Ann. Funct. Anal. 2 (2011), no. 2, 22–33

advertisement
Ann. Funct. Anal. 2 (2011), no. 2, 22–33
A nnals of F unctional A nalysis
ISSN: 2008-8752 (electronic)
URL: www.emis.de/journals/AFA/
SOME GEOMETRIC CONSTANTS OF ABSOLUTE
NORMALIZED NORMS ON R2
HIROYASU MIZUGUCHI AND KICHI-SUKE SAITO∗
Communicated by J. I. Fujii
Abstract. We consider the Banach space X = (R2 , k · k) with a normalized,
absolute norm. Our aim in this paper is to calculate the modified Neumann0
Jordan constant CN
J (X) and the Zbăganu constant CZ (X).
1. Introduction and preliminaries
Let X be a Banach space with the unit ball BX = {x ∈ X : kxk ≤ 1} and the
unit sphere SX = {x ∈ X : kxk = 1}. Many geometric constants for a Banach
space X have been investigated. In this paper we shall consider the following
constants;
kx + yk2 + kx − yk2 CN J (X) = sup
(x, y) 6= (0, 0) ,
2(kxk2 + kyk2 )
kx + yk2 + kx − yk2 0
CN J (X) = sup
x, y ∈ SX ,
4
kx + ykkx − yk CZ (X) = sup
x, y ∈ X, (x, y) 6= (0, 0) .
kxk2 + kyk2 The constant CN J (X), called the von Neumann-Jordan constant (hereafter referred to as NJ constant) have been considered in many papers ([3, 8, 10, 12]
and so on). The constant CN0 J (X), called the modified von Neumann-Jordan
constant (shortly, modified NJ constant) was introduced by Gao in [5] and does
not necessarily coincide with CN J (X) (cf. [1, 4, 7]). The constant CZ (X) was
introduced by Zbăganu ([15]) and was conjectured that CZ (X) coincides with
Date: Received: 31 March 2011; Accepted: 14 June 2011.
∗
Corresponding author.
2010 Mathematics Subject Classification. Primary 46B20; Secondary 46B25.
Key words and phrases. Zbăganu constant, absolute norm, von Neumann-Jordan constant.
22
GEOMETRIC CONSTANTS OF R2
23
the von Neumann-Jordan constant CN J (X), but Alonso and Martin [2] gave an
example that CN J (X) 6= CZ (X) (cf.[6, 9]).
A norm k · k on R2 is said to be absolute if k(a, b)k = k(|a|, |b|)k for any
(a, b) ∈ R2 , and normalized if k(1, 0)k = k(0, 1)k = 1. Let AN2 denote the family
of all absolute normalized norm on R2 , and Ψ2 denote the family of all continuous
convex function ψ on [0, 1] such that ψ(0) = ψ(1) = 1 and max{1 − t, t} ≤ ψ(t) ≤
1 for all 0 ≤ t ≤ 1. As in [11], it is well known that AN2 and Ψ2 are in a one-toone correspondence under the equation ψ(t) = k(1 − t, t)k (0 ≤ t ≤ 1). Denote
k · kψ be an absolute normalized norm associated with a convex function ψ ∈ Ψ2 .
For ψ, ϕ ∈ Ψ2 , we denote ψ ≤ ϕ if ψ(t) ≤ ϕ(t) for any t in [0, 1]. Let
ψ(t)
ψ2 (t)
and M2 = max
,
0≤t≤1
ψ2 (t)
ψ(t)
p
where ψ2 (t) = k(1 − t, t)k2 =
(1 − t)2 + t2 corresponds to the l2 -norm. In
[11], Saito, Kato and Takahashi proved that, if ψ ≥ ψ2 (resp. ψ ≤ ψ2 ), then
CN J (C2 , k · kψ ) = M12 (resp. M22 ).
We put X = (R2 , k · kψ ) for ψ ∈ Ψ2 . Our aim in this paper is to consider the
conditions of ψ that CN J (X) = CZ (X) or CN J (X) = CN0 J (X).
