See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/221663910

A simple proof of the $A_2$ conjecture

Article · February 2012

Source: arXiv

CITATIONS READS
46 75

1 author:

Andrei Lerner
Bar Ilan University
61 PUBLICATIONS 2,351 CITATIONS

SEE PROFILE

All content following this page was uploaded by Andrei Lerner on 24 November 2016.

The user has requested enhancement of the downloaded file.

 A SIMPLE PROOF OF THE A2 CONJECTURE

ANDREI K. LERNER
arXiv:1202.2824v3 [math.CA] 5 May 2012

Abstract. We give a simple proof of the A2 conjecture proved

recently by T. Hytönen. Our proof avoids completely the notion
of the Haar shift operator, and it is based only on the “local mean
oscillation decomposition”. Also our proof yields a simple proof of
the “two-weight conjecture” as well.

1. Introduction
Let T be an L2 bounded Calderón-Zygmund operator. We say that
w ∈ A2 if
kwkA2 = sup w(Q)w −1(Q)/|Q|2 < ∞.
Q⊂Rn
In this note we give a rather simple proof of the A2 conjecture recently
settled by T. Hytönen [7].
Theorem 1.1. For any w ∈ A2 ,
(1.1) kT kL2 (w) ≤ c(n, T )kwkA2 .
Below is a partial list of important contributions to this result. First,
(1.1) was proved for the following operators:
• Hardy-Littlewood maximal operator (S. Buckley [3], 1993);
• Beurling transform (S. Petermichl and A. Volberg [24], 2002);
• Hilbert transform (S. Petermichl [22], 2007);
• Riesz transform (S. Petermichl [23], 2008);
• dyadic paraproduct (O. Beznosova [2], 2008);
• Haar shift (M. Lacey, S. Petermichl and M. Reguera [16], 2010).
After that, the following works appeared with very small intervals:
• a simplified proof for Haar shifts (D. Cruz-Uribe, J. Martell and
C. Pérez [5, 6], 2010);
• the L2 (w) bound for general T by kwkA2 log(1 + kwkA2 ) (C. Pérez,
S. Treil and A. Volberg [21], 2010);
• (1.1) in full generality (T. Hytönen [7], 2010);
2010 Mathematics Subject Classification. 42B20,42B25.
Key words and phrases. Calderón-Zygmund operator, Haar shift operator, local
mean oscillation decomposition, A2 conjecture.
1
2 ANDREI K. LERNER

• a simplification of the proof (T. Hytönen et al. [12], 2010);

• (1.1) for the maximal Calderón-Zygmund operator T♮ (T. Hytönen
et al. [9], 2010).
The “Bellman function” proof of the A2 conjecture in a geometri-
cally doubling metric space was given by F. Nazarov, A. Reznikov and
A. Volberg [19] (see also [20]).
All currently known proofs of (1.1) were based on the representation
of T in terms of the Haar shift operators Sm,k D . Such representations
also have a long history; for general T it was found in [7]. The second
key element of all known proofs was showing (1.1) for Sm,k
D in place of T
with the corresponding constant depending linearly (or polynomially)
on the complexity. Observe that over the past year several different
proofs of this step appeared (see, e.g., [15, 25]).
In a very recent work [18], we have proved that for any Banach
function space X(Rn ),
(1.2) kT♮ f kX ≤ c(T, n) sup kAD,S |f |kX ,
D,S

where X
AD,S f (x) = fQkj χQkj (x)
j,k

(this operator is defined by means of a sparse family S = {Qkj } from a

general dyadic grid D; for these notions see Section 2 below).
Observe that for the operator AD,S f inequality (1.1) follows just in
few lines by a very simple argument. This was first observed in [5, 6]
(see also [18]). Hence, in the case when X = L2 (w), inequality (1.2)
easily implies the A2 conjecture. Also, (1.2) yields the “two-weight
conjecture” by D. Cruz-Uribe and C. Pérez; we refer to [18] for the
details.
The proof of (1.2) in [18] still depended on the representation of T
in terms of the Haar shift operators. In this note we will show that
this difficult step can be completely avoided. Our new proof of (1.2) is
based only on the “local mean oscillation decomposition” proved by the
author in [17]. It is interesting that we apply this decomposition twice.
First it is applied directly to T♮ , and we obtain that T♮ is essentially
pointwise dominated by the maximal operator M and a series of dyadic
type operators Tm . In order to handle Tm , we apply the decomposition
again to the adjoint operators Tm⋆ . After this step we obtain a pointwise
domination by the simplest dyadic operators AD,S .
Note that all our estimates are actually pointwise, and they do not
depend on a particular function space. This explains why we prefer to
write (1.2) with a general Banach function space X.
 A2 CONJECTURE 3

