# Lp MAPPING PROPERTIES OF THE BERGMAN PROJECTION ON THE

HARTOGS TRIANGLE
DEBRAJ CHAKRABARTI AND YUNUS E. ZEYTUNCU

Abstract. We prove optimal estimates for the mapping properties of the Bergman projection
on the Hartogs triangle in weighted Lp spaces when p > 43 , where the weight is a power of the
distance to the singular boundary point. For 1 < p ≤ 34 we show that no such weighted estimates
are possible.

1. Introduction
1.1. Results. In this note we describe the Lp regularity of the Bergman projection on the Hartogs
triangle, the pseudoconvex domain in C2 defined as

 H = (z1 . z2 ) ∈ C2 .

the map BH is bounded and surjective from Lp (H) to Ap (H. This work was partially supported by a grant from the Simons Foundation (#316632 to Debraj Chakrabarti). 2010 Mathematics Subject Classification. Lp regularity.2.1. 1 . Hartogs triangle.1 below). Bergman projection. Proposition 5. We begin with the case p ≥ 2: Theorem 1. Key words and phrases. See [Che13] for a generalization of part (a) to a class of domains closely related to H.1 we can recover the following folk result: Corollary 1. To state our results we use weighted Bergman spaces Ap (H. where ω > 0 is a continuous on H.ω) = H |f | ωdV < ∞. Let BH : L2 (H) → A2 (H) denote the Bergman projection. Let 1 < p ≤ 43 . ω) if R function p p kf kLp (H. δ1 is comparable to the distance to the singular point 0. and let λ > 0 be a continuous function on (0. then BH is bounded (and surjective) from Lp (H) to Ap (H). both Theorem 1. |z2 | < |z1 | < 1 . (a) If 34 < p < 4. This is because in this range the space Ap (H. (1) for 1 < p < ∞. Note that the surjectivity of BH means that we have the best possible estimates. ω). 32A25. and a holomorphic function f ∈ O(H) belongs to Ap (H. If p ≥ 2. where dV is Lebesgue measure of C2 . 32A07. it turns out that there is absolutely no way to control the Bergman projection of an Lp function on the Hartogs triangle using a weight depending on the distance to the singularity: Theorem 1. (2) 1 On H. then BH is not bounded from Lp (H) to Ap (H). Then BH is not bounded from Lp (H) to Ap (H.1. From Theorem 1. 1]. z2 ) = |z1 | . i.1 and Corollary 1. Note that for 2 ≤ p < 4. δ1p−2 ).1 apply. we have √2 |z| ≤ δ1 (z) ≤ |z|. (b) If p ≥ 4. λ ◦ δ1 ). For such p. This pathological phenomenon does not seem to have been noticed before. and also by an Early Career internal grant from Central Michigan University to Debraj Chakrabarti.e. Let δ1 be defined on C2 by δ1 (z1 . δ1p−2 ) coincides with Ap (H) (cf. We now consider what happens under Bergman projection if 1 < p ≤ 34 .