In §2, we consider the modified von Neumann-Jordan constant. We prove
that if ψ ≤ ψ2 , then CN0 J (X) = CN J (X) = M22 . If ψ ≥ ψ2 , then we present the
necessarily and sufficient condition that CN0 J (R2 , k · kψ ) = CN J (R2 , k · kψ ) = M12 .
Further, we consider the conditions that CN0 J (R2 , k · kψ ) = CN J (R2 , k · kψ ) =
M12 M22 . In §3, we study the Zbăganu constant. First, we show that, if ψ ≥ ψ2 ,
then CZ (R2 , k · kψ ) = CN J (R2 , k · kψ ) = M12 . If ψ ≤ ψ2 , then we give the
necessarily and sufficient condition for that CZ (R2 , k · kψ ) = CN J (R2 , k · kψ ) =
M22 . Further we study the conditions that CZ (R2 , k · kψ ) = CN J (R2 , k · kψ ) =
M12 M22 . In §4, we calculate the modified NJ-constant CN0 J (X) and the Zbăganu
constant CZ (X) for some normed liner spaces.
M1 = max
0≤t≤1
2. The modified NJ constant of R2
In this section, we consider the Banach space X = (R2 , k · kψ ). From the
definition of the modified NJ constant, it is clear that CN0 J (X) ≤ CN J (X). In
this section, we consider the condition that CN0 J (X) = CN J (X).
Proposition 2.1. Let ψ ∈ Ψ2 . If ψ ≤ ψ2 , then CN0 J (X) = CN J (X) = M22 .
Proof. For any x, y ∈ SX , by [11, Lemma 3],
kx + yk2ψ + kx − yk2ψ ≤ kx + yk22 + kx − yk22
= 2 kxk22 + kyk22
≤ 2M22 kxk2ψ + kyk2ψ = 4M22 .
Now let ψ2 /ψ attain the maximum at t = t0 (0 ≤ t0 ≤ 1), and put
x=
1
1
(1 − t0 , t0 ), y =
(1 − t0 , −t0 ).
ψ(t0 )
ψ(t0 )
24
H. MIZUGUCHI, K.-S. SAITO
Then x, y ∈ SX and
4(1 − t0 )2 + 4t20
ψ(t0 )2
ψ2 (t0 )2
=4
= 4M22 ,
2
ψ(t0 )
kx + yk2ψ + kx − yk2ψ =
which implies that CN0 J (X) = M22 . By [11, Theorem 1], we have this proposition.
If ψ ≥ ψ2 , by [11, Theorem 1], then CN J (X) = M12 . We now give the necessarily
and sufficient condition of CN0 J (X) = M12 .
Theorem 2.2. Let ψ ∈ Ψ2 such that ψ ≥ ψ2 . Then CN0 J (X) = M12 if and only
if there exist s, t ∈ [0, 1] (s < t) satisfying one of the following conditions:
(1) ψ(s) = ψ2 (s), ψ(t) = ψ2 (t) and, if we put r = ψ(s)t+ψ(t)s
, then ψψ(r)
=
ψ(s)+ψ(t)
2 (r)
ψ(1−r)
ψ2 (1−r)
= M1 .
(2) ψ(s) = ψ2 (s), ψ(t) = ψ2 (t) and, if we put r =
ψ(1−r)
ψ2 (1−r)
ψ(t)s+ψ(s)t
,
ψ(t)+ψ(s)(2t−1)
then
ψ(r)
ψ2 (r)
=
= M1 .
Proof. (=⇒) Suppose that CN0 J (X) = M12 . First, for any x, y ∈ SX , by [11,
Lemma 3], we have
kx + yk2ψ + kx − yk2ψ ≤ M12 (kx + yk22 + kx − yk22 )
= 2M12 kxk22 + kyk22
≤ 2M12 kxk2ψ + kyk2ψ = 4M12 .
Since X = (R2 , k · kψ ) is finite dimensional,
kx + yk2ψ + kx − yk2ψ
0
CN J (X) = max
4
x, y ∈ SX .
