A note on a conjecture of Calderón

Σχετικά έγγραφα
1. For each of the following power series, find the interval of convergence and the radius of convergence:

On Generating Relations of Some Triple. Hypergeometric Functions

A study on generalized absolute summability factors for a triangular matrix

On Certain Subclass of λ-bazilevič Functions of Type α + iµ

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

Lecture 17: Minimum Variance Unbiased (MVUB) Estimators

n r f ( n-r ) () x g () r () x (1.1) = Σ g() x = Σ n f < -n+ r> g () r -n + r dx r dx n + ( -n,m) dx -n n+1 1 -n -1 + ( -n,n+1)

Bessel function for complex variable

ST5224: Advanced Statistical Theory II

Every set of first-order formulas is equivalent to an independent set

Solve the difference equation

On Inclusion Relation of Absolute Summability

Last Lecture. Biostatistics Statistical Inference Lecture 19 Likelihood Ratio Test. Example of Hypothesis Testing.

SUPERPOSITION, MEASUREMENT, NORMALIZATION, EXPECTATION VALUES. Reading: QM course packet Ch 5 up to 5.6

Ψηφιακή Επεξεργασία Εικόνας

Statistical Inference I Locally most powerful tests

Other Test Constructions: Likelihood Ratio & Bayes Tests

Uniform Estimates for Distributions of the Sum of i.i.d. Random Variables with Fat Tail in the Threshold Case

The Heisenberg Uncertainty Principle


Homework 8 Model Solution Section

Introduction of Numerical Analysis #03 TAGAMI, Daisuke (IMI, Kyushu University)

A Note on Intuitionistic Fuzzy. Equivalence Relation

Homework for 1/27 Due 2/5

The Neutrix Product of the Distributions r. x λ

Econ 2110: Fall 2008 Suggested Solutions to Problem Set 8 questions or comments to Dan Fetter 1

C.S. 430 Assignment 6, Sample Solutions

4.6 Autoregressive Moving Average Model ARMA(1,1)

ORDINAL ARITHMETIC JULIAN J. SCHLÖDER

Presentation of complex number in Cartesian and polar coordinate system

Uniform Convergence of Fourier Series Michael Taylor

Example Sheet 3 Solutions

2 Composition. Invertible Mappings

Areas and Lengths in Polar Coordinates

F19MC2 Solutions 9 Complex Analysis

SOLUTIONS TO MATH38181 EXTREME VALUES AND FINANCIAL RISK EXAM

w o = R 1 p. (1) R = p =. = 1

Στα επόμενα θεωρούμε ότι όλα συμβαίνουν σε ένα χώρο πιθανότητας ( Ω,,P) Modes of convergence: Οι τρόποι σύγκλισης μιας ακολουθίας τ.μ.

3.4 SUM AND DIFFERENCE FORMULAS. NOTE: cos(α+β) cos α + cos β cos(α-β) cos α -cos β

COMMON RANDOM FIXED POINT THEOREMS IN SYMMETRIC SPACES

Lecture 2. Soundness and completeness of propositional logic

A New Class of Analytic p-valent Functions with Negative Coefficients and Fractional Calculus Operators

Homework 4.1 Solutions Math 5110/6830

Lecture 13 - Root Space Decomposition II

Biorthogonal Wavelets and Filter Banks via PFFS. Multiresolution Analysis (MRA) subspaces V j, and wavelet subspaces W j. f X n f, τ n φ τ n φ.

Fourier Series. MATH 211, Calculus II. J. Robert Buchanan. Spring Department of Mathematics

LAD Estimation for Time Series Models With Finite and Infinite Variance

Fractional Colorings and Zykov Products of graphs

Math221: HW# 1 solutions

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

6.1. Dirac Equation. Hamiltonian. Dirac Eq.

Math 446 Homework 3 Solutions. (1). (i): Reverse triangle inequality for metrics: Let (X, d) be a metric space and let x, y, z X.

Areas and Lengths in Polar Coordinates

2. Let H 1 and H 2 be Hilbert spaces and let T : H 1 H 2 be a bounded linear operator. Prove that [T (H 1 )] = N (T ). (6p)

Supplemental Material: Scaling Up Sparse Support Vector Machines by Simultaneous Feature and Sample Reduction

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

Solutions: Homework 3

b. Use the parametrization from (a) to compute the area of S a as S a ds. Be sure to substitute for ds!

