For a proper simply connected domain D andw ∈ D, let CR(w;D)denote the conformal radius of Dwith respect tow, i.e., CR(w;D) ≡ f′(0)for f the unique conformal mapD → Dwith f(0) = wand f′(0) > 0. Let rad(w;D) ≡ inf{r : Br(w) ⊇ D}denote the out-radius of Dwith respect tow. By the Schwarz lemma and the Koebe one-quarter theorem,
dist(w, ∂D)≤CR(w;D)≤ [4 dist(w, ∂D)] ∧rad(w;D). (2.15) Further (see e.g., [30, Theorem 1.3])
|ζ|
(1+ |ζ|)2 ≤ |f(ζ )−w|
CR(w;D) ≤ |ζ|
(1− |ζ|)2 (2.16) As a consequence,
|ζ|
4 ≤ |f(ζ )−w|
CR(w;D) ≤4|ζ| (2.17)
where the right-hand inequality above holds for|ζ| ≤1/2.
Finally, we state the Beurling estimate [16, Theorem 3.76] which we will frequently use in conjunction with the conformal invariance of Brownian motion.
Theorem 2.9 (Beurling estimate) Suppose that B is a Brownian motion inCand τD=inf{t≥0:B(t)∈∂D}. There exists a constant c<∞such that ifγ: [0,1] → Cis a curve withγ (0)=0and|γ (1)| =1, z∈D, andPzis the law of B when started at z, then
Pz[B([0, τD])∩γ ([0,1])=∅] ≤c|z|1/2. 3 The intersection of SLEκ(ρ)with the boundary 3.1 The upper bound
The main result of this section is the following theorem, which in turn implies Theo-rem1.8.
Theorem 3.1 Fixκ >0,ρ1,R >−2, andρ2,R∈Rsuch thatρ1,R+ρ2,R >κ2−4. Fix xR ∈ [0+,1)and letηbe anSLEκ(ρ1,R, ρ2,R)process with force points(xR,1). Let
α= 1
κ(ρ1,R+2)
ρ1,R+ρ2,R+4−κ 2
. (3.1)
For eachǫ > 0, letτǫ = inf{t ≥ 0 : η(t) ∈ ∂B(1, ǫ)}and, for each r > 0, let σr =inf{t ≥0:η(t)∈∂(rD)}. For eachδ ∈ [0,1)and r≥2fixed, let
Eǫδ,r = {τǫ< σr, Im(η(τǫ))≥δǫ}. (3.2) We have that
P[Eǫδ,r] =ǫα+o(1) as ǫ→0. (3.3) Theo(1)in the exponent of (3.3) tends to 0 asǫ → 0 and depends only onκ, δ,xR, and the weightsρ1,R,ρ2,R. Theo(1), however, is uniform inr ≥ 2. Taking ρ1,R> (−2)∨(κ2−4)andρ2,R=0, we have that
α= 1
κ(ρ+2)
ρ+4−κ 2
. (3.4)
Thus Theorem 3.1 leads to the upper bound of Theorem 1.6. We begin with the following lemma which contains the same statement as Theorem3.1except is restricted to the case thatδ∈(0,1)and, in particular, is not applicable forδ=0.
Lemma 3.2 Assume that we have the same setup and notation as in Theorem3.1.
Then for eachδ∈(0,1)and r≥2fixed, we have that P[Eǫδ,r] ≍ǫα
where the constants in≍depend only onκ,δ, xR, and the weightsρ1,R,ρ2,R. Proof Forη, the SLEκ(ρ1,R, ρ2,R)process with force points(xR,1), let(gt)be the associated Loewner evolution and letVtRdenote the evolution ofxR. From (2.6) we know that
Mt =
gt(1)−VtR gt′(1)
−α
gt(1)−Wt
gt(1)−VtR
−2κ(ρ1,R+ρ2,R+4−κ/2)
is a local martingale and the law ofηreweighted byMis that of an SLEκ(ρ1,R,ρ2,R) process whereρ2,R= −2ρ1,R−ρ2,R−8+κ. We writeK =KτǫandK = {z:z∈ K}. LetGbe the extension ofgτǫtoC\(K∪K)which is obtained by Schwarz reflection.