Here we use the same method to study the mapping properties of BH in Lp spaces. in Section 3 we consider the Bergman projection on the product D∗ × D with respect to a radial weight on the first factor. This technique was used in [MM92. for providing a very congenial atmosphere to work on this problem while we were participating in the workshop Cauchy-Riemann Equations in Several Variables in June 2014. there is a ∂-closed (0. this was done in [CS13]. 1)-form ∞ (H) which is smooth up to the boundary on H. In the following section. we give a proof of Theorem 1. Consequently. one can obtain estimates which take into account the behavior of functions and forms near the singular points. estimates were obtained for the Bergman projection on certain non-smooth domains in Lp spaces with weights which vanish at the singularities of the boundary. such that no solution u of the equation g ∈ C0. w) on U × U (called the Bergman kernel). where bH cannot even be represented as a graph of a continuous function. The Hartogs triangle. A survey of mapping properties on Sobolev spaces on smoothly bounded pseudoconvex domains can be found in [BS99]. KP08.1). A standard way of understanding function theory on domains with singular boundary is the use of weights which vanish or blow up at the singular points of the boundary. We refer to [Kra01] for the general theory. one can pull back problems on H to problems on D∗ × D.4. This is similar to our Theorem 1. . The Bergman projection. It follows from the Riesz representation theorem that the Bergman projection is an Rintegral operator with the kernel BU (z.2. in the last section. 1. where estimates for the canonical solution of the ∂-problem were obtained in weighted Sobolev spaces..3. It is natural to consider the mapping properties of BU on other spaces of functions on U . 1. Bar84. For the Hartogs triangle. The orthogonal projection operator BU : L2 (U ) → A2 (U ) is the Bergman projection associated with the domain U . The closure H does not have a Stein neighborhood basis. In [EL08]. After that. For results on the regularity of the Bergman projection in Lp space on bounded pseuodoconvex domains we refer to the following articles and the references therein [LS12. and one gets weights coming from the Jacobian factor. CS13] to study function theory on H. Finally. By the Bergman inequality. While the Hartogs triangle has a non-Lipschitz boundary.1 ∂u = g lies in C ∞ (H) (see [CC91]). Zey13]. The Hartogs triangle defined in (1) has remarkable geometric and function-theoretic properties. we obtain estimates on D∗ × D using operator-theoretic methods relating to norm-convergence in Lp spaces of Taylor series. as a complex manifold it has a very simple structure: it is biholomorphic to the product D∗ × D of the punctured unit disc D∗ = { z ∈ C | 0 < |z| < 1 } and the unit disc D. We would like to thank the American Institute of Mathematics. The Bergman space A2 (U ) is defined to be the intersection L2 (U ) ∩ O(U ) of the space L2 (U ) of square integrable functions on U (with respect to the Lebesgue measure of Cn ) with the space O(U ) of holomorphic functions on U . but the Hartogs triangle is not among the domains to which the results of [EL08] apply. PS77. We recall some basic definitions and facts regarding the Bergman projection operator. This paper is organized as follows. MS94. In Section 4 we prove Theorem 1.1.2. In Proposition 3. i. for example Lp spaces or Sobolev spaces. BU f (z) = U BU (z.1. using a duality argument. A2 (U ) is a closed subspace of L2 (U ).e. ZEYTUNCU 1. The ∂-problem on H is not globally regular.2 DEBRAJ CHAKRABARTI AND YUNUS E. Acknowledgements. Consequently. BS¸12. we deduce Corollary 1. and is a classical source of counterexamples in complex analysis.e. w)f (w)dV (w) for all f ∈ L2 (U ).1 from Theorem 1. The boundary bH of the domain H has a serious singularity at the point 0.1 using a description of weighted Bergman spaces on H (see Proposition 5. Palo Alto. Let U ⊂ Cn be a bounded domain.1 using biholomorphic mapping of H with D∗ × D. i.

For f ∈ Lq (U. v ∈ L2 (H) is the standard inner product of the Hilbert space L2 (H). kf kLp (U.1. p > 1. gi| kgkLp (U ) =1 sup |hf. BU gi| .2 For a continuous function ω > 0 on a domain U . Here R and later we denote hu. We use the standard duality of Lp spaces to prove the following: Lemma 2. which if u. p1 + 1q = 1. Then if BU is bounded from Lp (U ) to Ap (U. ω). ω 1−q ) to Aq (U ). Proof. Proof of Theorem 1. the space Lp (U. then BU is also bounded from Lq (U. Let U ⊂ Cn be a bounded domain. ω) consists of functions f on U for which Z p |f |p ωdV < ∞. where q is the conjugate exponent to p. and ω > 0 a weight function on U . vi = H uvdV . so that we have kBU f kLq (U ) = = sup |hBU f. We let g ∈ Lp (U ) range over the unit sphere kgkLp (U ) = 1. since BU is self-adjoint kgkLp (U ) =1 = sup kgkLp (U ) =1 ≤ sup kgkLp (U ) =1 .BERGMAN PROJECTION ON THE HARTOGS TRIANGLE 3 2. ω (1−q) ) we estimate kBU f kLq (U ) by using duality of Lp spaces on U . i.e.ω) = U where dV is Lebesgue measure.

