ebook img

An Interpolation Theorem for Sublinear Operators on Non-homogeneous Metric Measure Spaces PDF

0.14 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview An Interpolation Theorem for Sublinear Operators on Non-homogeneous Metric Measure Spaces

An Interpolation Theorem for Sublinear Operators on 2 Non-homogeneous Metric Measure Spaces 1 0 2 Haibo Lin and Dongyong Yang∗ n a J 0 3 Abstract. Let (X,d,µ) be a metric measure space and satisfy the so-called upper dou- bling condition and the geometrically doubling condition. In this paper, the authors ] P establish an interpolation result that a sublinear operator which is bounded from the A Hardy space H1(µ) to L1,∞(µ) and from L∞(µ) to the BMO-type space RBMO(µ) is . alsoboundedonLp(µ)forallp∈(1, ∞). Thisextensionisnotcompletelystraightforward h t and improves the existing result. a m 1 Introduction [ 1 v Spaces of homogeneous type were introduced by Coifman and Weiss [3] as a general 7 framework in which many results from real and harmonic analysis on Euclidean spaces 9 have their natural extensions; see, for example, [4, 5, 6]. Recall that a metric space (X, d) 0 6 equipped with a nonnegative Borel measure µ is called a space of homogeneous type if . (X, d, µ) satisfies the following measure doubling condition that there exists a positive 1 0 constant C , depending on µ, such that for any ball B(x, r) ≡{y ∈ X : d(x, y) < r} with µ 2 x ∈ X and r ∈ (0, ∞), 1 : 0 < µ(B(x, 2r)) ≤ Cµµ(B(x, r)). (1.1) v i X Themeasuredoublingcondition(1.1)playsakeyroleintheclassicaltheoryofCalder´on- r Zygmund operators. However, recently, many classical results concerning the theory of a Calder´on-Zygmund operators and function spaces have been proved still valid if the dou- bling condition is replaced by a less demanding condition such as the polynomial growth condition; see, for example [14, 16, 17, 15, 18] and the references therein. In particular, let µ be a non-negative Radon measure on Rn which only satisfies the polynomial growth condition that there exist positive constants C and κ ∈(0,n] such that for all x ∈ Rn and r ∈ (0, ∞), µ({y ∈ Rn : |x−y| < r})≤ Crκ. (1.2) Such a measure does not need to satisfy the doubling condition (1.1). We mention that theanalysis withnon-doublingmeasuresplayed astrikingroleinsolvingthelong-standing open Painlev´e’s problem by Tolsa in [18]. 2010 Mathematics Subject Classification. Primary 42B35; Secondary 42B25, 47B38. Key words and phrases. upperdoubling, geometrically doubling, RBMO(µ), sublinear, interpolation. ∗Corresponding author. 