2. Preliminaries
2.1. Calderón-Zygmund operators. By a Calderón-Zygmund op-
erator in Rn we mean an L2 bounded integral operator represented
as Z
T f (x) = K(x, y)f (y)dy, x 6∈ supp f,
Rn
with kernel K satisfying the following growth and smoothness condi-
tions:
c
(i) |K(x, y)| ≤ |x−y| n for all x 6= y;

(ii) there exists 0 < δ ≤ 1 such that

′ ′ |x − x′ |δ
|K(x, y) − K(x , y)| + |K(y, x) − K(y, x )| ≤ c ,
|x − y|n+δ
whenever |x − x′ | < |x − y|/2.
Given a Calderón-Zygmund operator T , define its maximal truncated
version by Z
T♮ f (x) = sup K(x, y)f (y)dy .
0<ε<ν ε<|y|<ν

2.2. Dyadic grids. Recall that the standard dyadic grid in Rn consists
of the cubes
2−k ([0, 1)n + j), k ∈ Z, j ∈ Zn .
Denote the standard grid by D.
By a general dyadic grid D we mean a collection of cubes with the
following properties: (i) for any Q ∈ D its sidelength ℓQ is of the form
2k , k ∈ Z; (ii) Q ∩ R ∈ {Q, R, ∅} for any Q, R ∈ D; (iii) the cubes of a
fixed sidelength 2k form a partition of Rn .
Given a cube Q0 , denote by D(Q0 ) the set of all dyadic cubes with
respect to Q0 , that is, the cubes from D(Q0 ) are formed by repeated
subdivision of Q0 and each of its descendants into 2n congruent sub-
cubes. Observe that if Q0 ∈ D, then each cube from D(Q0 ) will also
belong to D.
A well known principle says that there are ξn general dyadic grids Dα
such that every cube Q ⊂ Rn is contained in some cube Q′ ∈ Dα such
that |Q′ | ≤ cn |Q|. For ξn = 3n this is attributed in the literature to
M. Christ and, independently, to J. Garnett and P. Jones. For ξn = 2n
it can be found in a recent work by T. Hytönen and C. Pérez [11].
Very recently it was shown by J. Conde et al. [4] that one can take
ξn = n + 1, and this number is optimal. For our purposes any of such
variants is suitable. We will use the one from [11].
4 ANDREI K. LERNER

Proposition 2.1. There are 2n dyadic grids Dα such that for any cube
Q ⊂ Rn there exists a cube Qα ∈ Dα such that Q ⊂ Qα and ℓQα ≤ 6ℓQ .
The grids Dα here are the following:
Dα = {2−k ([0, 1)n + j + α)}, α ∈ {0, 1/3}n.
We outline briefly the proof. First, it is easy to see that it suffices to
consider the one-dimensional case. Take an arbitrary interval I ⊂ R.
Fix k0 ∈ Z such that 2−k0 −1 ≤ 3ℓI < 2−k0 . If I does not contain any
point 2−k0 j, j ∈ Z, then I is contained in some I ′ = [2−k0 j, 2−k0 (j + 1))
(since such intervals form a partition of R), and ℓI ′ ≤ 6ℓI . On the
other hand, if I contains some point j0 2−k0 , then I does not contain
any point 2−k0 (j + 1/3), j ∈ Z (since ℓI < 2−k0 /3), and therefore I is
contained in some I ′′ = [2−k0 (j + 1/3), 2−k0 (j + 4/3)), and ℓI ′′ ≤ 6ℓI .

2.3. Local mean oscillations. Given a measurable function f on Rn

and a cube Q, the local mean oscillation of f on Q is defined by

∗

ωλ (f ; Q) = inf (f − c)χQ λ|Q| (0 < λ < 1),
c∈R

where f ∗ denotes the non-increasing rearrangement of f .