D E.

1 .

.

−1 .

f · ω p . (BU g) · ω p .

If 1 < p ≤ 34 . Let f be the function on H given by f (z1 .ω) · kf k   q − Lq U. it follows that BU is bounded from Lq (U. It therefore suffices to present a function f ∈ L2 (H) ∩ Lq (H. (λ ◦ δ1 )1−q ) for any choice of λ. z2 ) = (χ ◦ δ1 ) · z1 . to prove the result we only need to show that BH is not bounded from Lq (H. 1] such that χ ≡ 0 near 0. so that f ∈ L2 (H) ∩ Lq (H.ω p   since by assumption the first factor is finite.2. z1 z2 i = χ(|z1 |)z1 · z1 m z2 n dV ZH = χ(r1 )r1 e−iθ1 r1m e−imθ1 r2n e−inθ2 r1 r2 dr1 dr2 dθ1 dθ2 H Z 2π  Z 2π  Z −i(m+1)θ1 −inθ2 = e dθ1 · e dθ2 · χ(r1 )r1m+2 r2n+1 dr1 dr2 . Therefore. by H¨older’s inequality ! = sup kgkLp (U ) =1 kBU gkLp (U. so that it suffices to show that unless m = −1 and n = 0. the function f is orthogonal to each z1m z2n . ω (1−q) ) to Aq (U ). and χ ≡ 1 near 1. We have.  Proof of Theorem 1.  −1 f · ω p Lq (U )  1 · (BU g) · ω p Lp (U ) . which is continuous on H and vanishes near the singularity 0. Let χ ≥ 0 be a continuous function on [0. Note that the system of monomials {z1m z2n } for m ≥ −(n + 1) and n ≥ 0 forms a complete orthogonal set in A2 (H). then the conjugate exponent q ≥ 4. by Lemma 2. (λ ◦ δ1 )1−q ) such that BH f 6∈ Aq (H). (λ◦δ1 )1−q ) to Aq (H). and we have − pq = 1 − q.1. We claim that there is a constant C > 0 such that BH (f ) = Cz1−1 . using polar coordinates: Z m n hf. 0 0 ∆ .

This integral vanishes unless m = −1.4 DEBRAJ CHAKRABARTI AND YUNUS E. Z . r2 ) : 0 ≤ r2 < r1 < 1}. we verify that z11 does not belong to Aq (H). if q ≥ 4. We have. n = 0. To complete the proof. ZEYTUNCU R2 where ∆ ⊂ is the subset defined by {(r1 .

.

q Z .

1.

1 .

.

dV = q r1 r2 dr1 dr2 dθ1 dθ2 .

z1 .

Note that when p ≥ 2. the Hardy space on T2 .g. as N goes to infinity. We have the following analog of a result of Zhu ([Zhu91. For p ≥ 1 and for any f ∈ Ap (D2 ). z2 ) is precisely SN g(·. (5) µ=0 ν=0 i. w2 ) = N X ∞ X aµ. we conclude that there is a constant C0 . (3)  3. For almost all z2 ∈ T. z2 ) is in Lp (T). Weighted Lp -regularity of the Bergman projection on D∗ × D. (4) µ. For p ≥ 2. Corollary 4]): Lemma 3. For f ∈ O(D2 ). Ap (D∗ × D) is identical to the space Ap (D2 ). z2 ). Suppose that g ∈ H p (T2 ). For a non-negative integer N .. the Bergman projection BD∗ ×D is bounded and surjective from Lp (D∗ × D. In this section we prove the following regularity result for the Bergman projection on the bidisc. we define a map SN : O(D2 ) → O(D2 ) in the following way. From the classical Riesz theory of convergence of Fourier series in Lp (T) (see e. As in (2). z2 ) ∈ C2 | |z1 | = |z2 | = 1}.1. SN f → f in the Lp -norm.ν w1µ w2ν .ν w1µ w2ν . independent of the function g such that we have for almost all z2 ∈ T that Z 2π . w2 ) = ∞ X aµ.5]). δ12−p ) onto Ap (D∗ × D). since each holomorphic function in Lp (D∗ × D) extends holomorphically to D2 (see the proof of Lemma 5. Denote by T the unit circle {|z| = 1} in the plane and by T2 the two-torus {(z1 .1.ν=0 we define the function SN f by setting SN f (w1 . [Gra08. we set δ1 (z) = |z1 |. Proposition 3. the function z1 7→ g(z1 . H H r1  Z r1 Z 1 1−q 2 r1 r2 dr2 dr1 = 4π = 2π 2 r1 =0 r2 =0 1 r13−q dr1 . SN is the N -th partial sum in the w1 variable. Section 3.e. and note that the N -th partial sum of the Fourier series representation of g(·. This means that g admits a holomorphic extension to D2 and has boundary values in Lp (T2 ). with Taylor representation f (w1 . Proof.1 below). 0 Z which diverges if q ≥ 4.