By (2.15), we have
G′(x)dist(x,K)≍dist(G(x),G(K∪K)). (3.5) Observe that G(K ∪K) = [OτL
ǫ,OτR
ǫ] where OtL (resp. OtR) is the image of the leftmost (resp. rightmost) point ofKt∩Rundergt. Note that (3.5) implies
ǫgτ′
ǫ(1)≍gτǫ(1)−OτR
ǫ.
It is clear thatgt(1)−Wt ≥ gt(1)−OtR ≥ gt(1)−VtR. On the eventEǫδ,r, we run a Brownian motion started from the midpoint of the line segment[1, η(τǫ)]. Then
0 xR 1 η
B(1 ) ϕ
−=ϕ(∞) ϕ(xR) ϕ(η)
B(− 1) =ϕ(B(1 ))
0
Fig. 9 The image of an SLEκ(ρ1,R, ρ2,R)process inHfrom 0 to∞with force points(xR,1)under ϕ(z) =ǫz/(1−z)has the same law as an SLEκ(ρL;ρR)process inHfrom 0 to∞with force points (−ǫ;ǫxR/(1−xR))whereρR=ρ1,RandρL=κ−6−(ρ1,R+ρ2,R)
this Brownian motion has uniformly positive (thoughδ-dependent) probability to exit H\K through each of the left side ofK, the right side ofK, the interval[xR,1], and the interval(1,∞). Consequently, by the conformal invariance of Brownian motion,
gτǫ(1)−Wτǫ ≍gτǫ(1)−OτR
ǫ ≍gτǫ(1)−VτR
ǫ on Eǫδ,r.
These facts imply that Mτǫ ≍ǫ−α onEǫδ,r where the constants in≍depend only onκ,δ,xR, and the weightsρ1,R,ρ2,R. Thus
P[Eǫδ,r] ≍ǫαE[Mτǫ1
Eδ,rǫ ] =ǫαP⋆[Eδ,rǫ ]
whereP⋆is the law ofηweighted by the martingale M. As we remarked earlier,P⋆ is the law of an SLEκ(ρ1,R,ρ2,R)with force points(xR,1).
We now perform a coordinate change using the Möbius transformationϕ(z) = ǫz/(1−z). Then the law of the image of a path distributed according toP⋆underϕis equal to that of an SLEκ(2+ρ1,R+ρ2,R;ρ1,R)process inHfrom 0 to∞with force points(−ǫ;ǫxR/(1−xR))(see Fig.9). Note that 2+ρ1,R+ρ2,R ≥ κ2 −2 by the hypotheses of the lemma. Letη⋆be an SLEκ(2+ρ1,R+ρ2,R;ρ1,R)process inHfrom 0 to∞with force points(−ǫ;ǫxR/(1−xR)). In particular, by Lemma2.1,η⋆almost surely does not hit(−∞,−ǫ). Under the coordinate change, the eventEǫδ,r becomes {σ1,ǫ⋆ < ξǫ,r⋆ , Im(η⋆(σ1,ǫ⋆ )) ≥ δ}whereσ1,ǫ⋆ is the first time thatη⋆ hits∂B(−ǫ,1), ξǫ,r⋆ is the first time thatη⋆hits∂B(−ǫr2/(r2−1), ǫr/(r2−1)). By Lemma2.4, the probability of the event{σ1,ǫ⋆ < ξǫ,r⋆ , Im(η⋆(σ1,ǫ⋆ ))≥δ}is bounded from below by a positive constant depending only onκ,δ,ρ1,R, andρ2,R. ThusP⋆[Eǫδ,r] ≍1 which impliesP[Eǫδ,r] ≍ǫαand the constants in≍depend only onκ,δ,xR, and the weights
ρ1,R,ρ2,R. ⊓⊔
Corollary 3.3 Fixκ >0,ρL >−2, ρ1,R >−2andρ2,R ∈Rsuch thatρ1,R+ρ2,R>
κ
2 −4. Fix xL ≤ 0,xR ∈ [0+,1)and let η be an SLEκ(ρL;ρ1,R, ρ2,R)process with force points(xL;xR,1). Let Eδ,rǫ be the event as in Theorem3.1, then for each δ∈(0,1)and r ≥2fixed, we have that
P[Eǫδ,r] ≍ǫα
where the constants in≍depend only onκ,δ, r , xL, xR, and the weightsρL, ρ1,R,ρ2,R. Proof Let(gt)be the Loewner evolution associated withηand letVtL,VtRdenote the evolution ofxL,xR, respectively, undergt. From (2.6) we know that