By a median value of f over Q we mean a possibly nonunique, real
number mf (Q) such that


max |{x ∈ Q : f (x) > mf (Q)}|, |{x ∈ Q : f (x) < mf (Q)}| ≤ |Q|/2.
It is easy to see that the set of all median values of f is either one
point or the closed interval. In the latter case we will assume for the
definiteness that mf (Q) is the maximal median value. Observe that it
follows from the definitions that
(2.1) |mf (Q)| ≤ (f χQ )∗ (|Q|/2).
#,d
Given a cube Q0 , the dyadic local sharp maximal function Mλ;Q 0
f is
defined by
#,d
Mλ;Q 0
f (x) = sup ωλ (f ; Q′ ).
x∈Q′ ∈D(Q 0)

We say that {Qkj } is a sparse family of cubes if: (i) the cubes Qkj are
disjoint in j, with k fixed; (ii) if Ωk = ∪j Qkj , then Ωk+1 ⊂ Ωk ; (iii)
|Ωk+1 ∩ Qkj | ≤ 21 |Qkj |.
The following theorem was proved in [18] (its very similar version
can be found in [17]).
 A2 CONJECTURE 5

Theorem 2.2. Let f be a measurable function on Rn and let Q0 be a

fixed cube. Then there exists a (possibly empty) sparse family of cubes
Qkj ∈ D(Q0 ) such that for a.e. x ∈ Q0 ,
X
|f (x) − mf (Q0 )| ≤ 4M #,d
1
;Q
f (x) + 2 k
1 (f ; Q )χ k (x).
ω n+2 j Qj
0 2
2n+2
k,j

The following proposition is well known, and it can be found in a

slightly different form in [13]. We give its proof here for the sake of the
completeness. The proof is a classical argument used, for example, to
show that T is bounded from L∞ to BMO. Also the same argument
is used to prove a good-λ inequality relating T and M.
Proposition 2.3. For any cube Q ⊂ Rn ,
∞  
1 1
X Z
(2.2) ωλ (T f ; Q) ≤ c(T, λ, n) mδ m Q|
|f (y)|dy
m=0
2 |2 2mQ

and
∞  
1 1
X Z
(2.3) ωλ (T♮ f ; Q) ≤ c(T, λ, n) mδ m Q|
|f (y)|dy .
m=0
2 |2 2mQ

Proof. Let f1 = f χ2√nQ and f2 = f − f1 . If x ∈ Q and x0 is the center

of Q, then by the kernel assumptions,
Z
|T (f2 )(x) − T (f2 )(x0 )| ≤ √
|f (y)||K(x, y) − K(x0 , y)|dy
Rn \2 nQ
∞
|f (y)| 1
Z X Z
≤ cℓδQ dy ≤ cℓQδ
|f (y)|dy
|x − y|n+δ m=0
(2 ℓQ )n+δ
m
Rn \2Q 2m+1 Q\2m Q
∞  
1 1
X Z
≤c |f (y)|dy .
m=0
2mδ |2m Q| 2m Q

From this and from the weak type (1, 1) of T ,

∗

T f − T (f2 )(x0 ) χQ λ|Q|
≤ (T (f1 ))∗ (λ|Q|) + kT (f2 ) − T (f2 )(x0 )kL∞ (Q)
∞  
1 1 1
Z X Z
≤c |f (y)|dy + c mδ m Q|
|f (y)|dy
|Q| 2√n|Q| m=0
2 |2 2mQ

∞  
1 1
X Z
′
≤c mδ m Q|
|f (y)|dy ,
m=0
2 |2 2mQ

which proves (2.2).

6 ANDREI K. LERNER

The same inequalities hold for T♮ as well, which gives (2.3). The
only trivial difference in the argument is that one needs to use the
sublinearity of T♮ instead of the linearity of T . 

3. Proof of (1.2)
Combining Proposition 2.3 and Theorem 2.2 with Q0 ∈ D, we get
that there exists a sparse family S = {Qkj } ∈ D such that for a.e.
x ∈ Q0 ,
∞
 X 1 
|T♮ f (x) − mQ0 (T♮ f )| ≤ c(n, T ) Mf (x) + T
mδ S,m
|f |(x) ,
m=0
2
where M is the Hardy-Littlewood maximal operator and
X
TS,m f (x) = f2m Qkj χQkj (x).
j,k
1
If f ∈ L , then it follows from (2.1) that |mQ (T♮ f )| → 0 as |Q| → ∞.
Therefore, letting Q0 to anyone of 2n quadrants and using Fatou’s
lemma, we get
∞
 X 1 
kT♮ f kX ≤ c(n, T ) kMf kX + sup kT S,m |f |k X
m=0
2mδ S∈D
(for the notion of the Banach function space X we refer to [1, Ch. 1]).
Hence, (1.2) will follow from
(3.1) kMf kX ≤ c(n) sup kAD,S f kX (f ≥ 0)
D,S