Z 2π .

.

p .

p .

.

.

iθ .

iθ (6) .

SN g(e . z2 ).

dθ ≤ C0 .

z2 ).g(e .

0 0 . dθ.

Now Z Z |SN f (w1 .1. [Zhu91. where we think of SN as an operator from Ap (D2 ) to itself. it is sufficient to show that there is a constant C > 0 such that kSN k ≤ C for all non-negative integers N . w2 )|p dV (w1 )dV (w2 ) kSN f kpLp (D2 ) = D D  Z 2π Z 2π . Proposition 1]). to prove Lemma 3.BERGMAN PROJECTION ON THE HARTOGS TRIANGLE 5 Following a standard functional-analytic argument (cf.

Z 1Z 1 .

p .

iθ1 iθ2 .

r1 r2 = .

r2 e ).SN f (r1 e .

dθ1 dθ2 dr1 dr2 0 0 0 0  Z 2π Z 2π .

Z 1Z 1 .

p .

iθ1 iθ2 .

r1 r2 ≤ C0 .

r2 e ).f (r1 e .

ν w1µ w2ν (8) µ. the sequence {tµ }∞ µ=0 is bounded. Let f ∈ Ap (D2 ). The map T is bounded from Ap (D2 ) to itself.1. 2 and in particular. represented as in (4). we see that ∞ X k=0 k(tk+1 − tk )Sk f kLp (D2 ) ≤ C ∞ X |tk+1 − tk | k=0 < ∞. we have used (6). Proof. Now let g = limN →∞ SN T f . Such C exists since SN f → f in Lp (D2 ). (7) µ=0 PN −1 Writing tN = µ=0 (tµ+1 − tµ ) + t0 . k=0 Note that each of the factors tN and SN f in the first term converges to a limit as N → ∞ so that the first term converges to a limit in Lp (D2 ) as N → ∞. and using (7) we see that the limit limN →∞ tN exists. Let C > 0 be such that kSN f kLp (D2 ) ≤ C for all N . it follows that . We will first show that the partial sums SN T f of T f (defined using the operator SN of (5)) converge in Ap (D2 ). the new function T f (w1 .2. and therefore has a limit as N → ∞.  We now give a sufficient condition for an operator on Ap (D2 ) defined by a certain multiplier sequence on the Taylor coefficients to be bounded. By the continuity of SM on Ap (D2 ) established in the proof of Lemma 3. where the limit is in the topology of Ap (D2 ). Indeed. where in the last-but-one line. w2 ) = ∞ X tµ aµ. Therefore. Let M ≥ 0 be an integer.ν=0 is also a holomorphic function on D2 . Let {tµ }∞ µ=0 be a sequence of complex numbers such that ∞ X |tµ+1 − tµ | < ∞. Lemma 3. This shows the second term is the (N − 1)-th partial sum of an absolutely convergent series in a Banach space. dθ1 dθ2 dr1 dr2 0 0 0 0 = C0 kf kpLp (D2 ) . we can write SN T f = tN · SN f − N −1 X (tk+1 − tk )Sk f. using summation by parts. It follows that for a holomorphic f on D .