1 Because measures satisfying (1.2) are only different, not more general than measures satisfying (1.1), the Calder´on-Zygmund theory with non-doubling measures is not in all respects a generalization of the corresponding theory of spaces of homogeneous type. To include the spaces of homogeneous type and Euclidean spaces with a non-negative Radon measure satisfying a polynomial growth condition, Hyto¨nen [8] introduced a new class of metric measure spaces which satisfy the so-called upper doubling condition and the geometrically doubling condition (see, respectively, Definitions 2.1 and 2.2 below), and a notion of the regularized BMO space, namely, RBMO(µ) (see Definition 2.4 below). Since then, more and more papers focus on this new class of spaces; see, for example [11, 12, 10, 1, 9, 7, 13]. Let(X, d, µ)beametricmeasurespacesatisfyingtheupperdoublingconditionandthe geometrically doubling condition. In [10], the atomic Hardy space H1(µ) (see Definition 2.5 below) was studied and the duality between H1(µ) and RBMO(µ) of Hyto¨nen was established. Some of results in [10] were also independently obtained by Anh and Duong [1] via different approaches. Moreover, Anh and Duong [1, Theorem 6.4] established an interpolation result that a linear operator which is bounded from H1(µ) to L1(µ) and from L∞(µ) to RBMO(µ) is also bounded on Lp(µ) for all p ∈ (1, ∞). The purpose of this paper is to generalize and improve the interpolation result for linear operators in [1] to sublinear operators in the current setting (X, d, µ), which is stated as follows. Theorem 1.1. Let T be a sublinear operator that is bounded from L∞(µ) to RBMO(µ) and from H1(µ) to L1,∞(µ). Then T extends boundedly to Lp(µ) for every p ∈(1, ∞). In Section 2, we collect preliminaries we need. In Section 3, for r ∈ (0,1), we first ♯ show that the maximal function M (f), which is a variant of the sharp maximal function r M♯(f)in[1],isboundedfromRBMO(µ)toL∞(µ),thenweestablishaweaktypeestimate between the doubling maximal function N(f) and M♯(f), and we also establish a weak typeestimate for N (f)with r ∈ (0,1), avariant of N(f). Usingthese results we establish r Theorem 1.1. We remark that the method for the proof of Theorem 1.1 is different from that of [1, Theorem 6.4]. Precisely, in the proof of [1, Theorem 6.4], the fact that the composite operator M♯◦T of the sharp maximal function M♯ and a linear operator T is a sublinear operator was used. However, as far as we know, when T is sublinear, whether the composite operator M♯ ◦T is a sublinear operator is unclear and so the proof of [1, Theorem 6.4] is not available. Throughoutthis paper, we denote by C a positive constant which is independentof the main parameters involved but may vary from line to line. The subscripts of a constant indicate the parameters it depends on. The notation f . g means that there exists a constant C > 0 such that f ≤ Cg. Also, for a µ-measurable set E, χ denotes its E characteristic function. 2 Preliminaries In this section, we will recall some necessary notions and notation and the Calder´on- Zygmund decomposition which was established in [1]. We begin with the definition of upper doubling space in [8]. 