0% found this document useful (0 votes)
13 views13 pages

A Proof of Lieb-Wehrl Entropy Conjecture For: Abstract

This paper confirms the Lieb-Wehrl entropy conjecture for the group SU(N, 1) by establishing sharp functional inequalities for coherent state transforms. The authors extend previous results related to the Faber-Krahn inequality within the context of Bergman spaces. Key theorems are presented, demonstrating the maximization of certain integrals involving convex functions and coherent states, and the paper is structured to explore these results systematically.

Uploaded by

gliosca.diego
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views13 pages

A Proof of Lieb-Wehrl Entropy Conjecture For: Abstract

This paper confirms the Lieb-Wehrl entropy conjecture for the group SU(N, 1) by establishing sharp functional inequalities for coherent state transforms. The authors extend previous results related to the Faber-Krahn inequality within the context of Bergman spaces. Key theorems are presented, demonstrating the maximization of certain integrals involving convex functions and coherent states, and the paper is structured to explore these results systematically.

Uploaded by

gliosca.diego
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

A PROOF OF LIEB–WEHRL ENTROPY CONJECTURE FOR SU (N, 1)

PRATYOOSH KUMAR AND MANDEEP SINGH

Abstract. We investigate the sharp functional inequalities for the coherent state transforms of
SU (N, 1). These inequalities are rooted in Wehrl’s definition of semiclassical entropy and his con-
jecture about its minimum value. Lieb resolved this conjecture in 1978, posing a similar question
arXiv:2501.10061v1 [math-ph] 17 Jan 2025

for Bloch coherent states of SU (2). The SU (2) conjecture was settled by Lieb and Solovej in 2014,
and the conjecture was extended for a wide class of Lie groups. The generalized Lieb conjecture
has been resolved for several Lie groups, including SU (N ), N ≥ 2, SU (1, 1), and its AX + B sub-
group. With sharp functional inequalities for the coherent state transforms of the group SU (N, 1),
we confirm this Lieb-Wehrl entropy conjecture for SU (N, 1), N ≥ 2. Additionally, we explore the
Faber-Krahn inequality, which applies to the short-time Fourier transform with a Gaussian window.
This inequality was previously proven by Nicola and Tilli and later extended by Ramos and Tilli
to the wavelet transform. In this paper, we further extend this result within the framework of the
Bergman space Aα .

1. Introduction
We begin with sharp functional inequalities for coherent state transforms and their connection
with Lieb -Wehrl entropy conjecture. To establish the connection, we consider the normalized
Gaussian functions ψz ∈ L2 (R) parametrized by z = (p, q) ∈ R2 given by
−(x−q)2
1
+ ipx
ψz (x) = (πℏ)− 4 e 2ℏ ℏ , for all x ∈ R,
where ℏ > 0 is a fixed constant. These vectors are (Schrödinger, Klauder, Glauber) coherent states
first considered by Schrödinger and these are related to irreducible unitary representations of the
Heisenberg-Weyl group. The analogues of these vectors for other Lie groups have been studied in
the literature (see e.g. [14, 13]). For a nonnegative operator ρ on L2 (R) with tr ρ = 1, the transform
hρ (z) = ⟨ψz , ρψz ⟩,
is known as the coherent state transforms, the Husimi function, or the covariant symbol. Corre-
sponding to ρ, Wehrl in [19] defined the classical entropy by
Z
− hρ (z) log hρ (z)dz,
R2
where he proved that it is positive, and conjectured that its minimum value occurs when ρ is an
projection operator onto any coherent state, that is when ρ = |ψz0 ⟩⟨ψz0 |, for some z0 ∈ R2 . Lieb
proved the conjecture shortly thereafter in [9], where he proved more generally that for s ≥ 1, the
quantity Z
− (hρ (z))s dz,
R2

2020 Mathematics Subject Classification. Primary 39B62, 22E45, Secondary 30H20, 81R30.
Key words and phrases. Coherent states, Lieb-Wehrl entropy conjecture, Faber-Krahn inequality.
Second Author is supported by Institute fellowships of IIT Guwahati.
1
2

is maximized when ρ = |ψz0 ⟩⟨ψz0 |, for some z0 ∈ R2 . Since the above quantity is independent
of ρ for s = 1. As a result, Wehrl’s entropy conjecture follows by differentiating at s = 1. Later,
Carlen gave an alternate proof of Lieb’s result based on logarithmic Sobolev inequality in [4], also he
characterized the cases of equality. Another proof is also given in [11]. Lieb and Solovej generalized
above result in [9] and proved that for any convex function Φ : [0, 1] → R with Φ(0) = 0, the
integral
Z
− Φ(hρ (z))dz,
R2