i. so that by composing the two maps. δ 2−p ) to Ap (D2 ) = Ap (D∗ × D).g.δ1 D2 D2 ) so that Lp (D∗ × D. δ12−p ) = Lp (D2 . δ12−p ) to Ap (D2 ) (note that Ap (D2 ) is identical to Ap (D∗ × D). Then U maps Ap (D2 ) to Lp (D∗ × D. |f | dV ≤ kf kLp (D2 ) = L (D . we have T f = g. w1m 1 D = δµm . for f ∈ Ap (D∗ × D).e. ZEYTUNCU SM g = SM (limN →∞ SN T f ) = limN →∞ SM SN T f = SM T f . [Zhu07. δ12−p ) is continuously embedded in Lp (D2 ). To construct U .6 DEBRAJ CHAKRABARTI AND YUNUS E. we construct an operator U : Ap (D∗ × D) → p L (D∗ × D. On D∗ × D we have 0 < δ1 < 1. FR75]). which being monotone and bounded certainly satisfies (7). we let U f = δ12 · T f.1. Indeed: Z . We first prove that BD∗ ×D is bounded from Lp (D∗ × D. the Bergman projection on D2 is bounded from Lp (D2 ) to Ap (D2 ). (9) m m hw2 . we see that BD∗ ×D is bounded from Lp (D∗ × D.) Recall that the Bergman projection on the unit disc is bounded from Lp (D) to Ap (D) (see e. by a simple application of Fubini’s theorem it follows that the Bergman projection on the bidisc D2 is bounded from Lp (D2 ) to Ap (D2 ). δ12−p ). and therefore for p ≥ 2. Z Z p p |f |p δ12−p dV = kf kp p 2 2−p . To show that BD∗ ×D surjective onto Ap (D∗ × D). we obtain. Using the fact that the Bergman kernel of the product D2 is the tensor product of the Bergman kernels of the factors. a computation shows that E D |w1 |2 w1µ . w2 iD tµ Defining T as in (8).  Proof of Proposition 3. δ 2−p ) such that BD∗ ×D ◦ U is the identity on Ap (D∗ × D). the function U f is projected by BD∗ ×D to f . But as noted above. Since this is true for each M . which shows the surjectivity of BD∗ ×D . in the equation (8) we let tµ = 1 + (µ + 1)−1 . the operator T maps Ap (D2 ) to itself. Since g ∈ Ap (D2 ). Further.

2 .

p 2−p p .

δ1 T f .

where in the last line we have used Lemma 3.δ1 ) D∗ ×D Z = δ12+2p |T f |p dV ∗ ZD ×D ≤ |T f |p dV ≤ D∗ ×D C kf kpLp (D2 ) < ∞. w2n iD 1 2 . w2 iD m. since {w1m w2n }∞ m.ν |w1 |2 w1µ w2ν .ν=0 = ∞ ∞ X X m. w2 ) = |w1 |2 tµ aµ.ν w1µ w2ν .ν=0 D tµ aµ.ν=0 ∞ X D∗ ×D E D tµ aµ. w1m iD · hw2n .n=0 = ∞ X µ. δ dV kU f k p ∗ 2−p = 1 L (D ×D. w1m hw1m .n=0 µ.n=0 µ.2. w1m iD D∗ ×D · w1m w2n hw1m .ν |w1 |2 w1µ . w2n iD m n w w hw2n . we see that * + ∞ ∞ X X wm wn BD∗ ×D (U f )(w1 .n=0 is an orthogonal set 2 2 in L (D ). w1m w2n · m m 1 2 n n hw1 . w1 iD · hw2 . Further. w1m w2n m. w2n iD E D · hw2ν .

and let Φ : U → V be a biholomorphic map. w2 ).n=0 7  4. Let U and V be two bounded domains in Cn .n w1m w2n = f (w1 .ν · δµm δνn w1m w2n . Change of Variables. 4.ν=0 ∞ X = am.BERGMAN PROJECTION ON THE HARTOGS TRIANGLE ∞ ∞ X X 1 = tµ aµ. m.1. We set .1. using (9) tµ m. Proof of Theorem 1.n=0 µ.