Mt =
gt(1)−VtR g′t(1)
−α
×
gt(1)−Wt gt(1)−VtR
−2κ(ρ1,R+ρ2,R+4−κ/2)
×(gt(1)−VtL)−ρκL(ρ1,R+ρ2,R+4−κ/2)
is a local martingale which yields that the law of ηreweighted by M is that of an SLEκ(ρL;ρ1,R,ρ2,R)process whereρ2,R = −2ρ1,R−ρ2,R−8+κ. Note that, by similar analysis in Lemma3.4, the termgτǫ(1)−VτLǫ is bounded both from below and above by positive finite constants depending only onr on the eventEǫδ,r. The rest of the analysis in the proof of Lemma3.2applies similarly in this setting. ⊓⊔
Throughout the rest of this subsection, we let:
T=R×(0,1). (3.6)
Lemma 3.4 Letηbe a continuous curve inHstarting from0with continuous Loewner driving function W and let(gt)be the corresponding family of conformal maps. For each t ≥0, let OtL(resp. OtR) be the leftmost (resp. rightmost) point of gt(η([0,t]))in R. There exists a universal constant C ≥1such that the following is true. Fixϑ >0 and letσ be the first time thatηexitsϑT. Then
|Wσ −Oσq| ≥ ϑ
C for q∈ {L,R}. (3.7)
Letζ be the first time thatηexitsD∩ϑT. Then
|Wt−Otq| ≤Cϑ for q∈ {L,R} and all t ∈ [0, ζ]. (3.8) Finally, ifηexitsD∩ϑTthrough the right side of∂D∩ϑT, then
|Wζ−OζL| ≥ 1
C. (3.9)
Proof Forz∈C, we letPz denote the law of a Brownian motionBinCstarted atz.
By [16, Remark 3.50] we have that
|Wσ −OσL| = lim
y→∞yPyi[BexitsH\η[0, σ]on the left side ofη([0, σ])].
|Wσ −OσL| ≥ lim
y→∞yPyi[Bτ ∈I]
×Pyi[BexitsH\η([0, σ])on the left side ofη([0, σ])|Bτ ∈I]. (3.10) We have,
ylim→∞yPyi[Bτ ∈ I]= lim
y→∞
I−ϑi
1 π
y(y−ϑ ) w2+(y−ϑ )2dw
=
I−ϑi
1
πdw= ϑ
π (3.11)
(recall the form of the Poisson kernel onH, see e.g., [16, Exercise 2.23]). It is easy to see that there exists a universal constant p0>0 such that for anyz∈I,
Pz[BexitsH\η[0, σ]on the left side ofη([0, σ])]≥ p0. (3.12) Combining (3.10) with (3.11) and (3.12) gives (3.7). The bounds (3.8) and (3.9) are
proved similarly. ⊓⊔
Lemma 3.5 Fixκ >0,ρL ∈(κ2−4,κ2−2), andρR>−2. Letηbe anSLEκ(ρL;ρR) process with force points(−ǫ;xR)for xR ≥ 0+andǫ > 0. Letσ1 =inf{t ≥ 0 : η(t)∈∂D}. Define, for u≥0, TuL =inf{t ≥0:Wt−VtL =u}, where VtLdenotes the evolution of xL. Let p2 = p2(12)be the constant from Lemma2.6. There exists constantsǫ0>0,ϑ0>0, and C>0such that for allǫ∈(0, ǫ0)andϑ∈(0, ϑ0)we have
P[σ1<T0L∧TϑL] ≤ p21/(Cϑ ). Proof LetEϑ = {σ1<T0L∧TϑL}. By definition, we have that
|Wt−VtL|< ϑ for all t ∈ [0, σ1] on Eϑ. (3.13) By (3.7) of Lemma3.4there exists a constantC1>0 such thatη([0, σ1])⊆C1ϑT.
Moreover,ηexitsD∩ C1ϑT
on its left side for allϑ >0 small enough because a Brownian motion argument [(analogous to (3.9)] implies there exists a constant C2 >0 such that|Wσ1−VσL
1| ≥C2on the event thatηexits through the right side, contradicting (3.13).