2 Definition 2.1. A metric measure space (X, d, µ) is called upper doubling if µ is a Borel measure on X and there exists a dominating function λ : X × (0, ∞) → (0, ∞) and a positive constant C such that for each x ∈ X, r → λ(x, r) is non-decreasing, and for all λ x ∈ X and r ∈ (0, ∞), µ(B(x, r))≤ λ(x, r)≤ C λ(x, r/2). λ Remark 2.1. (i) Obviously, a space of homogeneous type is a special case of the upper doubling spaces, where one can take the dominating function λ(x, r)≡ µ(B(x, r)). More- over, let µ be a non-negative Radon measure on Rn which only satisfies the polynomial growth condition (1.2). By taking λ(x, r)≡ Crκ, we see that (Rn, |·|, µ) is also an upper doubling measure space. (ii)Itwasprovedin[10]thatthereexistsadominatingfunctionλrelatedtoλsatisfying the property that there exists a positive constant Cλe such that λ ≤ λ, Cλe ≤ Cλ, and for all x, y ∈ X with d(x, y) ≤ r, e λ(x, r) ≤ Ceλ(y, r). e (2.1) λ Basedonthis,inthispaper,wealways assumethatthedominatingfunctionλalsosatisfies e e (2.1). Throughout the whole paper, we also always assume that the underlying metric space (X, d) satisfies the following geometrically doubling condition introduced in [8]. Definition 2.2. A metric space (X, d) is called geometrically doubling if there exists some N ∈ N+ ≡ {1, 2, ···} such that for any ball B(x, r) ⊂ X, there exists a finite ball 0 covering {B(x , r/2)} of B(x, r) such that the cardinality of this covering is at most N . i i 0 The following coefficients δ(B, S) for all balls B and S were introduced in [8] as ana- logues of Tolsa’s numbers K in [16]; see also [10]. Q,R Definition 2.3. For all balls B ⊂ S, let dµ(x) δ(B, S) ≡ . λ(c ,d(x, c )) Z(2S)\B B B where and in that follows, for a ball B ≡ B(c , r ) and ρ ∈(0, ∞), ρB ≡ B(c , ρr ). B B B B p In what follows, for each p ∈ (0,∞), L (µ) denotes the set of all functions f such loc that |f|p is µ-locally integrable. Definition 2.4. Let η ∈ (1,∞) and p ∈ (0,∞). A function f ∈ Lp (µ) is said to be in loc the space RBMOp(µ) if there exists a non-negative constant C and a complex number f η B for any ball B such that for all balls B, 1 |f(y)−f |pdµ(y) ≤ Cp B µ(ηB) ZB and that for all balls B ⊂ S, |f −f |≤ C[1+δ(B, S)]. B S Moreover, the RBMOp(µ) norm of f is defined to be the minimal constant C as above η and denoted by kfkRBMOpη(µ). 3 When p = 1, we write RBMO1(µ) simply by RBMO(µ), which was introduced by η Hyto¨nen in [8]. Moreover, the spaces RBMOp(µ) and RBMO(µ) coincide with equivalent η norms, which is the special case of [7, Corollary 2.1]. Proposition 2.1. Let η ∈ (1,∞) and p ∈ (0,∞). The spaces RBMOp(µ) and RBMO(µ) η coincide with equivalent norms. Remark 2.2. It was proved in [8, Lemma 4.6] that the space RBMO(µ) is independent of the choice of η. By this and Proposition 2.1, it is obvious that the space RBMOp(µ) is η independent of the choice of η. We now recall the definition of the atomic Hardy space introduced in [10]; see also [1]. Definition 2.5. Let ρ ∈ (1,∞) and p ∈ (1, ∞]. A function b ∈ L1(µ) is called a (p, 1) - λ atomic block if (i) there