Srednicki Chapter 55

SUPPLEMENT TO ROBUSTNESS, INFINITESIMAL NEIGHBORHOODS, AND MOMENT RESTRICTIONS (Econometrica, Vol. 81, No. 3, May 2013, )

Solution Series 9. i=1 x i and i=1 x i.

Μια εισαγωγή στα Μαθηματικά για Οικονομολόγους

Matrices and Determinants

Answer sheet: Third Midterm for Math 2339

PROPERTIES OF CERTAIN INTEGRAL OPERATORS. a n z n (1.1)

1. Matrix Algebra and Linear Economic Models

Section 8.3 Trigonometric Equations

CRASH COURSE IN PRECALCULUS

CHAPTER 103 EVEN AND ODD FUNCTIONS AND HALF-RANGE FOURIER SERIES

SOLVING CUBICS AND QUARTICS BY RADICALS

Congruence Classes of Invertible Matrices of Order 3 over F 2

Proof of Lemmas Lemma 1 Consider ξ nt = r

CHAPTER 101 FOURIER SERIES FOR PERIODIC FUNCTIONS OF PERIOD

MATH 38061/MATH48061/MATH68061: MULTIVARIATE STATISTICS Solutions to Problems on Matrix Algebra

A Two-Sided Laplace Inversion Algorithm with Computable Error Bounds and Its Applications in Financial Engineering

Supplement to A theoretical framework for Bayesian nonparametric regression: random series and rates of contraction

ANSWERSHEET (TOPIC = DIFFERENTIAL CALCULUS) COLLECTION #2. h 0 h h 0 h h 0 ( ) g k = g 0 + g 1 + g g 2009 =?

Online Appendix I. 1 1+r ]}, Bψ = {ψ : Y E A S S}, B W = +(1 s)[1 m (1,0) (b, e, a, ψ (0,a ) (e, a, s); q, ψ, W )]}, (29) exp( U(d,a ) (i, x; q)

SCITECH Volume 13, Issue 2 RESEARCH ORGANISATION Published online: March 29, 2018

INTEGRATION OF THE NORMAL DISTRIBUTION CURVE

Iterated trilinear fourier integrals with arbitrary symbols

Chapter 3: Ordinal Numbers

Tridiagonal matrices. Gérard MEURANT. October, 2008

MINIMAL CLOSED SETS AND MAXIMAL CLOSED SETS

Nowhere-zero flows Let be a digraph, Abelian group. A Γ-circulation in is a mapping : such that, where, and : tail in X, head in

Jordan Journal of Mathematics and Statistics (JJMS) 4(2), 2011, pp

IIT JEE (2013) (Trigonomtery 1) Solutions

A General Note on δ-quasi Monotone and Increasing Sequence

Finite Field Problems: Solutions

Oscillatory integrals

Reminders: linear functions

Chapter 2. Ordinals, well-founded relations.

forms This gives Remark 1. How to remember the above formulas: Substituting these into the equation we obtain with

Geodesic Equations for the Wormhole Metric

1. Introduction and Preliminaries.

Quadratic Expressions

THE SECOND ISOMORPHISM THEOREM ON ORDERED SET UNDER ANTIORDERS. Daniel A. Romano

SOME IDENTITIES FOR GENERALIZED FIBONACCI AND LUCAS SEQUENCES

Mock Exam 7. 1 Hong Kong Educational Publishing Company. Section A 1. Reference: HKDSE Math M Q2 (a) (1 + kx) n 1M + 1A = (1) =

Supplementary Materials: Trading Computation for Communication: Distributed Stochastic Dual Coordinate Ascent

Transcript:

A ote o a cojecture of Calderó Jiecheg CHEN & Xiagrog ZHU Dept of Math Xixi Campus), Zhejiag Uiversitry Abstract For f SR ) ad Ω L 1 S 1 ), S 1 Ωx )dx = 0, defie T Ω f)x) = lim ɛ 0+ x y ɛ Ωy/ y ) y fx y)dy I this paper, we shall prove that there are a class of fuctios i H 1 S 1 ) L l + LS 1 ) such that T Ω is weak type L 1 bouded Supported by 973 project ad NSFZJ 1