Therefore, CN0 J (X) = M12 if and only if there exist x, y ∈ SX (x 6= y) such that
kx + yk2ψ + kx − yk2ψ = 4M12 .
From the above inequality, the elements x, y ∈ SX (x 6= y) satisfy kxkψ = kxk2 =
1, kykψ = kyk2 = 1 and
kx + ykψ
kx − ykψ
=
= M1 .
kx + yk2
kx − yk2
Since k · kψ is absolute and x, y ∈ SX (x 6= y) satisfy kxk2 = kyk2 = 1, it is
sufficient to consider the following three cases:
(i) There exist s, t ∈ [0, 1] (s 6= t) satisfying x = ψ21(s) (1 − s, s) and y =
1
(1 − t, t).
ψ2 (t)
(ii) There exist s, t ∈ [0, 1] (s < t) satisfying x = ψ21(s) (1 − s, s) and y =
1
(−1 + t, t).
ψ2 (t)
GEOMETRIC CONSTANTS OF R2
25
(iii) There exist s, t ∈ [0, 1] (s > t) satisfying x = ψ21(s) (1 − s, s) and y =
1
(−1 + t, t).
ψ2 (t)
Case (i). We may suppose that s < t. Then there exist α, β ∈ [0, π2 ] (α < β) such
that
1
1
x=
(1 − s, s) = (cos α, sin α), y =
(1 − t, t) = (cos β, sin β).
ψ2 (s)
ψ2 (t)
Since kxk2 = kyk2 = 1, we have
x+y =(
s
t
α+β
α+β
1−s 1−t
+
,
+
) = ||x + y||2 (cos
, sin
).
ψ2 (s) ψ2 (t) ψ2 (s) ψ2 (t)
2
2
By [13, Propositions 2a and 2b], we remark that
1−s
1−t
s
t
≥
,
≤
.
ψ2 (s)
ψ2 (t) ψ2 (s)
ψ2 (t)
Since x − y is orthogonal to x + y in the Euclidean space (R2 , k · k2 ), we have
1−s
1−t
s
t
x−y =(
−
,
−
)
ψ2 (s) ψ2 (t) ψ2 (s) ψ2 (t)
α+β−π
α+β−π
= ||x − y||2 (cos
, sin
)
2
2
α+β
α+β
= ||x − y||2 (sin
, − cos
).
2
2
Thus we have
α+β
α+β
, sin
)||ψ
||x + y||ψ = ||x + y||2 ||(cos
2
2
sin α+β
α+β
α+β
2
= ||x + y||2 (cos
+ sin
)ψ(
).
2
2
cos α+β
+
sin α+β
2
2
Since ||x + y||ψ = M1 ||x + y||2 , we have
sin α+β
α+β
α+β
2
M1 = (cos
+ sin
)ψ(
).
α+β
2
2
cos 2 + sin α+β
2
Putting r =
We also have
cos
sin α+β
2
α+β
+sin α+β
2
2
, then it is clear that r =
||x − y||ψ = ||x − y||2 (sin
ψ(s)t+ψ(t)s
ψ(s)+ψ(t)
and M1 =
ψ(r)
.
ψ2 (r)
cos α+β
α+β
α+β
2
+ cos
)ψ(
).
α+β
2
2
cos 2 + sin α+β
2
Since ||x − y||ψ = M1 ||x − y||2 , we similarly have
M1 = (sin
cos α+β
α+β
α+β
ψ(1 − r)
2
+ cos
)ψ(
)=
.
α+β
α+β
2
2
ψ2 (1 − r)
sin 2 + cos 2
Case (ii). Then there exist α ∈ [0, π2 ] and β ∈ [ π2 , π] such that
x=
1
1
(1 − s, s) = (cos α, sin α), y =
(−1 + t, t) = (cos β, sin β).
ψ2 (s)
ψ2 (t)
26
H. MIZUGUCHI, K.-S. SAITO
Since kxk2 = kyk2 = 1, we have
x+y =(
1−s
1−t
s
t
α+β
α+β
−
,
+
) = ||x + y||2 (cos
, sin
).