SupposeC>0; we will set its value later in the proof. For each 1≤k≤ Cϑ1 , we let
Lk= {z∈H:Re(z)= −kCϑ} and ζk =inf{t≥0:η(t)∈Lk}.
gζk(·)−Wζk Wζk−VL
ζk
−kCϑ
−(k+1)Cϑ
η zkϑ=η(ζk)−ϑ
Lk
Lk+1
zϑk ϑ
0
−1 Lk+1
Fig. 10 Illustration of the justification of (3.14) in the proof of Lemma3.5
On Eϑ, we have thatζ1 < ζ2 <· · · < σ1 <T0L. For eachk, let Fk = {ζk <TϑL} and letFkbe theσ-algebra generated byη|[0,ζk]. To complete the proof, we will show that
P[ζk+1<T0L|Fk]1Fk ≤ p21Fk for each 1≤k≤ 1 Cϑ
where p2=p2(12)is the constant from Lemma2.6. To see this, we just need to show thatgζk(η|[ζk,ζk+1])satisfies the hypotheses of Lemma2.6and that with
Lk+1= gζk(Lk+1)−Wζk Wζk−VζL
k
we have thatLk+1∩2D=∅onFk. Therefore it suffices to prove
dist(Wζk,gζk(Lk+1)) Wζk−VζL
k
→ ∞ on Fk as C→ ∞. (3.14)
LetBbe a Brownian motion starting fromzϑk =η(ζk)−ϑand letHk+1= {z∈H: Re(z)≥ −(k+1)Cϑ}be the subset ofHwhich is to the right ofLk+1(see Fig.10).
The probability thatBexitsHk+1\η([0, ζk])through the right side ofη([0, ζk])(blue) is1, through(−(k+1)Cϑ,−kCϑ )(green) is1, and throughLk+1(orange) is 1/C (since this probability is less than the probability that the Brownian motion exits{z ∈ C : −(k+1)Cϑ < Re(z) < −kCϑ}through Lk+1which is less than 1/C). Let
zϑk ≡xkϑ+ykϑi ≡ gζk(zkϑ)−Wζk Wζk−VζL
k
.
By the conformal invariance of Brownian motion, we have that dist(zϑk,Lk+1)
ykϑ C. (3.15)
(gζk(Hk+1)−Wζk)/(Wζk −VζL
k)throughLk+1is bounded from below by a positive universal constant times the probability that a Brownian motion starting fromzϑk exits B(zϑk,d)∩H,d =dist(zϑk,Lk+1), through∂B(zϑk,d)∩H. This latter probability is bounded from below by a positive universal constant timesykϑ/d. Thus 1/Cykϑ/d, as desired.
The conformal invariance of Brownian motion and the estimates above also imply that sin(arg(zϑk))≍1, hence|zϑk| ≍ |ykϑ|. Combining this with (3.15) implies that
dist(zϑk,Lk+1)
|zϑk| C.
Thus, by the triangle inequality,
dist(Lk+1,0)C|zϑk|
(providedCis large enough). Since|zϑk| ≍1, this proves (3.14), hence the lemma.⊓⊔ Proof of Theorem3.1 Lemma3.2implies the lower bound in (3.3) because we can take, e.g.,δ =12. In order to prove the upper bound, it is sufficient to show
P[τǫ <∞] ≤ǫα+o(1) as ǫ→0.