1 Itroductio For f SR d ) ad Ω L 1 S d 1 ), S d 1 Ωx )dx = 0, defie Ωy/ y ) T Ω f)x) = lim ɛ 0+ x y ɛ y d fx y)dy 1) I [1], Calderó ad Zygmud proved that if Ω L l + LS d 1 ), ie S d 1 Ωx ) l + Ωx ) )dx <, ) T Ω is L p bouded for 1 < p < I [7] ad [5], Ricci, Weiss ad idepedetly Coett proved that if Ω H 1 S d 1 ), T Ω is L p bouded for 1 < p < Also see [4] But, it remaied ope for log time if T Ω is weak type L 1 bouded uder the correspogdig coditios for Ω L l + LS d 1 ), it is a cojecture of Calderó) I [3], Christ ad Rubio de Fracia proved that for Ω L l + LS d 1 ) d 7), T Ω is weak type L 1 bouded Ad at the same time, i [6], Hofma idepedetly proved that for Ω L q S d 1 ) d =, q > 1), T Ω is weak type L 1 bouded Fially, i [8], Seeger geeralized the result to all d ad Ω L l + LS d 1 ) We kow that L l + LS d 1 ) H 1 S d 1 ) So, it is atural to ask a harder questio: for Ω H 1 S d 1 ), is T Ω weak type L 1 bouded? I [9], Stefaov proved that if Ω is a fiite sum of H 1 S 1 ) atoms with the additioal assumptio that the atoms are supported o almost disjoit arcs of comparable size ote that such a Ω must be L S 1 ) fuctio), T Ω L 1 W L 1 essetially depeds oly o Ω H1 Precisely, he proved Theorem 1 Let N, l be positive itegers satisfyig N π l c 0 where c 0 is suitably choose, ad I deote the arc i S 1 with ceter e ad satisfyig I l, I I m 1 mi I, I m ) for m Suppose Ω = N 1 a where > 0 ad a is a H 1 S 1 ) atom o S 1 satisfyig suppa ) I, a l ad S 1 a θ)dθ = 0 The, T Ω L 1 W L 1 C N =1 ) where C is idepedet of N ad l I this paper, we shall prove that there are a class of fuctios i H 1 S 1 ) L l + LS 1 ) such that T Ω is weak type L 1 bouded We have

Theorem Let I deote the arc i S 1 with ceter e ad legth ρ, disjoit mutually, ad A = sup θ S 1 :θ / I ρ < ; 3) θ e let Ω = 1 a where 1 < ad a is a H 1 S 1 ) atom o S 1 satisfyig i) suppa ) I, ii) S 1 a θ)dθ = 0, iii) a ρ 1, 4) The, {x : T Ω f)x) > } C A ) 1 f 1 5) for all f SR ) ad > 0, where C is idepedet of f, ad Ω Theorem 3 Suppose ρ 0, 1) ad ρ < There are {e } S 1, arcs I S 1 with ceter e ad legth ρ such that for ay sequece of H 1 atoms {a } satisfyig 4))ad Ω = a, {x : T Ω f)x) > } C ρ f 1 6) for all f SR ) ad > 0, where C is idepedet of f, ad Ω From Theorem 3, we have Corollary 4 For ay icreasig fuctio ϕ : [0, ) [0, ) with ϕ0+) = 0, if ϕt) lim =, 7) t t the, there must be a Ω H 1 ϕl) such that T Ω is weak type L 1 bouded, where Some lemmas ϕl) = { } Ω : ϕ Ωθ) )dθ < S 1 Without loss of geerality, we always assume that 1 = 1 ad > 0 3

For a rectagle Q = Qy, r) R ) with ceter y ad sides legth r = r 1, r ), let mq = Qy, mr), dq) = maxr 1, r ) Set ad A = Q : the loger side of Q is parallel to e, the shorter side legth is ρ time of the loger side legth 1 M f)x) = sup x Q A Q Q fz) dz It is easy to see that M is weak type L 1 bouded ad sup M L 1 W L 1 we first give a modified Whitey s decompositio, < Now, Lemma 5 Suppose E R is ope There is m R + such that for ay, there are mutually disjoit rectagles {Q,i } A satisfyig i) ii) E = i Q,i 4Q,i E iii) mq,i E c { } iv) dq,i ) k : k = 0, ±1, ±, This lemma ca be proved alog the idea of the proof of the Whitey s decompositio, see[10] For f L 1 R ) ad > 0, let E = { } x : M f)x) > E = {x : Mf)x) > } E ) where M is the Hardy-Littlewood maximal operator For ay, we shall make C Z decompositio of f based o the modified Whitey s decompositio of E ot E, key poit) By Lemma 5, we have that E = i Q,i where {Q,i } satisfy the coditios i Lemma 5 Take We have b Q,i = fx) 1 g x) = i 1 Q,i Q,i gx) = fx)χ E cx) B,j = i:dq,i )= j b Q,i Q,i fy)dy)χ Q,i x) Q,i fy)dyχ Q,i x) 4