and
(3.2) sup kTS,mf kX ≤ c(n)m sup kAD,S f kX (f ≥ 0).
S∈D D,S

Inequality (3.1) was proved in [18]; we give the proof here for the sake
of the completeness. The proof is just a combination of Proposition 2.1
and the Calderón-Zygmund decomposition. First, by Proposition 2.1,
2 n
X
n
(3.3) Mf (x) ≤ 6 M Dα f (x).
α=1
Second, by the Calderón-Zygmund decomposition, if
Ωk = {x : M d f (x) > 2(n+1)k } = ∪j Qkj and Ejk = Qkj \ Ωk+1 ,
then the family {Qkj } is sparse and
X
M d f (x) ≤ 2n+1 fQkj χEjk (x) ≤ 2n+1 Af (x).
k,j
 A2 CONJECTURE 7

From this and from (3.3),

2 n
X
n
(3.4) Mf (x) ≤ 2 · 12 ADα ,Sα f (x),
α=1

where Sα ∈ Dα depends on f . This implies (3.1) with c(n) = 2 · 24n .

We turn now to the proof of (3.2). Fix a family S = {Qkj } ∈ D.
Applying Proposition 2.1 again, we can decompose the cubes Qkj into
2n disjoint families Fα such that for any Qkj ∈ Fα there exists a cube
Qkj,α ∈ Dα such that 2m Qkj ⊂ Qkj,α and ℓQkj,α ≤ 6ℓ2m Qkj . Hence,
2n
X X
n
TS,m f (x) ≤ 6 fQkj,α χQkj (x).
α=1 j,k:Qkj ∈Fα

Set X
Am,α f (x) = fQkj,α χQkj (x).
j,k

We have that (3.2) will follow from

(3.5) kAm,α f kX ≤ c(n)m sup kAD,S f kX (f ≥ 0).
D,S

Consider the formal adjoint to Am,α :

X 1 Z 
⋆
Am,α f = k
f χQkj,α (x).
j,k
|Qj,α | Q k
j

Proposition 3.1. For any m ∈ N,

kA⋆m,α f kL2 = kAm,α f kL2 ≤ 8kf kL2 .
Proof. Set Ejk = Qkj \ Ωk+1 . Observe that the sets Ejk are pairwise
disjoint and |Qkj | ≤ 2|Ejk |. From this,
Z X XZ
k
(Am,α f )gdx = fQkj,α gQkj |Qj | ≤ 2 (M Dα f )(M d g)dx
Rn k,j k,j Ejk
Z
≤ 2 (M Dα f )(M d g)dx.
Rn

From this, using Hölder’s inequality, the L2 boundedness of M d and

duality, we get the L2 bound for Am,α . 
Lemma 3.2. For any m ∈ N,
kA⋆m,α f kL1,∞ ≤ c(n)mkf kL1 .
8 ANDREI K. LERNER

Proof. Set Ω = {x : Mf (x) > α} and let Ω = ∪l Ql be a Whitney

decomposition such that 3Ql ⊂ Ω, where Ql ∈ D (see, e.g., [1, p.
348]). Set also X
bl = (f − fQl )χQl , b = bl
l
and g = f − b. We have
|{x : |A⋆m,α f (x)| > α}| ≤ |Ω| + |{x : |A⋆m,αg(x)| > α/2}|
(3.6) + |{x ∈ Ωc : |A⋆m,αb(x)| > α/2}|.
c(n)
Further, |Ω| ≤ α
kf kL1 , and, by the L2 boundedness of A⋆m,α ,
4 c c c
|{x : |A⋆m,α g(x)| > α/2}| ≤ 2
kA⋆m,α gk2L2 ≤ 2 kgk2L2 ≤ kgkL1 ≤ kf kL1
α α α α
(we have used here that g ≤ cα).
It remains therefore to estimate the term in (3.6). For x ∈ Ωc con-
sider XX 1 Z 
⋆
Am,α b(x) = k
b l χQkj,α (x).
l k,j
|Q j,α | Q k
j