We are first going to perform a change of coordinates. Letϕ: H → Hbe the Möbius transformationz → ϕ(z):=ǫz/(1−z). FixxR ∈ [0+,1)and letηbe an SLEκ(ρ1,R, ρ2,R)process with force points located at(xR,1)as in Theorem3.1. Then the law ofη=ϕ(η)is that of an SLEκ(ρL;ρR)process with force points(−ǫ;xR) wherexR=ǫxR/(1−xR)and
ρL =κ−6−
ρ1,R+ρ2,R
and ρR=ρ1,R. (3.16) Letσ1be the first time thatηhits∂Dand letVtL,VtRdenote the evoltuion ofxL,xR
undergt, respectively. Foru ≥ 0, defineTuL = inf{t ≥ 0 : Wt −VtL = u}(as in the statement of Lemma3.5). Then it is sufficient to proveP[σ1 <T0L] ≤ ǫα+o(1). Note that the exponentαcomes from the sum of the exponent of|VtL−VtR|and the exponent of|Wt−VtL|in the left martingaleMLfrom (2.7) with these weights. For u ≥0, defineτuL =inf{t ≥0: MtL =u}. Note thatτ0L =T0L. Fixβ∈(0,1)and set ϑ=ǫβ. Foru >0, we have the bound
P[σ1< τ0L] ≤P[τuL < τ0L] +P[σ1< τ0L < τuL]. (3.17) We claim that exists constantsC1>0 andγ >0 depending only onρL,ρR, and κsuch that
|Wt−VtL|γ ≤C1MtL for all t ∈ [0, σ1]. (3.18)
Sinceρ1,R+ρ2,R > κ2 −4 it follows thatρL < κ2 −2. Therefore the sign of the exponent of|VtL−VtR|in the definition ofMtLis the same as the sign ofρR. IfρR≥0, then the exponent has a positive sign. In this case,MtL ≥ |Wt−VtL|αso that we can takeγ =α. Now suppose thatρR <0. By (3.8) of Lemma3.4we know that there exists a constantC2>0 such that
|VtL−VtR| ≤C2 for all t∈ [0, σ1]. (3.19) Thus, in this case, there exists a constant C3 > 0 such that MtL ≥ C3|Wt − VtL|(κ−4−2ρL)/κ. Therefore we can takeγ =(κ−4−2ρL)/κ. This proves the claimed bound in (3.18).
Setu=ϑγ/C1. To bound the second term on the right side of (3.17), we first note by (3.18) that
P[σ1< τ0L < τuL] ≤P[σ1<T0L∧TϑL]. (3.20) By Lemma3.5, we know that
P[σ1<T0L∧TϑL] ≤ p1/(Cϑ )2 . (3.21) We will now bound the first term on the right side of (3.17). Since τ0L, τuL are stopping times for the martingaleMLandMτ0∧τu =uP[τuL < τ0L], we have that
P[τuL < τ0L] = 1 uE[MτL
0∧τu] = M0L
u = ǫα
u(1−xR)(κ−4−2ρL)ρR/(2κ). (3.22) Combining (3.17) with (3.21) and (3.22) we get that P[σ1 < T0L] ≤ ǫα+o(1), as
desired. ⊓⊔
Recall that (see for example [24, Section 4]) the β-Hausdorff measure of a set A⊆Ris defined as
Hβ(A)= lim
ǫ→0+
Hβ
ǫ(A)
where Hβ
ǫ(A):=inf
⎧⎨
⎩ j |Ij|β : A⊆ ∪jIj and |Ij| ≤ǫ for all j
⎫⎬
⎭.
Proof of Theorem1.6forκ ∈(0,4),upper bound Fixκ∈(0,4), ρ ∈(−2,κ2−2).
Letηbe an SLEκ(ρ)process with a single force point located at 0+. Letα∈(0,1) be as in (3.4). Fix 0<x <y. We are going to prove the result by showing that
dimH(η∩ [x,y])≤1−α almost surely. (3.23) For eachk∈Zandn∈Nwe letIk,n= [k2−n, (k+1)2−n]and letzk,nbe the center ofIk,n. LetInbe the set ofksuch thatIk,n⊆ [x/2,2y]and letEk,nbe the event that
everyn ≥n0we have that{Ik,n :k∈In, Ek,noccurs}is a cover ofη∩ [x,y]. Fixζ >0. Theorem3.1implies that there exists a constantC1>0 (independent ofn) andn1=n1(ζ )such that
P[Ek,n] ≤C12−(α−ζ )n for each n≥n1 and k∈In. Consequently, there exists a constantC2>0 such that
E
Hβ
2−n(η∩ [x,y])
≤E
⎡
⎣
k∈In
2−βn1Ek,n
⎤
⎦≤C22−βn×2n×2−(α−ζ )n.
By Fatou’s lemma, E
H1−α+2ζ(η∩ [x,y])
≤lim inf
n E
H1−α+2ζ
2−n (η∩ [x,y])
≤lim inf
n C22−nζ =0.
This implies thatH1−α+2ζ(η∩ [x,y])=0 almost surely. This proves (3.23) which
completes the proof of the upper bound. ⊓⊔