Lemma 6 For ay, f = g + g + i b Q,i, ad i) ii) R b Q,i x)dx = 0 bq,i R x) dx C Q,i bq,i i Q,i x) dx f 1 g C f 1, g C f 1 iii) E E + {x : Mf)x) > } C 1 f 1 Lemma 7 Uder the hypothesis of Theorem, for all f SR ) ad > 0 a ) g C A f 1 8) Proof We first prove that a ) ) ξ) C A 9) By a well-kow computatio see [10] page 39), a ) ) 1 ξ) = a θ) l S 1 θ, ξ ξ + πi sig θ, ξ dθ 10) which is idepedet of ξ So, we may assume ξ = 1 Let ξ S 1 deote ayoe of the TWO uit vectors orthogoal to ξ If ±ξ / I, sig θ, ξ is costat for θ I, thus without loss of geerality, we may assume that e, ξ ) < π/, i this case, e, ξ e ξ ) ) a ) ξ) = S 1 a θ) l θ, ξ 1 l e, ξ 1) dθ S 1 a θ) l1 + θ e, ξ e, ξ 1 )dθ C sup θ I θ e e ξ C ρ e ξ 11) By 11), we have ote that A 1) ) a ) ξ) = + ) ) a ) ξ) :ξ I or ξ I :±ξ / I C1 + ρ ) C A e ξ 1) 5

Thus a ) g = = S 1 Lemma 7 is proved m S 1 m m a ) ) g ) a ) C A m ĝ m C A m m ) ) ) ξ) g ) am ξ) m ξ) g m ) ξ)dξ g m ) ξ) ) a ) ξ) dξ ) m m f 1 C A f 1 Take β C c 1, )) such that 0 β 1 ad j β j t) = 1 for all t R +, ψ Cc 1, 1)) such that 0 ψ 1 ad ψ [ 1, 1 ] = 1 ad ψ k) ± 1 ) = 0 for all k = 0, 1,, Let a,j x) = β j x ) ax/ x ) x ) ϕ s x) = ψ s 13) 6 ρ 1, e ) x) We have Lemma 8 Uder the hypothesis of Theorem, a,j ϕ s B,j s C A f 1 14) s>0 j Proof By a similar estimate i [9] see sectio 5 below for details), we have a,j B,j s C s f j 1 15) Now, we estimate ϕ s ξ) where s 0 If ϕ s ξ) = ψ s 6 ρ 1 ξ ξ, e ) > 0, we have s 6 ρ 1 ξ ξ, e 1, thus ξ ξ, e s 6 ρ 16) Note that ξ ξ, e ξ e for all satisfyig ±ξ / I ad e, ξ ) < π/, we have ξ s e C 6 ξ ρ, ad thus ρ 1 e C s 6 By Theorem 3), the umber of satisfyig 16) does ot exceed CA s 3, thus ϕ s ξ) CA s 3 17) 6

By 15) ad 17), ) a,j ϕ s B,j s s>0 j s>0 a,j ϕ s B,j s j ) ) = ϕ s>0 s â,j B,j s j s>0 ϕ s ξ) ) 1/ ) 1/ â,j B,j s j CA s 3 1/ a s>0,j B,j s j ) ) ) CA s 3 s 1/ f 1 = CA f 1 s>0 Lemma 8 is proved Lemma 9 Uder the hypothesis of Theorem, a,j δ ϕ s ) B,j s 1 C f 1 18) s>0 j Proof By a similar estimate i [9] see sectio 5 below for details), we have a,j δ ϕ s ) 1 C s 1 19) So Lemma 9 is proved a,j δ ϕ s ) B,j s 1 s>0 j C s 1 B,j s 1 = C f 1 s>0 j 3 Proof of Theorem For fixed f ad, T Ω f)x) = T Ω g)x) + a ) g x) + i a ) b Q,i x), 7