is maximized when ρ = |ψz0 ⟩⟨ψz0 |, for some z0 ∈ R2 . And they called it generalized Wehrl conjecture.
In [8], Lieb observed that the analogue of Wehrl’s entropy conjecture should hold if we replace
Scrödinger coherent states by Bloch coherent states, which are related to the irreducible representa-
tions of SU (2). This conjecture was resolved in [9], although some special cases were already known
in [3, 17]. Lieb and Solovej used limting argument in [9] and noted that this approach does not work
for SU (N ), N > 2. Later, in [10], they proved the conjecture for all symmetric representations of
SU (N ) and conjectured that the analogue of Wehrl’s conjecture should hold, at least for a wide
class of Lie groups.
In recent work [6], Lieb and Solovej considered the group SU (1, 1) and its subgroup AX +B where
they formulated the Wehrl-type entropy conjecture into a problem about containment of Bergman
spaces on D. Kulikov proved this containment of Bergman spaces on D in [6], hence confirming
Wehrl’s conjecture for SU (1, 1). He also characterized the cases of equality when Φ is strictly
convex. Some particular cases for SU (1, 1) were already considered in [1, 2]. Inspired by Kulikov’s
technique, Frank applied a uniform approach to prove Wehrl’s conjecture for the Heisenberg-Weyl
group, SU (2), and SU (1, 1), and characterized the cases of equality when Φ is not affine linear (see
[5] ).
One of the main results of this paper is the following theorem, where we use the containment of
Bergman spaces on BN established by Li and Su in [7]. In this theorem, we generalize [5, Theorem
6] to higher dimensions and confirm the Wehrl conjecture for the group SU (N, 1).

Theorem A. Let k ∈ Z+ and consider the irreducible discrete series representation of SU (N, 1)
on H corresponding to k. Let Φ : [0, 1] → R be convex. Then,
Z 
2
sup Φ(|⟨ϕz , ϕ⟩| )dm(z) : ϕ ∈ H, ∥ϕ∥H = 1
BN
Z 1
N N N −1
= Φ(s)s− (N +1)k −1 1 − s1/(N +1)k ds,
(N + 1)k 0

and the supremum is attained for ϕ = eiθ ϕz0 for some z0 ∈ BN , θ ∈ R. If Φ is strictly convex and
if the supremum is finite, then it is attained only for such ϕ.

Note that if Φ(t) = ts for s ≥ 1, then the analogue of Wehrl’s original entropy conjecture follows
from Theorem A by taking minus a derivative at s = 1 as mentioned above. One consequence of
the above theorem is the following result, which generalizes [5, Corollary 7].
3

Theorem B. Let k ∈ Z+ and consider the irreducible discrete series representation of SU (N, 1)
on H corresponding to k. Let Φ : [0, 1] → R be convex. Then,
Z 
sup Φ(⟨ϕz , ρϕz ⟩)dm(z) : ρ ≥ 0 on H, tr ρ = 1
BN
Z 1
N N N −1
= Φ(s)s− (N +1)k −1 1 − s1/(N +1)k ds,
(N + 1)k 0

and the supremum is attained for ρ = |eiθ ϕz0 ⟩⟨ eiθ ϕz0 | for some z0 ∈ BN , θ ∈ R. If Φ is strictly
convex and if the supremum is finite, then it is attained only for such ρ.
In this paper, we also discuss another important topic, the Faber–Krahn inequality. Given f ∈
L2 (R), the Short-time Fourier transform (STFT), defined as
Z
νf (x, ω) = e−2πiyω f (y)ϕ(x − y)dy, x, ω ∈ R
R
1/4 −πx2
where ϕ(x) = 2 e is the Gaussian window. The STFT is also called Bargmann transform
in Analysis and the Coherent state transform in quantum mechanics. We refer to [12] and the
references provided for more details on this topic. The Faber–Krahn inequality for STFT, as given
in [12, equation 1.3], can be conveniently restated in terms of functions in the Fock space on C. In
[12], Nicola and Tilli prove the following result for the Fock space
Theorem 1.1. [12, Theorem 3.1] For every F ∈ F 2 (C) \ {0} and every measurable set E ⊂ R2 of
finite measure, we have
2
|F (z)|2 e−π|z| dz
R
E
≤ 1 − e−|E| .
∥F ∥2F 2
Moreover, equality occurs (for some F and for some E such that 0 < |E| < ∞) if and only if
F = cFz0 (for some z0 ∈ C and some nonzero c ∈ C) and E is equivalent, up to a set of measure
zero, to a ball centered at z0 .
Inspired by proof of the Faber-Krahn inequality for STFT in the work mentioned above, Ramos
and Tilli extended the Faber-Krahn inequality for Wavelet transform [15]. They further reduce this
problem to an optimization problem on Bergman spaces. Now, we state the second main result of
this paper. The following can be considered as an extension of the above result and a version of
Faber-Krahn inequality for Bergman spaces Aα on BN .
Theorem C. Let α > N and s > 0 be fixed. Then, for every ϕ ∈ Aα with ∥ϕ∥Aα = 1 and every
measurable set E ⊂ BN such that m(E) = s, we have
Z Z
2 α
2

1 − |z|2 dm(z),

|ϕ(z)| 1 − |z| dm(z) ≤ (1.1)
E Bs
where Bs is the ball centered at the origin such that m(Bs ) = s. Moreover, there is equality in (1.1)
if and only if ϕ ≡ eiθ ϕz0 for some z0 ∈ BN , θ ∈ R, and E is equivalent (up to measure 0) to a ball
centered at z0 , such that m(E) = s.
The paper is organized as follows. In the next section, we introduce some preliminary notation,
define Bergman spaces over BN , and list some of their properties. In Section 3, we start with the
group SU (N, 1) structure, its representation, and corresponding coherent states. Then, we proceed
to prove Theorem A and Theorem B. Finally, Section 4 includes a detailed discussion of the Feber-
Krahn inequality and its extension in the Bergman space framework, where we prove Theorem
C.
4