ψ2 (s) ψ2 (t) ψ2 (s) ψ2 (t)
2
2
By [13, Propositions 2a and 2b], we remark that
1−t
s
t
1−s
≥
,
≤
.
ψ2 (s)
ψ2 (t) ψ2 (s)
ψ2 (t)
Since x − y is orthogonal to x + y in the Euclidean space (R2 , k · k2 ), we have
1−s 1−t
s
t
x−y =(
+
,
−
)
ψ2 (s) ψ2 (t) ψ2 (s) ψ2 (t)
α+β−π
α+β−π
= ||x − y||2 (cos
, sin
)
2
2
α+β
α+β
= ||x − y||2 (sin
, − cos
).
2
2
Since cos α+β
≥ 0 and sin α+β
≥ 0, we have
2
2
α+β
α+β
, sin
)||ψ
2
2
sin α+β
α+β
α+β
2
= ||x + y||2 (cos
+ sin
)ψ(
).
2
2
cos α+β
+
sin α+β
2
2
||x + y||ψ = ||x + y||2 ||(cos
Since ||x + y||ψ = M1 ||x + y||2 , we have
sin α+β
α+β
α+β
2
+ sin
)ψ(
M1 = (cos
).
α+β
2
2
cos 2 + sin α+β
2
Putting r =
cos
We also have
sin α+β
2
α+β
α+β
+sin
2
2
, then it is clear that r =
||x − y||ψ = ||x − y||2 (sin
ψ(t)s+ψ(s)t
ψ(t)+ψ(s)(2t−1)
and M1 =
ψ(r)
.
ψ2 (r)
cos α+β
α+β
α+β
2
).
+ cos
)ψ(
α+β
2
2
cos 2 + sin α+β
2
Since ||x − y||ψ = M1 ||x − y||2 , we similarly have
M1 = (sin
cos α+β
α+β
α+β
ψ(1 − r)
2
+ cos
)ψ(
)=
.
α+β
α+β
2
2
ψ2 (1 − r)
sin 2 + cos 2
Case (iii). There exist s, t ∈ [0, 1] (s > t) satisfying x = ψ21(s) (1 − s, s) and
y = ψ21(t) (−1 + t, t). Then, we put s0 = t and t0 = s. We define x0 , y0 in SX by
x0 =
1
1
(1 − s0 , s0 ), y0 =
(−1 + t0 , t0 ).
ψ(s0 )
ψ(t0 )
Then we can reduce Case (ii).
(⇐=). If we suppose (1) (resp. (2)), then we put x = ψ21(s) (1 − s, s) (resp.
x = ψ21(s) (1 − s, s)) and y = ψ21(t) (1 − t, t) (resp. y = ψ21(t) (−1 + t, t)). Then
we have ||x||ψ = ||x||2 = 1, ||y||ψ = ||y||2 = 1, ||x + y||ψ = M1 ||x + y||2 and
GEOMETRIC CONSTANTS OF R2
27
||x − y||ψ = M1 ||x − y||2 . Hence it is clear to prove that CN0 J (X) = M12 . This
completes the proof.
We next study the modified NJ constant in the general case. If ψ ∈ Ψ, then
by [11, Therem 3], we have
max{M12 , M22 } ≤ CN J (X) ≤ M12 M22 .
However, by Theorem 2.2, there exist many ψ ∈ Ψ satisfying ψ ≥ ψ2 such that
CN0 J (X) < max{M12 , M22 } = CN J (X).
From [11, Theorem 3], CN J (X) = M12 M22 if either ψ/ψ2 or ψ2 /ψ attains a maximum at t = 1/2. Then, we have the following
Proposition 2.3. Let ψ ∈ Ψ2 and let ψ(t) = ψ(1 − t) for all t ∈ [0, 1]. If ψ/ψ2
attains a maximum at t = 1/2, then CN0 J (X) = CN J (X) = M12 M22 .
Proof. Suppose first M1 = ψ(1/2)/ψ2 (1/2). Take an arbitrary t ∈ [0, 1] and put
x=
1
1
(t, 1 − t) , y =
(1 − t, t).
