshow

(5) C(t,n (L,))$22max(1/p, i/p') -1.

Let 1 $P S 2. Put E = lp" (Lp). Then, by Lemma 3

(4)

11A : l:(E) . IZ(E)ll

S llI: l3(E) . I;(E)il 11A : l3(E) . I;f(E)Il 11I: l;t(E) . I;(E)11 S 211P-112 21/P'21/2-1/p•

= 211p,

where I denote the identity operators. Consequently, we obtain for all F and G in lp"(Lp) '

(6) (ll F+G ll3(p) + ll F-G ll3(pÅr)'!2 $ 2'iP (ll Fll;(p) + Il G I13(p)) '/2, which follows C(lp" (L,)) $ 22!P-i.

Let 2 ;$ P:S oo. Then,

llA : tg(l,n (L,)) - lg(l: (L,))ll = llA : l3(l:r (Lpr)) --" IZ(t:r (Lp•))ll $ 2ilp,.

Therefore, we have C(lp"(Lp))$ 22iP'-' in the same way as the preceeding case.

We now show that equality is attained in (5). Take disjoint measurable sets Si and S2 of finite positive measure and put f==Xs,, g={pt(Si)/pt(S2)}'iPXs,, where Xs, is the characteristic functions of Sj. If 1 $P$ 2, put F=:(f, O, "', O) and G=

(g, O, •••, O) E tp"(Lp). Then, we have

ll F+G l13(p) + ll F-G ll;(p) = llf+g l13 + ll f-gH;

= 2 • 22/Ppt(S,)2iP - 22iP(llfll3 + llgl13)

: 22iP (11 F l13(p) + 11 G II3(p)),

which implies C(lp"(Lp))=22iP-'. If 2SPS oo, a duality argument derives our conclusion. (Or we have only to take F=((f+g)/2, O, "',O) and G =((f-g)/2, O, "', O).) This completes the proof for IZ(Lp).

For lp(Lp), observe that the inequality (6) remains valid for all I7 and G in

lp(Lp) (let n.oo). The rest of the proof goes in the same way.

(5)

References

[ 1 ] J. Bergh and J. L6fstr6m, Interpolation spaces, Springer-Verlag, Berlin-Heidelberg-New York, 1976.

[2] J. A. Clarkson, Uniformly convex spac'es, Trans. Amer. Math. Soc. 40 (1936), 396-414.

[3] J. A. Clarkson, The von Neumann-Jordan constant for the Lebesgue spaces, Ann. Math. 38 (1937), 114-115.

[4] E. Hewitt and K Stromberg, Real and abstract analysis, Springer-Verlag, New York- Heidelberg-Berlin, 1965.

[5] P. Jordan and J. von Neumann, On inner products in linear metric spaces, Ann. Math. 36 (1935), 719-723.

[ 6 ] M. Kato, Generalized Clarkson's inequalities and the norms of the Littlewood matrices, Math.

Nachr. 114 (1983),163-170.

[7] M. Koskela, Some generalizations of Clarkson's inequalities, Univ. Beograd. Publ. Elek- trotechn. Fak Ser. Mat. Fiz. No. 634-677 (1979), 89'93.

[8] L. Maligranda and L. E. Persson, On Clarkson's inequalities and interpolation, Math. Nachr.

155 (1992), 187-197.

[ 9 ] K. Miyazaki and M. Kato, On a vector-valued interpolation theoretical proof of the generalized Clarkson inequalities, preprint.

[10] A. Tonge, Random Clarkson inequalities and Lp-versions of Grothendieck's inequality, Math.

Nachr. 131 (1987),335-343.

Department of Mathematics,

Kyushu Institute of Technology

参照

関連したドキュメント

For example, a maximal embedded collection of tori in an irreducible manifold is complete as each of the component manifolds is indecomposable (any additional surface would have to

The strategy to prove Proposition 3.4 is to apply Lemma 3.5 to the subspace X := (A p,2 ·v 0 ) ⊥ which is the orthogonal for the invariant form h·, ·i p,g of the cyclic space

The optimal interpolating vector σ is known as a vector-valued Lg- spline. The authors have defined a vector-valued Lg-spline to be the solu- tion of a variational

[3] Chen Guowang and L¨ u Shengguan, Initial boundary value problem for three dimensional Ginzburg-Landau model equation in population problems, (Chi- nese) Acta Mathematicae

By the algorithm in [1] for drawing framed link descriptions of branched covers of Seifert surfaces, a half circle should be drawn in each 1–handle, and then these eight half

We consider a bitangential interpolation problem for operator- valued functions defined on a general class of domains in C n (including as particular cases, Cartan domains of types I,

Applying the representation theory of the supergroupGL(m | n) and the supergroup analogue of Schur-Weyl Duality it becomes straightforward to calculate the combinatorial effect

To be specic, let us henceforth suppose that the quasifuchsian surface S con- tains two boundary components, the case of a single boundary component hav- ing been dealt with in [5]