2. Notation and Preliminaries


Most of our notation and results are standard and can be found in [20, 16, 13]. The symbols
Z+ , R, and C will respectively denote the set of all nonnegative integers, real numbers, and complex
numbers. Throughout the paper, we fix a positive integer N . For given z = (z1 , . . . , zN ) and
, w = (w1 , . . . , wN ) in CN , we define
⟨z, w⟩ = z1 w1 + · · · + zN wN ,
where wk is the complex conjugate of wk . We also write
p p
|z| = ⟨z, z⟩ = |z1 |2 + · · · + |zN |2 .
The open unit ball in CN is the set
BN = {z ∈ CN : |z| < 1}.
The boundary of BN will be SN = {z ∈ CN : |z| = 1} the unit sphere in CN . For z ∈ BN the
Möbius invariant measure on BN is given by
dv(z)
dm(z) = ,
(1 − |z|2 )N +1
where dv denotes the volume measure on BN normalized so that v(BN ) = 1. Now we define the
appropriate Bergman spaces.
Definition 2.1. For every α > N the Bergman space Aα is the Hilbert space of all holomorphic
functions ϕ : BN → C such that
Z
2
∥ϕ∥Aα = cα |ϕ(z)|2 (1 − |z|2 )α dm(z) < +∞,
BN

where cα is a normalizing constant so that ϕ(z) ≡ 1 have Bergman norm 1.


Using polar coordinates, the value of cα is given by
Γ(α)
cα = .
N !Γ(α − N )
The inner product on these spaces is given by
Z
⟨ϕ, ψ⟩ = cα ϕ(z)ψ(z)(1 − |z|2 )α dm(z).
BN

Given w ∈ BN , the reproducing kernel at w is the unique function Kw ∈ Aα such that for all
f ∈ Aα , f (w) = ⟨f, Kw ⟩. This kernel is given by
Kw (z) = (1 − ⟨z, w⟩)−α .
For any point z0 ∈ BN \{0} we define biholomorphic mappings of BN onto itself by
 
2 1/2
z0 − ⟨z,z 0⟩
|z0 |2 0
z − (1 − |z0 | ) z − ⟨z,z0 ⟩
z
|z0 |2 0
Υz0 (z) = .
1 − ⟨z, z0 ⟩
These mappings induce subjective isometries which are also involutive given as
α/2
1 − |z0 |2

(Tz0 ϕ) (z) = ϕ (Υz0 (z))
(1 − ⟨z, z0 ⟩)2
5

(see [18]), it turns out that the functions


α/2
1 − |z0 |2

ϕz0 (z) = (Tz0 1) (z) = ,
(1 − ⟨z, z0 ⟩)2
have Bergman norm 1 for all admissible values of α. Throughout the paper, for any ϕ ∈ Aα , we set
the function α
uϕ (z) := |ϕ(z)|2 1 − |z|2 , (2.1)
and its superlevel sets
Ωϕ,t := {z ∈ BN : uϕ (z) > t} , for t ≥ 0. (2.2)
The distribution function of uϕ is given by
µϕ (t) := m(Ωϕ,t ), for 0 ≤ t ≤ max uϕ , (2.3)
BN

and its decreasing rearrangement is defined by


u∗ϕ (s) := sup{t ≥ 0 : µϕ (t) > s} for s ≥ 0. (2.4)
We note that u∗ϕ is the inverse function of µϕ , i.e., u∗ϕ (s) = µ−1
ϕ (s), for s ≥ 0, which maps [0, +∞)
decreasingly and continuously onto (0, max uϕ ].
For any ϕ ∈ Aα , the sharp estimate for uϕ (z) in term of ∥ϕ∥Aα is given in the following proposition.
Proposition 2.2. [18] For any ϕ ∈ Aα and for every z ∈ BN
uϕ (z) ≤ ∥ϕ∥2Aα . (2.5)
The equality in (2.5) occurs at some point z0 ∈ BN if and only if ϕ = cϕz0 for some c ∈ C with
|c| = ∥ϕ∥Aα .

3. Proofs of Theorem A and Theorem B


Before the proofs of Theorem A and Theorem B, we recall some facts about the group SU (N, 1)
and their irreducible discrete series unitary representation (for details, we refer to [14, 13]). The
group SU (N, 1) is the group of (N + 1) × (N + 1) complex matrices of determinant 1 that leave the
form |z1 |2 + · · · + |zN |2 − |zN +1 |2 invariant. If g ∈ SU (N, 1) then g can be written as
 