ψ(t)
ψ(t)
Then x, y ∈ SX and
kx + ykψ =
2
1
2|2t − 1| 1
ψ( ) , kx − ykψ =
ψ( ).
ψ(t) 2
ψ(t)
2
Therefore we have
ψ(1/2)2
kx + yk2ψ + kx − yk2ψ = (2t − 1)2 + 1
4
ψ(t)2
2
2 ψ(1/2)
= 2ψ2 (t)
ψ(t)2
2
ψ2 (t)2 ψ(1/2)2
2 ψ2 (t)
=
=
M
.
1
ψ(t)2 ψ2 (1/2)2
ψ(t)2
Since t is arbitrary, we have CN0 J (X) ≥ M12 M22 which prove that CN0 J (X) =
M12 M22 .
In the case that M2 = ψ2 (1/2)/ψ(1/2), CN0 J (X) does not necessarily coincide
with M12 M22 . However, we have the following
Theorem 2.4. Let ψ ∈ Ψ2 and let ψ(t) = ψ(1 − t) for all t ∈ [0, 1]. Assume that
M2 = ψ2 (1/2)/ψ(1/2) and M1 > 1. Then CN0 J (X) = M12 M22 if and only if there
exist s, t ∈ [0, 1] (s < t) satisfying one of the following conditions:
, then
(1) ψ2 (s) = M2 ψ(s), ψ2 (t) = M2 ψ(t) and, if we put r = ψ(s)t+ψ(t)s
ψ(s)+ψ(t)
ψ(r) = M1 ψ2 (r).
ψ(t)s+ψ(s)t
, then
(2) ψ2 (s) = M2 ψ(s), ψ2 (t) = M2 ψ(t) and, if we put r = ψ(t)+ψ(s)(2t−1)
ψ(r) = M1 ψ2 (r).
28
H. MIZUGUCHI, K.-S. SAITO
Proof. (=⇒). For all x, y ∈ SX , we have
kx + yk2ψ + kx − yk2ψ ≤ M12 kx + yk22 + kx − yk22
= 2M12 kxk22 + kyk22
≤ 2M12 M22 kxk2ψ + kyk2ψ = 4M12 M22 .
From this inequality, CN0 J (X) = M12 M22 if and only if there exist x, y ∈ SX (x 6=
y) such that
kx + yk2ψ + kx − yk2ψ = 4M12 M22 .
Suppose that CN0 J (X) = M12 M22 . Then, the elements x, y ∈ SX (x 6= y) satisfy
kxk2 = kyk2 = M2 , kx + ykψ = M1 kx + yk2 , kx − ykψ = M1 kx − yk2 .
Since k · kψ is absolute, it is sufficient to consider the following three cases:
1
(1 − s, s) and y =
(i) There exist s, t ∈ [0, 1] (s 6= t) satisfying x = ψ(s)
1
(1 − t, t).
ψ(t)
1
(ii) There exist s, t ∈ [0, 1] (s < t) satisfying x = ψ(s)
(1 − s, s) and y =
1
(−1 + t, t).
ψ(t)
1
(iii) There exist s, t ∈ [0, 1] (s > t) satisfying x = ψ(s)
(1 − s, s) and y =
1
(−1 + t, t).
ψ(t)
As in the proof of Theorem 2.2, we can prove this theorem. This completes
the proof.
3. The Zbăganu constant of R2
The Zbăganu constant CZ (X) in [15] is defined by
kx + ykkx − yk CZ (X) = sup
x, y ∈ X, (x, y) 6= (0, 0) .
kxk2 + kyk2 Then it is clear that CZ (X) ≤ CN J (X) for any Banach space X. In this section,
we consider the condition that CZ (X) = CN J (X) for X = (R2 , k · kψ ). Then, we
have the following
Proposition 3.1. Let ψ ∈ Ψ2 . If ψ ≥ ψ2 , then CZ (X) = CN J (X) = M12 .