.

.

.

u = .

det Φ0 .

and v = .

(det Φ0 ) ◦ Φ−1 .

v α ) → Lp U.2. (2) BV is bounded and surjective from Lp (V ) to Ap (V. u2−p+α  (11) is an isometric isomorphism of Banach spaces. We prove the following: Lemma 4. Let α be a real number.1. The following are equivalent: (1) BU is bounded and surjective from Lp (U. u2−p ) to Ap (U. The map TΦ : Lp (V. Indeed. For a function h on V define TΦ h = det(Φ0 ) · (h ◦ Φ). (10) Let α be a real number. v α ). kTΦ f kpLp (U. We first note the following Lemma 4.u2−p+α ) Z = ZU = ZU = . Proof. u2−p+α ). .

.

.

.

.

(det Φ0 ) · (f ◦ Φ).

p .

det Φ0 .

2−p+α dV .

.

α .

.

2 |f ◦ Φ|p .

det Φ0 .

.

det Φ0 .

dV .

.

α |f |p .

det Φ0 ◦ Φ−1 .

v α ). (12) So that we have BV = TΦ−1 ◦ BU ◦ TΦ .  . BU is bounded from Lp (U. which can be written as BU ◦ TΦ = TΦ ◦ BV . The part 2 =⇒ 1 can be done exactly the same way. by hypothesis. so that we have (2). V where in the last line we use the change of variables formula. Recall the Bell transformation formula for the Bergman projection (cf. u2−p ) to Ap (U. u2−p+α ) to Ap (V.1. u2−p ). and we also know that TΦ−1 is an isometry from Ap (U.  Proof of Lemma 4. u2−p+α ). dV. Start with 1 =⇒ 2. [Bel81]). We know that TΦ is an isometry from Lp (V ) to Lp (U.

1. (15) 0 D so that δ α ∈ L1 (D). Lemma 5. Since −2 < α ≤ 0. We apply Lemma 4. Then we have Ap (D∗ . We first claim that each function in Ap (D∗ .1. (2) BD∗ ×D is bounded and surjective from Lp (D∗ × D.1 shows that the following statements are equivalent: (1) BH is bounded and surjective from Lp (H) to Ap (H.1.1 5. On the punctured unit disc D∗ = { z ∈ C | 0 < |z| < 1 } .1.1. ZEYTUNCU 4. (13) so that in Lemma 4. the result follows for p ≥ 2. w1 w2 ). we have u(w1 . Now . Let p ≥ 2 and −2 < α ≤ 0. δ α ) = Ap (D). Let Φ : D∗ × D → H be given by Φ(w1 . (14) and the two norms k·kLp (D) and k·kLp (D. Lemma 4. δ α ) has a removable singularity at 0. Weighted Bergman spaces on the punctured disc. z2 ) = |z1 | . Proof of Corollary 1. and we take α = p − 2.8 DEBRAJ CHAKRABARTI AND YUNUS E.δα ) on this space are equivalent.2. 5. Since the second statement was proved in Proposition 3. we have Z Z 1 |w|α dV (w) = 2π r1+α dr < ∞. so that to establish the claim it is sufficient to show that each function in A2 (D∗ ) extends holomorphically to 0. δ12−p ) to Ap (D∗ × D). δ1p−2 ).1. w2 ) = (w1 . It follows by H¨ older’s inequality that Ap (D∗ ) ⊂ A2 (D∗ ). Proof. Proof of Theorem 1. w2 ) = |w1 | and v(z1 . denote by δ the weight function δ(z) = |z|.

R 2π .

P P 2 2ν ν .

f (reiθ ).

we have 0 f ∈ A2 (D∗ ) we have Z 1 Z 2π . If ν=−∞ aν w .2 dθ = 2π ∞ if f ∈ O(D∗ ). writing f (w) = ∞ ν=−∞ |aν | r .

 .

2  .