A B
g=
C D
where A, B, C, D are N × N, N × 1, 1 × N, and 1 × 1 matrices, respectively, satisfying
AA+ − BB + = IN , AC + = BD+ , DD+ − CC + = 1,
here IN denote the N × N identity matrix and Z + denotes the transposed conjugate of a matrix
Z. The group SU (N, 1) acts transitively on BN by
z → zg = (A′ z + C ′ )(B ′ z + D′ )−1 ,
and the coherent states of the group will be parametrized by points on BN . For a positive integer
k, the group has irreducible discrete series unitary representation which can be realized on A(N +1)k
by
−(N +1)k
U k (g)ϕ(z) = [det(B ′ z + D′ )] ϕ(zg)
Coherent states corresponding to the lowest weight vector ϕ0 ≡ 1 are given by
−(N +1)k
ϕg (z) = U k (g)ϕ0 (z) = [det(B ′ z + D′ )] ,
6

and ϕg1 = eiθ ϕg2 if and only if g1 = g2 h, where θ ∈ R and h belong to the isotropy subgroup of
ϕ0 . Therefore, as mentioned before, every coherent state is now determined by a point w ∈ BN ,
ϕg (z) = eiθ ϕw (z), where
(N +1)k/2
(1 − |w|2 )

ϕw (z) = .
(1 − ⟨z, w⟩)2
Now we are ready to prove Theorem A.
Proof of Theorem A. Let k be positive integer and ϕ ∈ A(N +1)k . From the above discussion we have
⟨ϕz , ϕ⟩ = (1 − |z|2 )(N +1)k/2 f (z),
where Z
f (z) = c(N +1)k (1 − ⟨z, w⟩)−(N +1)k ϕ(w)(1 − |w|2 )(N +1)k dm(w).
BN
−(N +1)k
Since (1−⟨z, w⟩) is the reproducing kernel of A(N +1)k and ϕ ∈ A(N +1)k , therefore f (z) = ϕ(z)
for all z ∈ BN . Hence,
|⟨ϕz , ϕ⟩|2 = (1 − |z|2 )(N +1)k |ϕ(z)|2 = uϕ (z). (3.1)
Now in view of [7, Theorem 1.6], for ϕ ∈ A(N +1)k with ∥ϕ∥A(N +1)k = 1 we have the following
inequality: Z Z
2
Φ(|⟨ϕz , ϕ⟩| )dm(z) ≤ Φ(|⟨ϕz , 1⟩|2 )dm(z),
BN BN
and the supremum on the left-hand side is attained for ϕ ≡ 1. Since for every t > 0 the measure
m({uϕz0 (z) > t}) is independent of z0 , therefore the supremum on the left-hand side is attained for
ϕ = eiθ ϕz0 , for some z0 ∈ BN , θ ∈ R. A simple calculation by taking ϕ ≡ 1 shows that the explicit
value of the supremum is
 
Z   Z 1 Φ (1 − r2 )(N +1)k r2N −1
(N +1)k
Φ 1 − |z|2 dm(z) = 2N dr
BN 0 (1 − r2 )N +1
Z 1
N N N −1
= Φ(s)s− (N +1)k −1 1 − s1/(N +1)k ds.
(N + 1)k 0
To prove the second part of the theorem, we assume the supremum is finite and the given function
Φ is strictly convex. This implies that the derivative Φ′ is a strictly increasing function. Now, if
the supremum is attained for some ϕ, then we have
Z   Z 1 Z 1
2 (N +1)k) ′
2
µ1 (t)Φ′ (t)dt.

Φ |ϕ(z)| 1 − |z| dm(z) = µϕ (t)Φ (t)dt = (3.2)
BN 0 0

Also, from the norm condition ∥ϕ∥A(N +1)k = 1, we have


Z 1 Z 1
µϕ (t)dt = µ1 (t)dt. (3.3)
0 0

Choose a ∈ [0, 1] such that µϕ (t) ≥ µ1 (t) for t ≤ a and µϕ (t) ≤ µ1 (t) for t ≥ a. Since the function
1 1
g(t) = t (N +1)k (µϕN (t) + 1) is decreasing [7, Theorem 3.1] thus such an a always exist. Hence, the
function
h(t) := (Φ′ (t) − Φ′ (a)) (µϕ (t) − µ1 (t)) ≤ 0 for all t ∈ [0, 1]. (3.4)
7

Now, multiplying with −Φ′ (a) on both sides of the equation (3.3) and using (3.2), we obtain
Z 1 Z 1

−Φ (a) (µϕ (t) − µ1 (t)) dt + Φ′ (t) (µϕ (t) − µ1 (t)) dt = 0.
0 0
This implies,
Z 1 Z 1
h(t)dt = (Φ′ (t) − Φ′ (a)) (µϕ (t) − µ1 (t)) dt = 0.
0 0
Now from (3.4), we have h(t) ≡ 0. Since Φ′ is strictly-increasing, therefore µϕ (t) = µ1 (t) for all
t ∈ [0, 1], which means ∥uϕ ∥∞ = 1. Now, by appealing to Proposition 2.2, we have ϕ = eiθ ϕz0 for
some z0 ∈ BN , θ ∈ R. □
Now we will prove Theorem B, which is a consequence of Theorem A.
Proof of Theorem B. For a nonnegative operator ρ on A(N +1)k with tr ρ = 1 we have
X X
ρ= λi |ψi ⟩⟨ψi | with λi = 1, λi ≥ 0, ⟨ψi , ψj ⟩ = δi,j .
i i