Proof. For any x, y ∈ X,
2kx + ykψ kx − ykψ ≤ kx + yk2ψ + kx − yk2ψ
≤ M12 kx + yk22 + kx − yk22
= 2M12 kxk22 + kyk22
≤ 2M12 kxk2ψ + kyk2ψ .
Since ψ/ψ2 attains the maximum at t = t0 (0 ≤ t0 ≤ 1), we put x = (1 − t0 , 0)
and y = (0, t0 ), respectively. Then we have
kx + yk2ψ + kx − yk2ψ = 2ψ(t0 )2
= 2M12 ψ2 (t0 )2
= 2M12 kxk2ψ + kyk2ψ .
GEOMETRIC CONSTANTS OF R2
29
Since kx + ykψ = ψ(t0 ) = kx − ykψ , we have
2kx + ykψ kx − ykψ = kx + yk2ψ + kx − yk2ψ
= 2M12 kxk2ψ + kyk2ψ .
Therefore we have
kx + ykψ kx − ykψ
= M12 ,
kxk2ψ + kyk2ψ
which implies that CZ (X) = M12 .
We next consider the case that ψ ≤ ψ2 . We remark that the Zbăganu constant
CZ (X) is in the following form;
4kxkkyk
CZ (X) = sup
x, y ∈ X, (x, y) 6= (0, 0) .
kx + yk2 + kx − yk2 Then we have the following
Theorem 3.2. Let ψ ∈ Ψ2 . Assume that ψ ≤ ψ2 . Then CZ (X) = M22 if and
only if there exist s, t ∈ [0, 1] (s < t) satisfying one of the following conditions:
2 (r)
(1) ψ(s) = ψ2 (s), ψ(t) = ψ2 (t) and, if we put r = ψ(s)t+ψ(t)s
, then ψψ(r)
=
ψ(s)+ψ(t)
ψ(1−r)
ψ2 (1−r)
= M2 .
(2) ψ(s) = ψ2 (s), ψ(t) = ψ2 (t) and, if we put r =
ψ(1−r)
ψ2 (1−r)
ψ(t)s+ψ(s)t
,
ψ(t)+ψ(s)(2t−1)
then
= M2 .
Proof. For any x, y ∈ X,
4kxkψ kykψ ≤ 2 kxk2ψ + kyk2ψ
≤ 2 kxk22 + kyk22
= kx + yk22 + kx − yk22
≤ M22 kx + yk2ψ + kx − yk2ψ .
Since X = (R2 , k · kψ ) is finite dimensional,
(
4kxkψ kykψ
CZ (X) = max
kx + yk2ψ + kx − yk2ψ
)
x, y ∈ X, (x, y) 6= (0, 0) .
Then CZ (X) = M22 if and only if there exist x, y ∈ SX (x 6= y) such that
4kxkψ kykψ
= M22 .
2
2
kx + ykψ + kx − ykψ
From the above inequality, kxk2 = kxkψ = kykψ = kyk2 and
kx + yk2
kx − yk2
=
= M22 .
kx + ykψ
kx − ykψ
Hence we may assume that
kxk2 = kxkψ = kykψ = kyk2 = 1.
ψ2 (r)
ψ(r)
=
30
H. MIZUGUCHI, K.-S. SAITO
As in the proof of Theorem 2.2, it is sufficient to consider the following three
cases:
(i) There exist s, t ∈ [0, 1] (s 6= t) satisfying x = ψ21(s) (1 − s, s) and y =
1
(1 − t, t).
ψ2 (t)
(ii) There exist s, t ∈ [0, 1] (s < t) satisfying x = ψ21(s) (1 − s, s) and y =
1
(−1 + t, t).
ψ2 (t)
(iii) There exist s, t ∈ [0, 1] (s > t) satisfying x = ψ21(s) (1 − s, s) and y =
1
(−1 + t, t).
ψ2 (t)
As in the proof of Theorem 2.2, we can similarly prove this theorem.
We next study the Zbăganu constant CZ (X) in general case. If ψ ∈ Ψ, by [11,
Theorem 3], then we have
max{M12 , M22 } ≤ CZ (X) ≤ CN J (X) ≤ M12 M22 .