exists some ball B such that supp(b) ⊂ B; (ii) b(x)dµ(x) = 0; X (iii) fRor j = 1, 2, there exist functions a supported on balls B ⊂ B and λ ∈ C such j j j that b = λ a +λ a , 1 1 1 2 and ka k ≤ [µ(ρB )]1/p−1[1+δ(B , B)]−1. j Lp(µ) j j Moreover, let |b| ≡ |λ |+|λ |. H1,p(µ) 1 2 atb A function f ∈L1(µ) is said to belong to the atomic Hardy space H1,p(µ) if there exist atb (p, 1)λ-atomic blocks {bj}j∈N such thatf = ∞j=1bj and ∞j=1|bj|H1,p(µ) < ∞. Thenorm atb 1,p of f in H (µ) is defined by atb P P kfk ≡ inf |b | , H1,p(µ)  j H1,p(µ) atb atb Xj  where the infimum is taken over all the possible decompositions of f as above. Remark 2.3. It was proved in [10] that for each p ∈ (1, ∞], the atomic Hardy space 1,p 1,p H (µ) is independent of the choice of ρ, and that for all p ∈ (1, ∞), the spaces H (µ) atb atb 1,∞ 1,p and H (µ) coincide with equivalent norms. Thus, in the following, we denote H (µ) atb atb simply by H1(µ). At the end of this section, we recall the (α, β)-doubling property of some balls and the Calder´on-Zygmund decomposition established by Anh and Duong [1, Theorem 6.3]. Given α, β ∈ (1, ∞), a ball B ⊂ X is called (α, β)-doubling if µ(αB) ≤ βµ(B). It was log α proved in [8] that if a metric measure space (X, d, µ) is upper doubling and β > C 2 ≡ λ 4 αν, then for every ball B ⊂ X, there exists some j ∈ Z ≡ N ∪ {0} such that αjB is + (α, β)-doubling. Moreover, let (X, d) begeometrically doubling, β > αn with n ≡ log N 2 0 and µ a Borel measure on X which is finite on boundedsets. Hyto¨nen [8] also showed that for µ-almost every x ∈ X, there exist arbitrarily small (α, β)-doubling balls centered at x. Furthermore, the radius of these balls may be chosen to be of the form α−jr for j ∈ N and any preassigned number r ∈ (0, ∞). Throughout this paper, for any α ∈ (1, ∞) and ball B, Bα denotes the smallest (α, β )-doubling ball of the form αjB with j ∈ Z , where α + β ≡max{αn, αν}+30n +30ν = αmax{n,ν}+30n +30ν. (2.2) α e Lemma 2.1. Let p ∈ [1, ∞), f ∈ Lp(µ) and ℓ ∈ (0,∞) (ℓ > ℓ ≡ γ kfk /µ(X) if 0 0 Lp(µ) µ(X) < ∞, where γ is any fixed positive constant satisfying that γ > max{C3log26, 63n}, 0 0 λ C is as in (2.2) and n = log N ). Then λ 2 0 (i) there exists an almost disjoint family {6B } of balls such that {B } is pairwise j j j j disjoint, 1 ℓp |f(x)|pdµ(x) > forallj, µ(62B ) γ j ZBj 0 1 ℓp |f(x)|pdµ(x) ≤ foralljandallη > 1, µ(62ηB ) γ j ZηBj 0 and |f(x)| ≤ ℓ forµ−almosteveryx ∈ X \(∪ 6B ); j j (ii) for each j, let S be a (3×62, Clog2(3×62)+1)-doubling ball concentric with B sat- j λ j isfying that r > 62r , and ω ≡ χ /( χ ). Then there exists a family {ϕ } of Sj Bj j 6Bj k 6Bk j j functions such that for each j, supp(ϕ ) ⊂ S , ϕ has a constant sign on S and j j j j P ϕ (x)dµ(x) = f(x)ω (x)dµ(x), j j ZX Z6Bj |ϕ (x)| ≤ γℓ forµ−almosteveryx ∈ X, j j X where γ is some positive constant depending only on (X, µ), and there exists a positive constant C, independent of f, ℓ and j, such