Therefore
! * +
X X X
⟨ϕz , ρϕz ⟩ = ⟨ϕz , λi |ψi ⟩⟨ψi | ϕz ⟩ = ϕz , λi ⟨ψi , ϕz ⟩ψi = λi |⟨ψi , ϕz ⟩|2
i i i

Now, by using (3.1) we have


X
⟨ϕz , ρϕz ⟩ = λi uψi (z)
i
Since the function Φ is a convex function thus for any z ∈ BN we have
!
X X
Φ(⟨ϕz , ρϕz ⟩) = Φ λi uψi (z) ≤ λi Φ (uψi (z)) . (3.5)
i i
R
If we denote by S(may be ∞), the supremum in Theorem A, then BN Φ (uψi (z)) dm(z) ≤ S and
Z X Z X
Φ(⟨ϕz , ρϕz ⟩)dm(z) ≤ λi Φ (uψi (z)) dm(z) ≤ λi S = S. (3.6)
BN i BN i
iθ iθ 2
Note that if ρ = |e ϕz0 ⟩⟨ e ϕz0 | then ⟨ϕz , ρϕz ⟩ = |⟨ϕz0 , ϕz ⟩| = uϕz0 (z). Now, by Theorem A the
supremum will be achieved for ρ = |eiθ ϕz0 ⟩⟨ eiθ ϕz0 | for some z0 ∈ BN , θ ∈ R.
Now, we discuss the equality case. If S < ∞ and it is attained for some ρ, then there is equality
everywhere in (3.5) and (3.6). Furthermore, if Φ is strictly convex, then from the last inequality
in (3.5), for every i we have uψi ≡ uψ1 and hence, ψi (z) = ψ1 (z) for every i and z ∈ BN . Since
⟨ψi , ψ1 ⟩ = δi,1 , therefore ψi ≡ 0 for all i ̸= 1. We conclude that ψ1 = eiθ ϕz0 for some z0 ∈ BN , θ ∈ R
and hence ρ = |eiθ ϕz0 ⟩⟨ eiθ ϕz0 |. □

4. Faber-Krahn type inequality


In this section, we establish the Faber-Krahn inequality and prove Theorem C in the setting of
Hilbert spaces, namely Aα . We begin with some preliminary lemmas that will be needed in the
proof of Theorem C. We state our first lemma, which is an appropriate adaptation of the proof of
Lemma 3.2 in [12]. This lemma links the derivative of the distribution function and the derivative
of the rearrangement function with the integral over the boundary of the superlevel sets. The
8

expression of a derivative of the distribution function can also be found in the proof of Theorem
3.1 in [7].
Lemma 4.1. The function µϕ (t) is absolutely continuous on (0, max uϕ ], and
Z −1

−µϕ (t) = ∇u
e ϕ dσg . (4.1)
{uϕ =t} g

Similarly, the function u∗ϕ is absolutely continuous on [0, +∞), with


Z !−1
−1
−(u∗ )′ (s) =
ϕ ∇u
e ϕ dσg . (4.2)
{uϕ =u∗ϕ (s)} g

As Kulikov pointed out in [6], the above lemma has an interesting geometric interpretation in
the sense that when t increases by a small number ϵ, then the superlevel set Ωuϕ ,t−ϵ expands in

the direction orthogonal to ∂Ωuϕ ,t−ϵ by the value proportional to ϵ |∇u
e ϕ |g . The next result is the
following lemma which gives the expression for the derivative of the integral function of uϕ over its
superlevel sets and follows as Lemma 3.4 in [12] by obvious modifications.
Lemma 4.2. The function
Z
Iϕ (s) = uϕ (z)dm(z), for s ∈ [0, +∞), (4.3)
Ωϕ,u∗ (s)
ϕ

1
is of class C on [0, +∞), and
Iϕ′ (s) = u∗ϕ (s) for s ≥ 0. (4.4)
Let Bs denote a ball centered at the origin with m(Bs ) = s and hyperbolic radius ρ . Specifically,
Bs = {z ∈ BN : |z| < tanh ρ}.
Then, we have the following relation between ρ and s which show that the measure of a ball has
exponential growth in the hyperbolic radius
Z Z
1
s= dm(w) = dv(w)
2 N +1
{|w|<tanh ρ} {|w|<tanh ρ} (1 − |w| )
Z tanh ρ
κ2N −1 (4.5)
=2N N +1

2
(1 − κ )
0

= (sinh ρ)2N .
We shall utilize the above relation between s and ρ in the following lemma, where we establish the
expression for the first and second order derivatives of the integral function of (1 − |z|2 )α over the
balls of measure s. The explicit expression of this integral function can be computed directly for
N = 1, and the required derivatives can be obtained with the help of that expression. However,
the situation is different for N > 1, and we have included a detailed proof of the lemma using an
approach similar to that used in the proof of Theorem 2.1 from [6].
Lemma 4.3. Let α > N be fixed and Bs be a ball centered at the origin such that m(Bs ) = s.
Define the function J(s) as Z