2 iθ .

kf kA2 (D∗ ) = .

f re .

i. Since this is finite. ν ≥ 0.e. We claim that the injective map in (16) is in fact surjective. . Since every element of Ap (D∗ ) has a removable singularity at 0. we have δ α ≥ 1 on D. δ α ) . so we have a continuous embedding Ap (D∗ .→ Ap (D∗ ) = Ap (D). we must have aν = 0 unless 2ν+1 > −1. dθ rdr 0 = ∞ X 0 |aν | 2 1 Z r2ν+1 dr. it follows that Ap (D∗ ) = Ap (D).δα ) = |f (w)|p |w|α dV (w) ZD Z = |f (w)|p |w|α dV (w) + |f (w)|p |w|α dV (w) |w|< 21 Z ≤C |w|α dV (w) + |w|≥ 12 1 2α Z |w|≥ 12 |f (w)|p dV (w). 0 ν=−∞ by the monotone convergence theorem. Let f ∈ We have Z kf kpLp (D∗ . (16) Ap (D). which verifies the claim that the singularity of f at 0 is removable. Now since α ≤ 0..

δ1p−2 ) H |z1 | Z 1 Z r1 1 2 = 4π 2 r1 r2 dr1 dr2 r1 =0 r2 =0 r1 = 2π 2 . we see from Lemma 4. so that the result follows. Irregularity of the Bergman projection on a smooth bounded domain in C2 . 48(1):167–175. [Bel81] Steven R. . Consequently to prove part (b) of Corollary 1.1: Proposition 5. TΦ : Ap (H. and the two norms k·kLp (H) and k·kLp (H. But we already saw that Ap (D∗ × D. δ1p−2 ) is the same as Ap (H) as a topological vector space. Applying Lemma 2. Noting that Tφ maps holomorphic functions to holomorphic functions. (17) Let the map Φ : D∗ × D → H be given by (13). then −2 < 2 − p ≤ 0. the space Ap (H. if 2 ≤ p < 4. δ1p−2 ) → Ap (D∗ × D) is also an isometric isomorphism. 119(2):431–436. 1981. Letting α = 0. and letting α = p − 2. Z 1 p p−2 kf k p = dV p |z1 | L (H. δ12−p ) and Ap (D∗ × D) are the same space of functions on D∗ × D. Proper holomorphic mappings and the Bergman projection. so that we have Ap (D∗ × D. and let TΦ be as in (10).1. References [Bar84] David E. Ann. On the other hand. Proof of Corollary 1. we combine Theorem 1. For part (a).1.  We deduce the following from Lemma 5.1. δ1p−2 ). δ1p−2 ) and by the latter. Bell.2. we obtain that BH is bounded (and hence surjective) from Lp (H) to Ap (H) when 34 < p < 2. we see that TΦ : Ap (H) → Ap (D∗ × D.2 that for each real α.1 with Proposition 5. we see that Ap (D∗ × D. Such a function is f (z1 .1. for p ≥ 4. The computation leading to (2) shows that f 6∈ Ap (H) if p ≥ 4.. Then by an application of Fubini’s theorem to the bidisc and Lemma 5. If 2 ≤ p < 4 then Ap (H. Therefore the map in (16) is a continuous bijection and the result follows by the open mapping theorem.BERGMAN PROJECTION ON THE HARTOGS TRIANGLE 9 The first of these integrals is finite (see (15)) and so is the second one since by hypothesis f ∈ Lp (D). Barrett. z2 ) = z11 . We showed in Section 3 above that BH is bounded and surjective from Lp (H) onto Ap (H. therefore the result follows for 2 ≤ p < 4. δ1α ) = Ap (D2 ). by a direct computation. δ12−p ) is an isometric isomorphism. we see that TΦ : Ap (H. there is a function f ∈ Ap (H. Duke Math. δ12−p ) = Ap (D2 ). Let δ1 be as in (2). δ1p−2 ) which is not in Ap (H). By the former.  5. of Math. δ1p−2 ) = Ap (H). provided p ≥ 2 and −2 < α ≤ 0.1 with w ≡ 1. The result follows. (2). if 2 ≤ p < 4. δ1α ) → Ap (D∗ × D. and the two Banach-space norms are equivalent. the operator BH is bounded from Lp (H) to Ap (H. δ12−p+α ) is an isometric isomorphism of Banach spaces. 1 Proof. In particular. J.δα ) on this space are equivalent.1 it suffices to show that. 1984. with equivalence of norms.