The second sum is taken over those cubes Qkj for which Qkj ∩ Ql 6= ∅. If
Ql ⊆ Qkj , then (bl )Qkj = 0. Therefore one can assume that Qkj ⊂ Ql . On
the other hand, Qkj,α ∩Ωc 6= ∅. Since 3Ql ⊂ Ω, we have that Ql ⊂ 3Qkj,α.
Hence
ℓQl ≤ 3ℓQkj,α ≤ 18 · 2m ℓQkj .
The family of all dyadic cubes Q for which Q ⊂ Ql and ℓQl ≤ 18 · 2m ℓQ
can be decomposed into m+4 families of disjoint cubes of equal length.
Therefore, X
χQkj ≤ (m + 4)χQl .
k,j:Qkj ⊂Ql ⊂3Qkj,α

From this we get

2
|{x ∈ Ωc : |A⋆m,α b(x)| > α/2}| ≤ kA⋆m,α bkL1 (Ωc )
α
2X 2(m + 4) X
X Z Z
≤ |bl |dx ≤ |bl |dx
α l k k Qkj α l Ql
k,j:Qj ⊂Ql ⊂3Qj,α

4(m + 4)
≤ kf kL1 .
α
The proof is complete. 
Lemma 3.3. For any cube Q ∈ Dα ,
ωλn (A⋆m,α f ; Q) ≤ c(n)mfQ (λn = 1/2n+2).
 A2 CONJECTURE 9

Proof. For x ∈ Q,
X  1 Z  X  1
Z 
f χQkj,α (x) = f ≡ c.
|Qkj,α| Qkj |Qkj,α | Qkj
k,j:Q⊆Qkj,α k,j:Q⊆Qkj,α

Hence
1
X  Z 
|A⋆m,αf (x)−c|χQ (x) = f χQkj,α (x) ≤ A⋆m,α (f χQ )(x).
|Qkj,α | Qkj
k,j:Qkj,α⊂Q

From this and from Lemma 3.2,

inf ((A⋆m,α f − c)χQ )∗ (λn |Q|) ≤ (A⋆m,α (f χQ ))∗ (λn |Q|) ≤ c(n)mfQ ,
c

which completes the proof. 

We are ready now to prove (3.5). One can assume that the sum
defining Am,α is finite. Then mA⋆m,α f (Q) = 0 for Q big enough. Hence,
By Lemma 3.3 and Theorem 2.2, for a.e. x ∈ Q (where Q ∈ Dα ),
A⋆m,α f (x) ≤ c(n)m Mf (x) + ASα,Dα f (x) .

From this and from (3.4), for any g ≥ 0 we have

Z Z
(Am,α f )gdx = f (A⋆m,αg)dx
Rn Rn
n +1 Z
2X
≤ cn m f (ADα ,Sα g)dx
α=1 Rn
n +1 Z
2X
= cn m (ADα ,Sα f )gdx ≤ c′n m sup kAD,S f kX kgkX ′ .
α=1 Rn D,S

Taking here the supremum over g with kgkX ′ = 1 completes the proof.

Added in proof. We have just learned that T. Hytönen, M. Lacey

and C. Pérez [10] have also found a proof of the A2 conjecture avoiding
a representation of T in terms of Haar shifts. The first step in this
proof is the same: the “local mean oscillation decomposition” combined
with Proposition 2.3 which reduces the problem to operators Am,α . In
order to handle Am,α , the authors use the result from [8] where it
was observed that this operator can be viewed as a positive Haar shift
operator of complexity m. As we have mentioned previously, our proof
avoids completely the notion of the Haar shift operator, and to bound
Am,α we apply the decomposition again (as it is shown starting with
Lemma 3.2).
10 ANDREI K. LERNER

