prev

next

out of 21

View

0Download

0

Embed Size (px)

Math. Z. (2007) 257:525–545 DOI 10.1007/s00209-007-0132-5 Mathematische Zeitschrift

Doubling measures, monotonicity, and quasiconformality

Leonid V. Kovalev · Diego Maldonado · Jang-Mei Wu

Received: 25 September 2006 / Accepted: 2 January 2007 / Published online: 16 March 2007 © Springer-Verlag 2007

Abstract We construct quasiconformal mappings in Euclidean spaces by integration of a discontinuous kernel against doubling measures with suitable decay. The differ- entials of mappings that arise in this way satisfy an isotropic form of the doubling condition. We prove that this isotropic doubling condition is satisfied by the distance functions of certain fractal sets. Finally, we construct an isotropic doubling measure that is not absolutely continuous with respect to the Lebesgue measure.

Mathematics Subject Classification (2000) 30C65 · 28A75 · 42A55 · 47H05

1 Introduction

Given a nonatomic positive Radon measure µ on R, define an increasing function fµ(x) =

∫ x 0 dµ(z), so that f

′ µ = µ in the sense of distributions. It is well-known that µ is

L.V.K. was supported by an NSF Young Investigator award under grant DMS 0601926. J.-M.W. was supported by the NSF grant DMS 0400810.

L. V. Kovalev (B) Department of Mathematics, Texas A&M University, Mailstop 3368, College Station, TX 77843-3368, USA e-mail: lkovalev@math.tamu.edu

D. Maldonado Department of Mathematics, University of Maryland, College Park, MD 20742, USA

Present Address: D. Maldonado Department of Mathematics, Kansas State University, Manhattan, KS 66506, USA e-mail: dmaldona@math.ksu.edu

J.-M. Wu Department of Mathematics, University of Illinois, 1409 West Green Street, Urbana, IL 61801, USA e-mail: wu@math.uiuc.edu

526 L. V. Kovalev et al.

doubling if and only if fµ is quasisymmetric. We extend this relation between doubling measures and quasisymmetric mappings to higher dimensions. Observe that

fµ(x) = 12 ∫

R

{ x − z |x − z| +

z |z| }

dµ(z).

For a nonatomic Radon measure µ on Rn, n ≥ 1, we define fµ : Rn → Rn by

fµ(x) = 12 ∫

Rn

{ x − z |x − z| +

z |z| }

dµ(z). (1.1)

where |·| is now interpreted as the Euclidean norm. When n ≥ 2 the integrand in (1.1) is not compactly supported with respect to z. The integral (1.1) converges provided that µ satisfies the decay condition

∫

|z|>1 |z|−1 dµ(z) < ∞. (1.2)

For 0 < γ < n, let

Iγ µ(x) = ∫

Rn

|x − z|γ−n dµ(z)

be the Riesz potential of µ of order γ . Condition (1.2) is equivalent to In−1µ being finite almost everywhere. Here and in the sequel the words “almost everywhere” or “a.e.” refer to the Lebesgue measure. A positive Radon measure µ on Rn is called doubling if there is a constant C ≥ 1 such that µ(2B) ≤ Cµ(B) for every ball B ⊂ Rn. Here 2B stands for the ball that has the same center as B and twice its radius.

Theorem 1.1 Let µ be a doubling measure on Rn that satisfies the decay condition (1.2). The mapping fµ defined by (1.1) is η-quasisymmetric with η depending only on the dou- bling constant of µ. Furthermore, fµ is δ-monotone and for a.e. x ∈ Rn

1 12C3

In−1µ(x) ≤ ‖Dfµ(x)‖ ≤ πIn−1µ(x), (1.3) where C is the doubling constant of µ.

Theorem 1.1 expands the class of weights that are known to be comparable to Jacobians of quasiconformal mappings. The quasiconformal Jacobian problem posed by David and Semmes in [9] asks for a characterization of all such weights (see also [4,6,7,16,18]). The authors of [6] point out that such a characterization would give a good idea of which metric spaces are bi-Lipschitz equivalent to Rn.

A mapping f : Rn → Rn is called monotone if 〈F(x) − F(y), x − y〉 ≥ 0 for all x, y ∈ Rn, where 〈·, ·〉 is the inner product [2,27]. In other words, F is monotone if the angle formed by the vectors F(x) − F(y) and x − y is at most π/2. A stronger version of this condition, introduced by Sobolevskii in [19], requires the angle to be bounded by a constant less than π/2.

Definition 1.2 Let δ ∈ (0, 1]. A mapping F from a convex domain � ⊂ Rn into Rn is called δ-monotone if for all x, y ∈ �

〈F(x) − F(y), x − y〉 ≥ δ|F(x) − F(y)||x − y|. (1.4)

Doubling measures, monotonicity, and quasiconformality 527

Let � be a domain in Rn. An injective mapping f : � → Rn is called η-quasisymmetric if there is a homeomorphism η : [0, ∞) → [0, ∞) such that

|f (x) − f (z)| |f (y) − f (z)| ≤ η

( |x − z| |y − z|

)

, z ∈ �, x, y ∈ � \ {z}. (1.5)

In the case n ≥ 2, f is called K-quasiconformal if f ∈ W1,nloc (�; Rn) and the operator norm of the derivative Df satisfies ‖Df (x)‖n ≤ K det Df (x), a.e. in � for some K ≥ 1.