that kϕjkL∞(µ)µ(Sj) ≤C |f(x)ωj(x)|dµ(x), ZX and if p ∈ (1, ∞), 1/p |ϕ (x)|pdµ(x) [µ(S )]1/p′ ≤ C |f(x)ω (x)|pdµ(x); j j ℓp−1 j (ZSj ) ZX (iii) if for any j, choosing S in (ii) to be the smallest (3×62, Clog2(3×62)+1)-doubling j λ ball of (3×62)B , then h ≡ (fω −ϕ ) ∈ H1,p(µ) and there exists a positive constant j j j j atb C, independent of f and ℓ, such that P C p khk ≤ kfk . Ha1t,bp(µ) ℓp−1 Lp(µ) 5 3 Proof Theorem 1.1 To prove Theorem 1.1, we also need some maximal functions in [8, 1] as follows. Let f ∈ L1 (µ) and x ∈ X. The doubling Hardy-Littlewood maximal function N(f)(x) and loc the sharp maximal function M♯(f)(x) are respectively defined by setting, 1 N(f)(x) ≡ sup |f(y)|dµ(y), µ(B) B(6,β6B)−∋dxoubling ZB and 1 M♯(f)(x) ≡ sup |f(y)−me (f)|dµ(y) µ(5B) B6 B∋x ZB |m (f)−m (f)| B S + sup , 1+δ(B, S) x∈B⊂S B,S(6,β6)−doubling where for any f ∈ L1 (µ) and ball B, m (f) means its average over B, namely, m (f) ≡ loc B B 1 f(x)dµ(x). It was showed in [9, Lemma 2.3] that for any p ∈ [1, ∞), Nf is µ(B) boundZeBd from Lp(µ) to Lp,∞(µ). Lemma 3.1. Let f ∈ RBMO(µ), r ∈ (0,1) and M♯(f) ≡ [M♯(|f|r)]1/r. Then we have r M♯f ∈ L∞(µ), and moreover, r kMr♯fkL∞(µ) . kfkRBMO(µ). Proof. From Remark 2.2, we deduce that for any ball B, 1 fBe6 −mBe6(f) ≤ µ(B6)ZBe6 f(x)−fBe6 dµ(x) . kfkRBMO(µ). (cid:12) (cid:12) (cid:12) (cid:12) (cid:12) (cid:12) (cid:12) (cid:12) On the other hand, by Proposition 2.1 and Remark 2.2, we see that e kfkRBMO(µ) ∼ kfkRBMOr(µ). 5 From these facts, it follows that 1 |f(x)|r −me (|f|r) dµ(x) µ(5B) B6 ZB 1 (cid:12) (cid:12) ≤ (cid:12) |f(x)|r − me(cid:12)(f) r + me (f) r −me (|f|r) dµ(x) µ(5B) B6 B6 B6 ZB 1 (cid:2)(cid:12) (cid:12) (cid:12) (cid:12) (cid:12)(cid:12) (cid:12) (cid:12)(cid:3) . µ(5B) |f(cid:12)(x)−fB|r(cid:12) dµ(x)+(cid:12) (cid:12)fB(cid:12)−(cid:12) fBe6 r +(cid:12) fBe6 −mBe6(f(cid:12)) r ZB 1 (cid:12) (cid:12) (cid:12) (cid:12) + f(x)−fe r dµ(x)(cid:12) (cid:12) (cid:12) (cid:12) µ(B6) ZBe6 B6 (cid:12) r (cid:12) . 1+δ B,B(cid:12)6 kfkr (cid:12) . kfkr , e RBMO(µ) RBMO(µ) h (cid:16) (cid:17)i e 6 where the last inequality follows from the fact that δ(B,B6) . 1, which holds by [10, Lemma 2.1]. On the other hand, for any (6,β )-doubling balls B ⊂ S,e 6 |m (|f|r)−m (|f|r)| ≤ |m (|f|r)−|f |r|+||f |r −|f |r|+||f |r −m (|f|r)| B S B B B S S S ≤ m (|f −f |r)+|f −f |r +m (|f −f |r) B B B S S S . [1+δ(B,S)]rkfkr . RBMO(µ) Combining these two inequalities finishes the proof of Lemma 3.1. Lemma 3.2. Let p ∈ [1, ∞) and f ∈ L1 (µ) such that f(x)dµ(x) = 0 if µ(X) < ∞. loc X If for any R > 0, R sup ℓpµ({x ∈ X : N(f)(x) > ℓ})< ∞, 0<ℓ<R we then have supℓpµ({x ∈ X : N(f)(x) > ℓ}) . supℓpµ x ∈ X : M♯(f)(x) > ℓ . ℓ>0 ℓ>0 (cid:16)n o(cid:17) Proof. Recall the λ-good inequality in [1] that for some fixed constant ν ∈ (0,1) and all ǫ ∈ (0,∞), there exists some δ > 0 such that for any ℓ > 0, µ x ∈ X : N(f)(x) > (1+ǫ)ℓ, M♯(f)(x) ≤ δℓ ≤ νµ({x ∈X : N(f)(x) > ℓ}). (cid:16)n o(cid:17) From this, it then follows that for R large enough and any ǫ > 0, sup ℓpµ({x ∈ X : N(f)(x) >ℓ}) 0<ℓ<R ≤ sup [(1+ǫ)ℓ]pµ({x ∈ X : N(f)(x) > (1+ǫ)ℓ}) 0<ℓ<R ≤ ν(1+ǫ)p sup ℓpµ({x ∈ X : N(f)(x) > ℓ}) 0<ℓ<R +(1+ǫ)psupℓpµ x ∈ X : M♯(f)(x) > δℓ . ℓ>0 (cid:16)n o(cid:17) Choosing ǫ small enough such that ν(1+ǫ)p < 1, our assumption then implies that sup ℓpµ({x ∈ X : N(f)(x) >ℓ}) . supℓpµ x ∈ X : M♯(f)(x) >ℓ . 