J(s) = 1 − |z|2 dm(z). (4.6)
Bs
9

Then, we have
−α
J ′ (s) = 1 + s1/N , (4.7)
and
αsJ ′ (s)
J ′′ (s) = − 2N −1 . (4.8)
Ns N + N s2
Proof. If ρ is the hyperbolic radius of Bs , then by relation (4.5) between s and ρ, we can express
the elements of Bs as follows:
( )  
2
 α 1 2 α
 1
Bs = z ∈ BN : 1 − |z| > α = z ∈ BN : 1 − |z| > α .
1 + sinh2 ρ (1 + s1/N )
α e 1 |−1 and apply the Coarea formula as
For u1 (z) = (1 − |z|2 ) , we consider h(z) = χBs (z)u1 (z)|∇u g
given in [7], to write
Z Z 1 Z 
−1
J(s) = χBs (z)u1 (z)dm(z) = −α
u1 (z)|∇u
e 1 |g dσg dk,
BN (1+s1/N ) u1 (z)=κ

where dσg denotes the hyperbolic surface area measure on ∂Bs = {z ∈ BN : u1 (z) = κ} induced by
the Bergmann metric on BN . We deduce that
Z
′ α 1/N −2α−1 N
 1
−1 e 1 |−1 dσg .
J (s) = 1+s s |∇u g (4.9)
N u1 (z)=(1+s 1/N
)
−α

Next, we apply Cauchy-Schwartz inequality which is indeed an equality in the case of the hyperbolic
surface area of ∂Bs to get
R 2
Z  dσ g
e 1 |−1 dσg = R ∂Bs
|∇u .
g
∂Bs
∂Bs
|∇u1 |g dσg
e

Using the isoperimetric inequality on BN (see [7]), the numerator term in the above expression is
2N −1
4N 2 s N + 4N 2 s2 . Now we compute the denominator term. Let ν be the outward unit normal to
∂Bs with respect to the Bergman metric. Then, we have |∇u e 1 |g = −⟨∇u
e 1 , ν⟩g . Since for z ∈ ∂Bs
 −α
we have u1 (z) = 1 + s1/N , we obtain

|∇u
e 1 |g |∇u
e 1 |g ⟨∇u
e 1 , ν⟩g
−α = = − = −⟨∇
e log u1 , ν⟩g .
(1 + s1/N ) u1 u1
Using the Gauss divergence theorem, we have
!
|∇u
e 1 |g
Z Z
e 1 |g dσg = 1 + s1/N −α

|∇u dσg
∂Bs ∂Bs u1
Z 
1/N −α

=− 1+s ⟨∇ log u1 , ν⟩g dσg
e
∂Bs
Z 
1/N −α

=− 1+s ∆e (log u1 ) dm(z) .
Bs
10

Note that u1 (z) ̸= 0 for z ∈ Bs . As a result, log u1 (z) is well defined on Bs and ∂Bs , and a simple
computation shows that ∆ e (log u1 ) = −4N α. Therefore, we obtain
Z
e 1 |g dσg = 4N αs 1 + s1/N −α .

|∇u
∂Bs

From the above calculation, we deduce from equation (4.9) that


−α
J ′ (s) = 1 + s1/N .
After differentiating the expression of J ′ (s) w.r.t. s, we arrive at the desired equation (4.8). □
Now we prove the last lemma of this section that is required to establish the equality cases in
Theorem C.
Lemma 4.4. Let ϕ ∈ Aα with ∥ϕ∥Aα = 1 and consider Iϕ (s) and J(s) as defined in (4.3) and (4.6)
respectively. Then, for the function G(s) := Iϕ (s) − J(s), the following statements are equivalent
(1) G′ (0) = 0.
(2) ϕ ≡ eiθ ϕz0 for some z0 ∈ BN , θ ∈ R.
(3) G(s) = 0 for all s ≥ 0.
Proof. (1 ⇐⇒ 2) We have G′ (0) = I ′ (0) − J ′ (0) = u∗ (0) − 1 = ∥u∥∞ − 1. Therefore, G′ (0) = 0 if
and only if ∥u∥∞ = 1. It follows from the equality part of Proposition 2.2 that ∥u∥∞ = 1 if and
only if ϕ ≡ eiθ ϕz0 for some z0 ∈ BN , θ ∈ R.
(2 =⇒ 3) If we consider ϕ ≡ eiθ ϕz0 for some z0 ∈ BN , θ ∈ R, then

(1 − |z0 |2 ) (1 − |z|2 )

iθ 2 2 α
 2 α

uϕ (z) = e ϕz0 (z) 1 − |z| = = 1 − |Υz 0 (z)| .
|1 − ⟨z, z0 ⟩|2
Therefore, Ωϕ,u∗ϕ (s) is a ball centered at z0 and m(Ωϕ,u∗ϕ (s) ) = µϕ (u∗ϕ (s)) = s. Using the change of
variable as given in [20, (1.26)], we have, for every s ≥ 0,
Z Z
2 α

1 − |z|2 dm(z) = J(s).