M. Geom.7898. Liwei Chen. 2013. Lp regularity of weighted Bergman projections. Duality of Bloch spaces and norm convergence of Taylor series. Michigan Math. Operator theory in function spaces. Illinois J. Analysis and geometry on worm domains.. H. Duke Math... R´egularit´e h¨ old´erienne de l’op´erateur ∂ sur le triangle de Hartogs. 2001. second edition. 2013. Krantz. Dearborn. Pleasant. 44(3):695–704. MI 48859 E-mail address: chakr2d@cmich. M. volume 138 of Mathematical Surveys and Monographs. Mapping properties of the Bergman projection on convex domains of finite type. Ann. New York. Inst. Sci. Press. 365(6):2959– 2976.. Projections on spaces of holomorphic functions in balls. Straube. MI 48128 E-mail address: zeytuncu@umich. 2012. Math. Yunus E. 2007. The Lp boundedness of the Bergman projection for a class of bounded Hartogs domains. Frank Forelli and Walter Rudin. C k+a -estimates for the ∂-equation on the Hartogs triangle. 1974/75. 2012. Math.. second edition. J. Math. Indiana Univ. J. Department of Mathematics. Publ.. 73(1):177–199. 1991. Boas and Emil J. David Barrett and S¨ onmez S ¸ ahuto˘ glu. D.. D.. 356(1):241–258. Peloso. Anal. Zeytuncu) University of Michigan . Res. J. volume 37 of Math. Nachr. J. arXiv:1304. Function theory of several complex variables. 61(1):187–198. Math.. RI. J. Chollet. Zeytuncu. Chaumat and A. 281(7):916–929. Springer. Providence. Inst. In Several complex variables (Berkeley. 1991. Steven G. J. ZEYTUNCU Harold P. 294(4):661–675. Estimates for the Bergman and Szeg¨ o projections on strongly pseudo-convex domains. Math. Steven G.10 [BS99] [BS ¸ 12] [CC91] [Che13] [CS13] [EL08] [FR75] [Gra08] [KP08] [Kra01] [LS12] [MM92] [MS94] [PS77] [Zey13] [Zhu91] [Zhu07] DEBRAJ CHAKRABARTI AND YUNUS E. CA. 1977. Ann. Sobolev regularity of the ∂-equation on the Hartogs triangle. volume 249 of Graduate Texts in Mathematics. Lp -estimates for the Bergman projection on strictly pseudoconvex nonsmooth domains. 18(2):478–510. Mt. 1999.-M. Phong and E...edu .edu (Yunus E. Stein. Loredana Lanzani and Elias M. Ke He Zhu. 56(1):127–154 (2013). 1992. Krantz and Marco M. Duke Math. Michigan Math. pages 79–111. 1995–1996). Reprint of the 1992 edition. Lan Ma and Joachim Michel. Providence. Cambridge Univ. Loukas Grafakos. Classical Fourier analysis. Global regularity of the ∂-Neumann problem: a survey of the L2 Sobolev theory. 38(1):89– 101. 2008. Debraj Chakrabarti and Mei-Chi Shaw.Dearborn. Math. Cambridge. McNeal and E. Irregularity of the Bergman projection on worm domains in Cn . Department of Mathematics and Statistics. Trans. (Debraj Chakrabarti) Central Michigan University. AMS Chelsea Publishing. Kehe Zhu. The Bergman projection in Lp for domains with minimal smoothness. Soc. Ann.. 24:593–602. Stein. J. 1994. Dariush Ehsani and Ingo Lieb. Stein. J. 2013. Amer. American Mathematical Society. 2008. Fourier (Grenoble). 41(4):867–882. 2008. RI.