0<ℓ<R ℓ>0 (cid:16)n o(cid:17) LettingR → ∞thenleadstotheconclusion, whichcompletes theproofofLemma3.2. Lemma 3.3. Let r ∈ (0, 1) and N (f) ≡ [N(|f|r)]1/r. Then for any p ∈ [1, ∞), there r exists a positive constant C, depending on r, such that for suitable function f and any ℓ > 0, µ({x ∈ X : N (f)(x) > ℓ}) ≤ Cℓ−psupτpµ({x ∈ X : |f(x)| > τ}). r τ≥ℓ 7 Proof. For each fixed ℓ > 0 and function f, decompose f as f(x)= f(x)χ (x)+f(x)χ ≡ f (x)+f (x). {x∈X:|f(x)|≤ℓ} {x∈X:|f(x)|>ℓ} 1 2 By the boundedness of N form Lp(µ) to Lp,∞(µ), we obtain that µ({x ∈X : N (f)(x)> 21/rℓ}) ≤ µ({x ∈ X : N(|f |r)(x) > ℓr}) r 2 . ℓ−rp |f (x)|rpdµ(x) 2 ZX . µ({x ∈ X : |f(x)| > ℓ}) ∞ +ℓ−rp τrp−1µ({x ∈ X : |f(x)|> τ})dτ Zℓ . µ({x ∈ X : |f(x)| > ℓ}) +ℓ−psupτpµ({x ∈ X : |f(x)| > τ}), τ>ℓ which implies our desired result. Proof of Theorem 1.1. BytheMarcinkiewiczinterpolationtheorem,weonlyneedtoprove that for all f ∈ Lp(µ) with p ∈ (1, ∞) and ℓ > 0, µ({x ∈ X : |Tf(x)| > ℓ}) .ℓ−pkfkp . (3.1) Lp(µ) We further consider the following two cases. Case (i)µ(X)= ∞. LetL∞(µ)bethespace of bounded functions with bounded supports b and L∞ (µ)≡ f ∈ L∞(µ) : f(x)dµ(x) = 0 . b,0 b (cid:26) ZX (cid:27) Then in this case, L∞ (µ) is dense in Lp(µ) for all p ∈ (1,∞). Let r ∈ (0, 1) and b,0 N (g) ≡ [N(|g|r)]1/r for any g ∈ Lr (µ). Notice that |Tf|≤ N (Tf) µ-almost everywhere r loc r on X. Then by a standard density argument, to prove (3.1), it suffices to prove that for all f ∈ L∞ (µ) and p ∈ (1, ∞), b,0 supℓpµ({x ∈ X : N (Tf)(x) > ℓ}) . kfkp . (3.2) r Lp(µ) ℓ>0 For each fixed ℓ > 0, applying Lemma 2.1, we obtain that f = g + h, where h is as Lemma 2.1 and g ≡ f −h, such that kgkL∞(µ) . ℓ, h ∈ H1(µ) (3.3) and khk . ℓ1−pkfkp . (3.4) H1(µ) Lp(µ) For each r ∈ (0, 1), define M♯(f) ≡ {M♯(|f|r)}1/r. Then, (3.3) together with the bound- r edness of T from L∞(µ) to RBMO(µ) and Lemma 3.1 shows that the function M♯(Tg) is r bounded by a multiple of ℓ. Hence, if c is a sufficiently large constant, we have 0 x ∈ X : M♯(Tg)(x) > c ℓ = ∅. (3.5) r 0 n o 8 On the other hand, since both f and h belong to H1(µ), we see that g ∈ H1(µ) and kgk ≤ kfk +khk . kfk +ℓ1−pkfkp . H1(µ) H1(µ) H1(µ) H1(µ) Lp(µ) By this together with the boundedness of T from H1(µ) to L1,∞(µ) and Lemma 3.3, we have that for any p ∈ (1, ∞) and R > 0, sup ℓpµ({x ∈ X : N (Tg)(x) > ℓ}) . sup ℓp−1supτµ({x ∈ X : |Tg(x)| > τ}) < ∞. r 0<ℓ<R 0<ℓ<R τ≥ℓ From this, (3.5), Lemma 3.2 and the fact that N ◦T is quasi-linear, we deduce that there r exists a positive constant C such that supℓpµ x ∈ X : N (Tef)(x)> Cc