Iϕ (s) = 1 − |Υz0 (z)| dm(z) =
Ωϕ,u∗ (s) Bs
ϕ

(3 =⇒ 1) This is obvious. With that we conclude the proof of the lemma. □


Now, we are ready to prove the main result of this section.
Proof of Theorem C. We will begin by proving the inequality part of the theorem. Let ϕ ∈ Aα be
a function with Bergman norm 1, and let E ⊂ BN be a set with m(E) = s. Then we can write
E = (E ∩ Ωϕ,u∗ϕ (s) ) ∪ (E\Ωϕ,u∗ϕ (s) ) and Ωϕ,u∗ϕ (s) = (Ωϕ,u∗ϕ (s) ∩ E) ∪ (Ωϕ,u∗ϕ (s) \E).
where, Ωϕ,u∗ϕ (s) = {uϕ > u∗ϕ (s)}. Observe that
uϕ > u∗ϕ (s) on Ωϕ,u∗ϕ (s) \E and uϕ ≤ u∗ϕ (s) on E\Ωϕ,u∗ϕ (s) .
In addition,
m(E) = m(Ωϕ,u∗ϕ (s) ) = s implies m(E\Ωϕ,u∗ϕ (s) ) = m(Ωϕ,u∗ϕ (s) \E).
So we have the following reduction
Z Z
uϕ (z)dm(z) ≤ Iϕ (s) = uϕ (z)dm(z). (4.10)
E Ωϕ,u∗ (s)
ϕ
11

Now to prove the the inequality (1.1) it is enough to show that G(s) ≤ 0 for s > 0, where G(s) is
defined as in the previous lemma. From equations (4.2) and (4.4), we have
Z !−1
−1
I ′′ (s) = −
ϕ ∇u
e ϕ dσg .
{uϕ =u∗ϕ (s)} g

By Cauchy Schwartz inequality, we have


R 
{uϕ =u∗ϕ (s)}
|∇u
e |
ϕ g dσ g
Iϕ′′ (s) ≥ − R 2 . (4.11)
{uϕ =u∗ (s)}
dσg
ϕ

Using the isoperimetric inequality on BN (see [7]), the denominator term in the above expression is
2N −1
bounded below by 4N 2 s N + 4N 2 s2 . To estimate the numerator term, let ν be the outward unit
normal to ∂Ωϕ,u∗ϕ (s) with respect to the Bergman metric. Then, we have |∇u
e ϕ |g = −⟨∇u
e ϕ , ν⟩g and

for z ∈ ∂Ωϕ,u∗ϕ (s) , we have uϕ (z) = uϕ (s). Therefore,
|∇u
e ϕ |g |∇u
e ϕ |g ⟨∇u
e ϕ , ν⟩g

= =− = −⟨∇
e log uϕ , ν⟩g .
uϕ (s) uϕ uϕ
Applying Gauss divergence theorem, we have
   
Z Z Z
∗ ∗
|∇u
e ϕ |g dσg = −u (s) 
ϕ ⟨∇e log uϕ , ν⟩g dσg  = −u (s) 
ϕ ∆
e (log uϕ ) dσg  .
∂Ωϕ,u∗ (s) ∂Ωϕ,u∗ (s) Ωϕ,u∗ (s)
ϕ ϕ ϕ

(4.12)
As uϕ (z) ̸= 0 for z ∈ Ωϕ,u∗ϕ (s) , log uϕ (z) is well defined on Ωϕ,u∗ϕ (s) as well on ∂Ωϕ,u∗ϕ (s) . We
have ∆e log uϕ (z) = 2∆e log |ϕ(z)| + α∆e log (1 − |z|2 ). As ϕ(z) ̸= 0 for z ∈ Ωϕ,u∗ (s) , the first term
ϕ

2∆e log |ϕ(z)| = 0 while the second is α∆ e log (1 − |z|2 ) = −4N α. Thus, the right-hand side of (4.12)
is equal to 4N αsu∗ (s). Finally, we have
′′
αsIϕ′ (s)
Iϕ (s) ≥ − 2N −1 . (4.13)
N s N + N s2

If we take h(s) = 1 + s1/N , then it turns out that (hG′ )′ (s) ≥ 0 for s ≥ 0, that is hG′ is an
increasing function. Note that,
Iϕ (0) = J(0) = 0 and lim Iϕ (s) = lim J(s) = 1.
s→+∞ s→+∞

This implies,
G(0) = 0 and lim G(s) = 0.
s→+∞
It follows from (2.5) that Iϕ′ (0) = ∥uϕ ∥∞ ≤ ∥ϕ∥Apα = 1, which further implies
G′ (0) = Iϕ′ (0) − J ′ (0) ≤ 0.
From Lemma 4.4, G′ (0) = 0 if and only if G(s) = 0 for all s ≥ 0. The other possibility is the case
when G′ (0) < 0. In this case we show that G(s) < 0 for all s > 0. If possible, let G(r1 ) ≥ 0 for
some r1 > 0. If, we set
r0 := inf{r > 0 : G(r) ≥ 0}.
then G(r0 ) = 0. Since G(0) = 0, therefore by Rolle’s theorem there exists s0 ∈ (0, r0 ) such that
G′ (s0 ) = 0. Clearly G(s0 ) < 0 and hence G(r0 ) > G(s0 ). Therefore we can find an s1 ∈ (s0 , r0 )
12