so, {x : T Ω f)x) > } I + II + III I = {x : T Ω g)x) > /3} { } II = x : a ) g x) > /3 { } III = x : i a ) b Q,i x) > /3 0) By L boudedess of T Ω, I C T Ω g) C Ω H 1 g C ) f 1 = C 1 f 1 1) Similarly, we have For III, we have II C a ) g x) CA 1 f 1 ) a ) b i Q,i x) = a,j B,j s x) s0 j + a,j B,j s x) s>0 j = a,j B,j s x) s0 j + a,j ϕ s B,j s x) s>0 j + a,j δ ϕ s ) B,j s x) s>0 j 3) If dq,i ) j, suppa,j b Q,i ) 4Q,i, so supp a,j B,j s ) i 4Q,i = E s0 j Noticig that E C 1 f 1, we have { } III E + x : a,j ϕ s B,j s x) s>0 j > /9 { } + x : a,j δ ϕ s ) B,j s x) > /9 s>0 = E + IV + V j 4) 8

By Lemma 8, IV C a,j ϕ s B,j s s>0 j CA 1 f 1 5) By Lemma 9, V C 1 a,j δ ϕ s ) B,j s CA 1 f 1 6) 1 s>0 j Combiig 0)-) ad 4)-6), we get {x : T Ω f)x) > } CA 1 f 1 Theorem is proved 4 Proof of Theorem 3 Without loss of geerality, we may assume that {ρ } is decreasig ad ρ < π/64 Let d m = 1 sup m m ρ We have Lemma 10 m d m 16 ρ Proof For m, ρ 1 i ρi ) im ρi ), thus d m m m m im ρi ) O the other had, m ρi ) m ρi ρj m im m jm ij = ρj ρi ) m j ij m j 4 j 4 j j 1 ρj ij ρi ) ρ j ) 1/ j 1 ρi ) ) 1/, j ij so d m m m m im ρi ) 16 j ρ j Lemma 10 is proved 9

By Lemma 10 ad the assumptio ρ < π/64, we get m d m < π/ So, we ca choose {e m } S 1, such that e m+1 e m = d m ad 0 < arg e m < arg e m+1 < π/ for all m I additio, by the fact d m ρ m + ρ m+1, {I m } are disjoit mutually We shall first apply iductio to prove that e m e for m > For m = + 1, e m e = d > m ) m ρ, we have m ) m ρ 7) +1 ρ = m ) m e m+1 e = e m+1 e m + e m e d m + m ) m ρ > m+1 ) m+1 ρ So, 7) holds for all m > Now, we shall prove that For θ S 1, we first cosider sup θ S 1 :θ / I m ρ Suppose e m e + m ) ) m ρ ρ < 8) θ e N + θ N θ N 0 θ def = { : 0 arg θ arg e π/} def = { : 0 arg θ arg e π/} def = { : arg θ arg e > π/} Label the elemets i N + θ by sub-idex such that < θ e < θ e 1 < θ e 0 The, N + θ = { < 1 < 0 } or N + θ = { K < < 1 < 0 } By 7), i the secod case, θ e l > e K e l > l K l ) K ρ l, thus : N + θ,θ / I ρ θ e i the fist case, θ e l > l ρ l, so 1 + K 1 K 1 l=0 l=1 ρ θ e : N + θ,θ / I 1 + l=1 ) l K l ) K l K l ) K ) C < ; l ) = C < 9) 10

Label the elemets i N θ by sub-idex such that θ e 0 < θ e 1 < θ e < The, N + θ = { 0 < 1 < } or N + θ = { 0 < 1 < < K } By 7), it is easy to show that I additio, : N 0 θ,θ / I : N θ,θ / I ρ C < 30) θ e ρ θ e C : N 0 θ,θ / I ρ = C < 31) From 9)-31), we get 8) By 8) ad Theorem, we get Theorem 3 Fially, we prove Corollary 4 By 7), we ca choose {t } such that 1 < t 1 < t < ad ϕt ) > t Set =, ρ = t 1, Ω = a where {a } are H 1 atoms satisfyig 4) ad a θ) = ρ 1 ie Ω / ϕl) for θ I The, Ω H 1 S 1 ), but S 1 ϕ Ωθ) )dθ = ρ ϕ ρ 1 ) =, 5 Appedix I the proofs of Lemmas 8-9, we apply the estimates 15) ad 19) without proofs I what follows, we shall give details of their proofs alog the ideas developped i [], [3], [6], [8] ad [9] Proof of 15) We have j a,j B,j s = a,j B,j s, a,i B,i s j i B,j s, ã,j a,i B,i s j ij B,j s 1 ã,j a,i B,i s j ij f 1 sup j ã,j a,i B,i s, ij 3) 11