However, by Theorem 3.2, there exist many ψ ∈ Ψ satisfying ψ ≥ ψ2 such that
CZ (X) < CN J (X) ≤ max{M12 , M22 }.
From [11, Theorem 3], CN J (X) = M12 M22 if either ψ/ψ2 or ψ2 /ψ attains a maximum at t = 1/2. Then, we have the following
Proposition 3.3. Let ψ ∈ Ψ2 and let ψ(t) = ψ(1 − t) for all t ∈ [0, 1]. If
2 (1/2)
M2 = ψψ(1/2)
, then CZ (X) = CN J (X) = M12 M22 .
Proof. From the definition, we have CZ (X) ≤ CN J (X) = M12 M22 . Take an arbitrary t ∈ [0, 1] and put x = (t, 1 − t) and y = (1 − t, t). Then kxkψ = kykψ = ψ(t)
and kx + ykψ = k(1, 1)kψ = 2ψ(1/2), kx − ykψ = k(2t − 1, 1 − 2t)kψ = 2|2t −
1|ψ(1/2). Hence we have
2 kxk2ψ + kyk2ψ
4kxkψ kykψ
=
kx + yk2ψ + kx − yk2ψ
kx + yk2ψ + kx − yk2ψ
ψ(t)2
(1 + (2t − 1)2 ) ψ(1/2)2
ψ(t)2
=
2ψ2 (t)2 ψ(1/2)2
2
ψ(t)2 ψ2 (1/2)2
2 ψ(t)
=
=
M
2
ψ2 (t)2 ψ(1/2)2
ψ2 (t)2
=
Since t is arbitrary, we have CZ (X) ≥ M12 M22 . Therefore we have CZ (X) =
M12 M22 . This completes the proof.
In case that M1 = ψ(1/2)/ψ2 (1/2), we have the following theorem as in the
proof of Theorem 2.2 and so omit the proof.
Theorem 3.4. Let ψ ∈ Ψ2 and let ψ(t) = ψ(1−t) for all t ∈ [0, 1]. If M1 = ψψ(1/2)
2 (1/2)
and M2 > 1, then CZ (X) = M12 M22 if and only if there exist s, t ∈ [0, 1] (s < t)
satisfying one of the following conditions:
GEOMETRIC CONSTANTS OF R2
31
(1) ψ2 (s) = M2 ψ(s), ψ2 (t) = M2 ψ(t) and, if we put r = ψ(s)t+ψ(t)s
, then
ψ(s)+ψ(t)
ψ(r) = M1 ψ2 (r).
ψ(t)s+ψ(s)t
(2) ψ2 (s) = M2 ψ(s), ψ2 (t) = M2 ψ(t) and, if we put r = ψ(t)+ψ(s)(2t−1)
, then
ψ(r) = M1 ψ2 (r).
4. Examples
In this section, we calculate CN0 J (X) and CZ (X) of some Banach spaces X =
(R2 , k · kψ ), where ψ ∈ Ψ. First, we consider the case that ψ = ψp .
Example 4.1. Let 1 ≤ p ≤ ∞ and 1/p + 1/q = 1. We put t = min(p, q). Then
2
CN0 J (R2 , k · kp ) = CZ (R2 , || · ||p ) = CN J (R2 , k · kp ) = 2 t −1 .
2
Suppose that 1 ≤ p ≤ 2. Since ψp ≥ ψ2 , we have CZ (R2 , || · ||p ) = 2 p −1 by
Proposition 3.1. On the other hand, as in Theorem 2.2, we take s = 0 and t = 1.
1
− 12
1
p
Since r = ψ(0)·1+ψ(1)·0
=
and
M
=
ψ
(1/2)/ψ
(1/2)
=
2
, by Theorem 2.2,
1
p
2
ψ(0)+ψ(1)
2
2
we have CN0 J (R2 , k · kp ) = M12 = 2 p −1 .