such that G′ (s1 ) > 0. Finally, G(r0 ) = 0 and lims→+∞ G(s) = 0 again by Rolle’s theorem there
will be an s2 ∈ (r0 , +∞) such that G′ (s2 ) = 0. Therefore, we have (hG′ )(s0 ) = 0 = (hG′ )(s2 ), and
(hG′ )(s1 ) > 0. Which leads to a contradiction as hG′ is an increasing function. This prove that
G(s) ≤ 0 for s > 0, and the inequalty (1.1).
Now, we discuss the equality case. From the above discussion, we note that either G(s) < 0
for all s > 0 or G(s) = 0 for all s > 0. Suppose there is equality in (1.1) for some s0 > 0 then
G(s0 ) = 0. This means G(s) = 0 for all s > 0. Hence from Lemma 4.4 it follows that ϕ ≡ eiθ ϕz0 for
some z0 ∈ BN , θ ∈ R. Also, E must coincide (up to measure 0) with Ωϕ,u∗ϕ (s) (otherwise we would
have strict inequality in (4.10)).
Conversely, suppose ϕ ≡ eiθ ϕz0 for some z0 ∈ BN , θ ∈ R. Then, it follows from Lemma 4.4 that
G(s) = 0 for all s ≥ 0 and {uϕ > u∗ϕ (s)} is a ball centered at z0 . Also, since E is equivalent (up to
measure 0) to a ball centered at z0 , with m(E) = s, there is equality in (4.10). As a result, there
is equality in (1.1). □

References
[1] Jogia Bandyopadhyay. Optimal concentration for SU(1, 1) coherent state transforms and an analogue of the
Lieb-Wehrl conjecture for SU(1, 1). Comm. Math. Phys., 285(3):1065–1086, 2009.
[2] Frédéric Bayart, Ole Fredrik Brevig, Antti Haimi, Joaquim Ortega-Cerdà, and Karl-Mikael Perfekt. Contractive
inequalities for Bergman spaces and multiplicative Hankel forms. Trans. Amer. Math. Soc., 371(1):681–707,
2019.
[3] Bernhard G. Bodmann. A lower bound for the Wehrl entropy of quantum spin with sharp high-spin asymptotics.
Comm. Math. Phys., 250(2):287–300, 2004.
[4] Eric A. Carlen. Some integral identities and inequalities for entire functions and their application to the coherent
state transform. J. Funct. Anal., 97(1):231–249, 1991.
[5] Rupert L. Frank. Sharp inequalities for coherent states and their optimizers. Adv. Nonlinear Stud., 23(1):Paper
No. 20220050, 28, 2023.
[6] Aleksei Kulikov. Functionals with extrema at reproducing kernels. Geom. Funct. Anal., 32(4):938–949, 2022.
[7] Xiaoshan Li and Guicong Su. Contraction property on complex hyperbolic ball. arXiv:2411.01911v1, 2024.
[8] Elliott H. Lieb. Proof of an entropy conjecture of Wehrl. Comm. Math. Phys., 62(1):35–41, 1978.
[9] Elliott H. Lieb and Jan Philip Solovej. Proof of an entropy conjecture for Bloch coherent spin states and its
generalizations. Acta Math., 212(2):379–398, 2014.
[10] Elliott H. Lieb and Jan Philip Solovej. Proof of the Wehrl-type entropy conjecture for symmetric SU (N ) coherent
states. Comm. Math. Phys., 348(2):567–578, 2016.
[11] Shunlong Luo. A simple proof of Wehrl’s conjecture on entropy. J. Phys. A, 33(16):3093–3096, 2000.
[12] Fabio Nicola and Paolo Tilli. The Faber-Krahn inequality for the short-time Fourier transform. Invent. Math.,
230(1):1–30, 2022.
[13] A. Perelomov. Generalized coherent states and their applications. Texts and Monographs in Physics. Springer-
Verlag, Berlin, 1986.
[14] A. M. Perelomov. Coherent states for arbitrary Lie group. Comm. Math. Phys., 26:222–236, 1972.
[15] João P. G. Ramos and Paolo Tilli. A Faber-Krahn inequality for wavelet transforms. Bull. Lond. Math. Soc.,
55(4):2018–2034, 2023.
[16] Walter Rudin. Function theory in the unit ball of Cn . Classics in Mathematics. Springer-Verlag, Berlin, 2008.
Reprint of the 1980 edition.
[17] Peter Schupp. On Lieb’s conjecture for the Wehrl entropy of Bloch coherent states. Comm. Math. Phys.,
207(2):481–493, 1999.
[18] Dragan Vukotić. A sharp estimate for Apα functions in Cn . Proc. Amer. Math. Soc., 117(3):753–756, 1993.
[19] Alfred Wehrl. On the relation between classical and quantum-mechanical entropy. Rep. Math. Phys., 16(3):353–
358, 1979.
[20] Kehe Zhu. Spaces of holomorphic functions in the unit ball, volume 226 of Graduate Texts in Mathematics.
Springer-Verlag, New York, 2005.
13

Pratyoosh Kumar,
Department of Mathematics, IIT Guwahati,
Guwahati - 781039, Assam, India.
Email address: [email protected], [email protected]

Mandeep Singh,
Department of Mathematics, IIT Guwahati,
Guwahati - 781039, Assam, India.
Email address: [email protected], [email protected]

You might also like