where ã,j x) = a,j x) We first estimate i 3 ã,0 a,i B,i s 0) We have ã,0 a,i B,i s 0) = R B,i s y) R a,0 y + z)a,i z)dz)dy = B,i s y) + R S 1 def = S 1 a θ)t i,s a )θ)dθ 0 a y+tθ y+tθ ) β y+tθ ) β i t) y+tθ t dt)a θ)dθ)dy where So, by 4), T i,s a )θ) = R B,i s y)l y i a )θ)dy L y i a )θ) = + i 3 ã,0 a,i B,i s 0) 0 a y+tθ y+tθ ) β y+tθ ) β i t) y+tθ t dt i 3 S 1 a θ)t i,s a )θ)dθ sup θ i 3 T i,sa )θ) 33) For coveiece, let Q θ) def = Θ,i,θ def = { } y : 1 y, 4 y 3, θ 4ρ { y, t) : a y+tθ y+tθ ) β y+tθ ) β i t) y+tθ t 0 } 34) where θ be oe of the two uit vectors orthogoal to θ The suppl y i a )θ)) Q θ) 35) Actually, if L y i a )θ) 0, i 3 ad y, t) Θ,i,θ, we have y y + tθ + tθ + i+1 3, y y + tθ tθ 1 i+1 1 4, ad Let y, θ = y + tθ, θ y + tθ y+tθ y+tθ, θ y + tθ y+tθ y+tθ θ ) y+tθ y + tθ y+tθ e + e θ 4ρ Q s θ) = { } y : y 4 ad y, θ ρ s where s > 0, the i 3 T i,s a )θ) i 3 i 3 R B,i s y)l y i a )θ)dy + ) j:q,j Q s θ),dq,j )= i s j:q,j Q s θ)=,dq,j )= i s R b,j y)l y i a )θ)dy def = I + II 36) 1

For I, oticig that we have L y i a )θ) I + 0 ρ 1 β y + tθ ) β i t) y + tθ t Cρ 1 i 3 j:q,j Q s θ),dq,j )= i s Cρ 1 Cρ 1 j:q,j Q sθ),dq,j ) s Q,j ρ s = C s dt Cρ 1, Q,j To estimate II, for y Q,j Q θ) where dq,j ) = i s, y θ such that y y θ //θ, y θ y Q //θ where y Q is the ceter of Q,j, thus 37) y y θ = y y Q, θ i s y θ y Q = y y Q, θ C i s ρ yy θ Q s θ) = y θ y Q Q s θ) = By Lemma 11 below ad 38), we have L y i a )θ) L y θ i a )θ) i s sup θ Lz i a )θ) C s ρ 1 z yy θ L y θ i a )θ) L y Q i a )θ) i s ρ sup L z θ i a )θ) C s ρ 1 z y θ y Q 38) 39) So, II = i 3 R b,j y)l y i a )θ)dy j:q,j Q s θ)=,dq,j )= i s R b,j y) L y i a )θ) L y Q i a )θ) dy i 3 j:q,j Q sθ)=,dq,j )= i s C i 3 j:q,j Q s θ)=,q,j Q θ),dq,j )= i s sup y L y i a )θ) L y θ i a )θ) + L y θ i C s ρ 1 j:q,j Q θ),dq,j ) s Q,j C Q,j a )θ) L y ) Q a )θ) i s 40) By 33), 36), 38) ad 40), we get i 3 ã,0 a,i B,i s 0) C s 13