References
[1] C. Bennett and R. Sharpley, Interpolation of Operators, Academic Press, New
York, 1988.
[2] O.V. Beznosova, Linear bound for the dyadic paraproduct on weighted Lebesgue
space L2 (w), J. Funct. Anal., 255 (2008), no. 4, 994-1007.
[3] S.M. Buckley, Estimates for operator norms on weighted spaces and reverse
Jensen inequalities, Trans. Amer. Math. Soc., 340 (1993), no. 1, 253–272.
[4] J.M. Conde, J. Garcı́a-Cuerva and J. Parcet, Sharp dyadic cover-
ings and nondoubling Calderón-Zygmund theory, preprint. Available at
http://arxiv.org/abs/1201.3513
[5] D. Cruz-Uribe, J.M. Martell and C. Pérez, Sharp weighted estimates for ap-
proximating dyadic operators, Electron. Res. Announc. Math. Sci. 17 (2010),
12–19.
[6] D. Cruz-Uribe, J.M. Martell and C. Pérez, Sharp weighted estimates for clas-
sical operators, Adv. Math., 229 (2012), no. 1, 408–441.
[7] T.P. Hytönen, The sharp weighted bound for general Calderón-
Zygmund operators, to appear in Annals of Math. (2012). Available at
http://arxiv.org/abs/1007.4330
[8] T.P. Hytönen and M. Lacey, The Ap − A∞ inequality for general Calderón–
Zygmund operators, to appear in Indiana Univ. Math. J. Available at
http://arxiv.org/abs/1106.4797
[9] T.P. Hytönen, M.T. Lacey, H. Martikainen, T. Orponen, M.C. Reguera, E.T.
Sawyer and I. Uriarte-Tuero, Weak and strong type estimates for maximal
truncations of Calderón-Zygmund operators on Ap weighted spaces, preprint.
Available at http://arxiv.org/abs/1103.5229
[10] T.P. Hytönen, M.T. Lacey and C. Pérez, Non-probabilistic proof of the A2
theorem, and sharp weighted bounds for the q-variation of singular integrals,
preprint. Available at http://arxiv.org/abs/1202.2229
[11] T.P. Hytönen and C. Pérez, Sharp weighted bounds involving A∞ , to appear
in J. Analysis&PDE. Available at http://arxiv.org/abs/1103.5562
[12] T.P. Hytönen, C. Pérez, S. Treil and A. Volberg, Sharp weighted es-
timates for dyadic shifts and the A2 conjecture, preprint. Available at
http://arxiv.org/abs/1010.0755
[13] B. Jawerth and A. Torchinsky, Local sharp maximal functions, J. Approx.
Theory, 43 (1985), 231–270.
[14] M.T. Lacey, An Ap -A∞ inequality for the Hilbert transform, preprint. Available
at http://arxiv.org/abs/1104.2199
[15] M.T. Lacey, On the A2 inequality for Calderón-Zygmund operators, preprint.
Available at http://arxiv.org/abs/1106.4802
[16] M.T. Lacey, S. Petermichl and M.C. Reguera, Sharp A2 inequality for Haar
Shift Operators, Math. Ann. 348 (2010), no. 1, 127-141.
[17] A.K. Lerner, A pointwise estimate for the local sharp maximal function with
applications to singular integrals, Bull. London Math. Soc., 42 (2010), no. 5,
843–856.
[18] A.K. Lerner, On an estimate of Calderón-Zygmund operators by dyadic positive
operators, preprint. Available at http://arxiv.org/abs/1202.1860
 A2 CONJECTURE 11

[19] F. Nazarov, A. Reznikov and A. Volberg, The proof of A2 conjec-

ture in a geometrically doubling metric space, preprint. Available at
http://arxiv.org/abs/1106.1342
[20] F. Nazarov and A. Volberg, A simple sharp weighted estimate of the
dyadic shifts on metric spaces with geometric doubling, preprint. Available at
http://arxiv.org/abs/1104.4893
[21] C. Pérez, S. Treil and A. Volberg, On A2 conjecture and corona decomposition
of weights, preprint. Available at http://arxiv.org/abs/1006.2630
[22] S. Petermichl, The sharp bound for the Hilbert transform on weighted Lebesgue
spaces in terms of the classical Ap - characteristic, Amer. J. Math., 129 (2007),
no. 5, 1355–1375.
[23] S. Petermichl, The sharp weighted bound for the Riesz transforms, Proc. Amer.
Math. Soc., 136 (2008), no. 4, 1237–1249.
[24] S. Petermichl and A. Volberg, Heating of the Ahlfors-Beurling operator: weakly
quasiregular maps on the plane are quasiregular, Duke Math. J. 112 (2002),
no. 2, 281–305.
[25] S. Treil, Sharp A2 estimates of Haar shifts via Bellman function, preprint.
Available at http://arxiv.org/abs/1105.2252

Department of Mathematics, Bar-Ilan University, 52900 Ramat Gan,

Israel
E-mail address: aklerner@netvision.net.il

View publication stats