Every sense-preserving quasisymmetric mapping f : Rn → Rn, n ≥ 2, is quas- iconformal and vice versa ([12, Chap. 10], [24, p. 98]). The following result relates δ-monotone and quasisymmetric mappings.

Theorem 1.3 [14, Theorem 6] Suppose that n ≥ 2 and f : � → Rn is a nonconstant δ-monotone mapping. If B is a closed ball such that 2B ⊂ �, then f is η-quasisymmetric on B with η depending only on δ.

Quasiconformality is known to be related to the doubling condition in several ways [5,17,20]. For example, if f : Rn → Rn is quasiconformal, then ‖Df‖n is a dou- bling weight, and moreover an A∞ weight [10]. Consequently, ‖Df‖ is doubling as well. Since δ-monotonicity is a stronger property than quasiconformality, one can expect that the differential of a δ-monotone mappings exhibits a stronger doubling behavior. Our Theorem 1.5 confirms this. Before stating it, we observe that a Radon measure µ is doubling if and only if there exists a constant A such that

A−1 ≤ µ(Q1) µ(Q2)

≤ A

for any congruent cubes Q1 and Q2 with nonempty intersection. Recall that two sub- sets of Rn are called congruent if there is an isometry of Rn that maps one of them onto the other.

Definition 1.4 A Radon measure µ on Rn is isotropic doubling if there is a constant A ≥ 1 such that

A−1 ≤ µ(R1) µ(R2)

≤ A (1.6)

whenever R1 and R2 are congruent rectangular boxes with nonempty intersection.

Theorem 1.5 Let f : Rn → Rn be a nonconstant δ-monotone mapping, n ≥ 2. Then the weight ‖Df‖ is isotropic doubling.

The requirement (1.6) for all boxes regardless of their orientation and aspect ratio imposes a very strong condition on the measure when n ≥ 2. In particular, in any cube the projection of an isotropic doubling measure to any (n−1)-dimensional face of the cube is comparable to the (n − 1)-dimensional Lebesgue measure Ln−1 (Lemma 3.1). Theorem 1.6 For every n ≥ 2 there exists an isotropic doubling measure µ on Rn that is purely singular with respect to Ln.

Theorem 1.7 For every n ≥ 2 there exists an isotropic doubling measure µ on Rn and a bi-Lipschitz mapping f : Rn → Rn such that the pushforward measure f#µ is not isotropic doubling.

528 L. V. Kovalev et al.

Theorems 1.1, 1.5, 1.6, and 1.7 are proved in sections 2, 3, 5 and 6, respectively. We also prove that the distance functions of certain fractal subsets of Rn give rise to iso- tropic doubling weights (Proposition 3.6). By virtue of this result, the sets constructed by Semmes [18] and Laakso [16] provide examples of isotropic doubling weights that are not comparable to ‖Df‖ for any δ-monotone, or even quasiconformal, mapping f : Rn → Rn.

2 Doubling measures and monotone mappings

The balls, cubes, and rectangular boxes considered in this paper are assumed closed. A nonnegative locally integrable function on Rn is called a weight. A weight is dou- bling if the measure w(x)dLn(x) is doubling, where Ln is the n-dimensional Lebesgue measure. Throughout the paper we only consider positive nonzero measures.

In this section we study the mapping fµ defined by (1.1). In particular, we prove Theorem 1.1 and its more general version, Theorem 2.1. Given three distinct points x, y, z ∈ Rn, let l(x, y, z) = |x−y|+ |x−z|+ |y−z| denote the perimeter of the triangle xyz. Also let ẑxy be the angle between the vectors y − x and z − x, and define

τ(x, y, z) := π − max{ẑxy, ẑyx}.

Theorem 2.1 Let � ⊂ Rn be a convex domain, n ≥ 2. Let µ be a nonatomic Radon measure on Rn that satisfies (1.2). Suppose that there exists κ > 0 such that

lim inf y→x

⎛

⎝ ∫

Rn

τ 2(x, y, z) dµ(z) l(x, y, z)

⎞

⎠ / ⎛

⎝ ∫

Rn

τ(x, y, z) dµ(z) l(x, y, z)

⎞

⎠ ≥ κ (2.1)

for all x ∈ �. Then fµ is δ-monotone in � with δ = κ/(2π2).

The proof is preceded by an elementary lemma.

Lemma 2.2 Fix z ∈ Rn and define gz(x) = (x − z)/|x − z| for x ∈ Rn \ {z}. For any distinct points x, y ∈ Rn \ {z} we have

2|x − y| π l(x, y, z)

τ (x, y, z) ≤ |gz(x) − gz(y)| ≤ 4|x − y|l(x, y, z) τ (x, y, z) (2.2)

and

2|x − y|2 π2l(x, y, z)

τ (x, y, z)2 ≤ 〈gz(x) − gz(y), x − y〉 ≤ 4|x − y| 2

l(x, y, z) τ (x, y, z)2. (2.3)

Proof First we prove (2.2). By the sine theorem

sin x̂zy |x − y| =

sin ẑyx + sin ẑxy |x − z| + |y − z| . (2.4)

Express the sum of sines as a product:

sin ẑyx + sin ẑxy = 2 cos x̂zy 2

cos ẑyx