By traslatio argumets, ã,0 a,i B,i s By dilatio argumets, i 3 L R ) C s ã,j a,i B,i s C ij s 41) Combiig with 3), it gives 15) Lemma 11 For θ suppa ), y, θ ρ s where s > 0, y 4, we have i) ii) θ Ly i a )θ) i ρ 1 L y θ i a )θ) i+ s ρ Proof Without loss of geerality, we may assume θ = 1, 0) Let e 1 = 1, 0), e = 0, 1), the y = y 1 e 1 + y e, y ρ s Settig w = y+tθ y+tθ = y+te 1 y+te 1, we have Jw, y) = dw dt = y y + te 1 4) Note that y + te 1, w = 0, so t = y,w e 1,w Obviously, for y, t) Θ,i,θ where Θ,i,θ is defied by 34), we have t y 1 = 1 t y = e,w e 1,w = e 1,w e,w Cρ 1 s We first estimate y 1 L y i a )θ) By 43), y+te 1) y 1 = 0 = y+te 1 Thus, y 1 Jw, y) = y y 1 y 1 y + te 1 ) = 0, ad for y, t) Θ,i,θ, β y + te 1 ) β i t) y 1 y + te 1 ) t C β i t) ) y 1 t C i t + 1 t C i Therefore y 1 L y i a )θ) a w) + a w) y 1 β y+tθ ) β i t) dw y+tθ t ) Jw,y) β y+tθ ) β i t) Jw,y) y 1 y+tθ t Jw,y) dw Jw,y) Cρ 1 i w[ i 1, i+1 ]) Cρ 1 i i+1 i 1 dt = Cρ 1 i dw Jw,y) 43) 44) 14

Now, we estimate y L y i a )θ) At first, by 43), we have y+te 1 y = y+te 1, C y+te y 1 ) y+te 1 t y + 1) Cρ 1 s y y + te 1 ) 45) Thus, by 43) ad 45), for y, t) Θ,i,θ, y β i t) t ) C i t+1 t t y χ i 1 t i+1t) C i ρ 1 s y β y+te 1 ) ) y+te 1 C y+te 1 + y+te 1 y+te 1 ) y+te 1 4 y χ 1 y+te 1 Cρ 1 s, which meas that β y + te 1 ) β i t) y y + te 1 ) t C i ρ 1 s 46) I additio, for y, t) Θ,i,θ such that y, θ ρ s, we have thus, By 46) ad 47), we get Lemma 11 is proved Proof of 19) Jw,y) y = y y y+te 1 ) C1 + y ρ 1 ) y+te1 + y y y+te 1 ) y y+te 1 y ) Jw, y) Jw, y) y C1 + y ρ 1 y y L y i a )e 1 ) a w) + a w) C i+ s ρ 1 +Cρ 1 ) y β y+tθ ) β i t) dw y+tθ t ) Jw,y) β y+tθ ) β i t) Jw,y) y y+tθ t Jw,y) ρ 1 dw w[ i 1, i+1 ]) Jw,y) dw w[ i 1, i+1 ]) Jw,y) s ρ 1 C i+ s ρ Cρ 1 s 47) i+1 i 1 dt = C i+ s ρ dw Jw,y) For simplicity, we omit the subidex By the defiitio of ϕ s ie ϕ s, see 13)) ad ψ, we have δ ϕ s ) ) = 1 ψ s 6 ρ 1, e ) = φ s 6 m ρ 1, e ) m 0 15

where φ C0, 1 ) 1, )) is defied by φ {u: 1 < u 1} = 1 ψ, φ {u:1 u <} = ψ ), φ {u: 1 < u <}c = 0 which satisfies m 0 φ m u) = 1 ψu) for all u R 1 Take a oegative fuctio ς { } C0 1 < < ) such that + ς k u) = 1 for all u R 1, ad set L k = ς k )), the a j δ ϕ s ) = Cm s,k ) a j L k m 0 k 48) Cm s,k y) = ς k y)φ s 6 m ρ 1 y y, e ) where a j is just a,j defied by 13) with ceter e = e ad radius ρ = ρ We first estimate Cm s,k ) 1 Defie a iversible liear operator A km : R R by A k,m y = k+ s 6 +m ρy 1, k y ) where y 1 //e ad y e Let h = s 6 +m ρ, by Sobolev imbeddig theorem f C 1 α α fq )) where Q is a osigular liear trasform o R ) C s,k m ) 1 C = C I the supports of ς ad φ, we have which meas that so From 49)-51), we get C s,k m ) 1 C y α α y α α C s,k m A k,m y)) ςhy 1 + y )φ y 1,e hy 1 +y ) ) 1 hy 1 + y, 1 y 1 hy 1 + y, y 1 4, y, h hy 1 + y y 1 49), 50) ) y 1, e α y ςhy 1 + y )φ hy 1 + y ) C 51) α y α ) y 1, e ςhy 1 + y )φ hy 1 + y ) C 5) 16