ℓ r 0 ℓ>0 (cid:16)n o(cid:17) ≤ supℓpµ({x ∈X : Nr(Tg)(x)e> c0ℓ})+supℓpµ({x ∈X : Nr(Th)(x) > c0ℓ}) ℓ>0 ℓ>0 . supℓpµ x ∈X : M♯(Tg)(x) > c ℓ +supℓpµ({x ∈ X : N (Th)(x) > c ℓ}) r 0 r 0 ℓ>0 ℓ>0 (cid:16)n o(cid:17) . supℓpµ({x ∈X : N (Th)(x) > ℓ}). (3.6) r ℓ>0 From the boundednessof N from L1(µ)to L1,∞(µ) andthe boundednessof T fromH1(µ) to L1,∞(µ), it follows that µ({x ∈X : N (Th)(x) > ℓ}) r ℓr ≤ µ x ∈ X : N |Th|rχ > 1 {x∈X:|(Th)(x)|>ℓ/2r} 2 (cid:18)(cid:26) (cid:27)(cid:19) (cid:16) (cid:17) r . ℓ−r (Th)(x)χ (x) dµ(x) 1 {x∈X:|(Th)(x)|>ℓ/2r} ZX (cid:12) (cid:12) (cid:12) ℓ(cid:12)/2r1 . ℓ−rµ (cid:12)x ∈ X : |(Th)(x)| > ℓ/21r (cid:12) sr−1ds (cid:16)n∞ o(cid:17)Z0 +ℓ−r sr−1µ({x ∈ X : |(Th)(x)| > s})ds 1 Zℓ/2r 1 1 . µ x ∈ X : |(Th)(x)| > ℓ/2r + sup sµ({x ∈ X : |(Th)(x)| > s}) ℓ 1 (cid:16)n o(cid:17) s≥ℓ/2r khk . H1(µ) .ℓ−pkfkp , ℓ Lp(µ) which together with (3.6) yields (3.2). Case (ii) µ(X) < ∞. In this case, assume that f ∈ L∞(µ). Notice that if ℓ ∈ (0,ℓ ], b 0 where ℓ is as in Lemma 2.1, then (3.1) holds trivially. Thus, we only have to consider the 0 ♯ case when ℓ > ℓ . Let r ∈ (0, 1), N (f) be as in Lemma 3.3 and M as in Case (i). For 0 r r each fixed ℓ > ℓ , applying Lemma 2.1, we obtain that f = g+h with g and h satisfying 0 9 (3.3) and (3.4), which together with the boundedness of T from L∞(µ) to RBMO(µ) and ♯ Lemma 3.1 yields (3.5) for M (Tg). We now claim that r 1 F ≡ |Tg(x)|rdµ(x) . ℓr, (3.7) µ(X) ZX where the constant depends on µ(X) and r. In fact, since µ(X) < ∞, we regard X as a ball. Then g ≡ g − 1 g(x)dµ(x) ∈ H1(µ). On the other hand, |T1|r ∈ RBMO(µ) 0 µ(X) X because of the fact that T1 ∈ RBMO(µ) and Lemma 3.1. This together with µ(X) < ∞ R implies that |T1(x)|rdµ(x) < ∞. ZX Then by the boundedness of T from H1(µ) to L1,∞(µ) and (3.3), we have r 1 |Tg(x)|rdµ(x) ≤ |Tg (x)|r + T g(y)dµ(y) (x) dµ(x) 0 µ(X) ZX ZX (cid:26) (cid:12) (cid:20) ZX (cid:21) (cid:12) (cid:27) . r kg0kH1(µ)/µ(X)(cid:12)(cid:12)(cid:12)tr−1µ({x ∈ X : |Tg0(x)| > t(cid:12)(cid:12)(cid:12)}) dt Z0 ∞ +r tr−1µ({x ∈ X : |Tg (x)| > t}) dt+ℓr 0 Zkg0kH1(µ)/µ(X) . µ(X) kg0kH1(µ)/µ(X)tr−1dt+kg k ∞ tr−2dt+ℓr 0 H1(µ) Z0 Zkg0kH1(µ)/µ(X) . [µ(X)]1−rkg kr +ℓr . ℓr, 0 H1(µ) which implies (3.7). Observe that (|Tg|r −F)dµ(x) = 0 and for any R > 0, X sRup ℓpµ({x ∈ X : N(|Tg|r −F)(x) >ℓ})≤ Rpµ(X) < ∞. 0<ℓ<R ♯ From this together with Lemma 3.2, M (F) = 0, (3.7) and an argument similar to that r used in Case (i), we conclude that there exists a positive constant c such that supℓpµ({x ∈ X : N (Tf)(x) > cc ℓ}) ≤ supℓpµ({x ∈X : N(|Tg|r −F)(x) > (c ℓ)r}) r 0 0 e ℓ>ℓ0 ℓ>ℓ0 +supℓpµ({x ∈ X : N (Th)(x) > c ℓ}) e r 0 ℓ>ℓ0 . supℓpµ x ∈ X : M♯(Tg)(x) > c ℓ r 0 ℓ>0 (cid:16)n o(cid:17) +supℓpµ({x ∈ X : N (Th)(x) > c ℓ}) r 0 ℓ>0 . supℓpµ({x ∈ X : N (Th)(x) > ℓ}) . kfkp , r Lp(µ) ℓ>0 whereinthefirstinequalitywechoosec largeenoughsuchthatF≤ (c ℓ)r. Thiscompletes 0 0 the proof of Theorem 1.1. 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.