Now, a j L k 1 = ay/ y )β j y ) y = S 1 0 R S 1 j+1 C C β j r) aθ) r sup r [ j 1, j+1 ],θ suppa) L k x y)dy dx L kx rθ)drdθ dx j 1 j aθ) L k x rθ) L k x re) drdθdx R L k x rθ) L k x re) dx sup r θ e L k 1 C k+j ρ r [ j 1, j+1 ],θ suppa) 53) By 5)-53), C s,k m ) a j L k 1 C k+j ρ 54) But, for larger k say, k + j 0), the estimate 54) is ot eough, we eed some other estimate Applyig Sobolev imbeddig thorem agai, we get by 50)-51) Write C s,k m ) a j L k 1 C C y α α α C s,k m A k,m y)â j A k,m y)) α y â j A k,m )) χ suppc s,k m A k,m )) 55) Note that y = y 1 e + y e suppc s,k m A k,m )), θ = θ 1 e + θ e suppa), θ 1 1, θ ρ ς k A k,m y) 0 = A k,m y k φ s 6 m ρ 1 A k,my A k,m y, e ) 0 = A k,my A k,m y, e s 6 +m ρ I additio, θ e < ρ ad s 6 + m > 0, so A k,my A k,m y, θ s 6 +m ρ A k,m y, θ s 6 +m+k ρ, 56) 57) ad ) α y A k,m y, θ = y α θ 1 s 6 +m+k ρy 1 + θ k y C s 6 +m+k ρ) α1 k ρ) α C s 6 +m+k ρ) α 58) 17

Now, by itegratio by parts, â j A k,m y) = aθ) β j r) r e ir Ak,my,θ drdθ = aθ) i A k,m y,θ ) 3 3 r β j r) r )e ir Ak,my,θ drdθ So, by 57)-58) ad the fact r 3 β j r) )r α 1 C j3 α +1), we have r y α â j A k,m y)) C aθ) ir) γ y γ A k,m y,θ i A 0γα k,m y,θ ) 3 3 r β j r) r )e ir Ak,my,θ drdθ C 0γα ρρ 1 s 6 +m+k ρ) 3 j3 γ +1) j s 6 +m+k ρ) γ = C s 6 +m+k+j ρ) 3+ γ 0γα 59) By 55), 50) ad 59), we have C s,k m ) a j L k 1 C α s 6 +m+k+j ρ) 3+ α = C 3 s 6 +m+k+j ρ) l 60) l=1 So, C s,k m ) a j L k m 0 k 1 C { } mi k+j 3 ρ, s 6 +m+k+j ρ) l m 0 k l=1 C k+j ρ m 0 k: k+j ρ 1 s m +C 3 s 6 +m+k+j ρ) l m 0 k: k+j ρ> 1 s m l=1 C s 1 m = C s m 0 1 61) From 48), 61), we get 19) 18

Refereces [1] A P Calderó ad A Zygmud, O sigular itegrals, Amer Math Soc 781956), 89-309 [] M Christ, Weak type bouds for rough kerels, A of Math 181988), 19-4 [3] M Christ ad J-L Rubio de Fracia, Weak type 1,1) bouds for rough operators II, Ivet Math 931988), 9-37 [4] R R Coifma ad G Weiss, Extesios of Hardy spaces ad their use i aalysis, Bull Amer Math Soc 831977), 569-645 [5] W C Coett, Sigular itegrals ear L 1, Proc Sympos Pure Math of Amer Math Soc S Waiger ad G Weiss eds), Vol35 1) 1979), 163-169 [6] S Hofma, Weak 1,1) boudedess of sigular itegrals with osmooth kerels, Proc Amer Math Soc 1031988), 60-64 [7] F Ricci ad G Weiss, A characterizatio of H 1 1 ), Proc Sympos Pure Math of Amer Math Soc S Waiger ad G Weiss eds), Vol35 1) 1979), 89-94 [8] A Seeger, Sigular itegral operator with rough covolutio kerels, J Amer Math Soc 91996), 99-109 [9] A Stefaov, Weak type estimates for certai Calderó-Zygmud sigular itegral operators, Studia Math 1471)001), 1-13 [10] E M Stei, Sigular itegrals ad differetiability properties of fuctios, Priceto Uiversity Press, Priceto N J, 1970 19