If 2 ≤ p ≤ ∞, then we similarly have, by Proposition 2.1 and Theorem 3.2,
2
0
CN J (R2 , k · kp ) = CZ (R2 , || · ||p ) = CN J (R2 , k · kp ) = 2 p −1 .
In [14, Example], C. Yang and H. Li calculated the modified NJ constant of
the following normed linear space. From our theorems, we have
Example 4.2. Let λ > 0 and Xλ = R2 endowed with norm
||(x, y)||λ = (||(x, y)||2p + λ||(x, y)||2q )1/2 .
(i) If 2 ≤ p ≤ q ≤ ∞, then CN J (Xλ ) = CN0 J (Xλ ) = CZ (Xλ ) =
2(λ+1)
.
22/p +λ22/q
2/p
2/q
2 +λ2
.
2(λ+1)
(ii) If 1 ≤ p ≤ q ≤ 2, then CN J (Xλ ) = CN0 J (Xλ ) = CZ (Xλ ) =
To see this, first, we remark that (p, q) is not necessarily a Hölder pair. We
define the normalized norm || · ||0λ by
||(x, y)||λ
||(x, y)||0λ = √
.
1+λ
Then ||·||0λ is absolute and so put the corresponding function ψλ (t) = ||(1−t, t)||0λ .
(i) Suppose that 2 ≤ p ≤ q ≤ ∞. Since ψλ ≤ ψ2 , by Proposition 2.1, we have
2(λ+1)
CN J (Xλ ) = CN0 J (Xλ ) = M22 = 22/p
. On the other hand, in Theorem 3.2, we
+λ22/q
take s = 0 and t = 1. Then we have r = 1/2 and
M22
ψ2 (1/2)
ψλ (1/2)
= M2 . Thus we have
22/p +λ22/q
.
2(λ+1)
CZ (Xλ ) =
=
(ii) Suppose that 1 ≤ p ≤ q ≤ 2. Since ψλ ≥ ψ2 , by Theorem 2.2 and Proposition
3.1, we similarly have (ii).
Example 4.3. Put


ψ2 (t)
(0 ≤ t ≤ 1/2),
ψ(t) =
 2 − √2 t + √2 − 1 (1/2 ≤ t ≤ 1).
√ √
Then CN0 J (R2 , k · kψ ) < CZ (R2 , k · kψ ) = CN J (R2 , k · kψ ) = 2 2( 2 − 1).
32
H. MIZUGUCHI, K.-S. SAITO
In fact, ψ ∈ Ψ2 and the norm of k · kψ is

p

|a|2 + |b|2
(|a| ≥ |b|)
k(a, b)kψ =
 √
2 − 1 |a| + |b| (|a| ≤ |b|) .
√ √
Since ψ ≥ ψ2 , by Proposition 3.1, we have CZ (R2 , k · kψ ) = M12 = 2 2 2 − 1 .
We assume that CN0 J (R2 , k · kψ ) = M12 . By Theorem 2.2, we can choose r ∈
[0, 1] such that ψψ(r)
= ψψ(1−r)
= M1 . This is impossible by the definition of ψ.
2 (r)
2 (1−r)
0
2
Therefore we have CN J (R , k · kψ ) < M12 .
Example 4.4. Let 1/2 ≤ β ≤ 1. We define a convex function ψβ ∈ Ψ2 by
ψβ (t) = max{1 − t, t, β}.
By [11, Example 4], we have
 2
2
 β + (1 − β)
(β ∈ [ 12 , √12 ])
2
2
β
CN J (R , k · kψβ ) =
 2
2(β + (1 − β)2 ) (β ∈ ( √12 , 1]).
Indeed,
M1 =
(
1
ψβ (1/2)
ψ2 (1/2)
=
β√
1/ 2
=
√
(β ∈ [ 12 , √12 ])
2β
(β ∈ ( √12 , 1])
and
ψ2 (β)
1
= {(1 − β)2 + β 2 }1/2 .
ψβ (β)
β
√
If 1/2 ≤ β ≤ 1/ 2, then ψβ ≤ ψ2 and so, by Proposition 2.1, we have
M2 =
β 2 + (1 − β)2
.
β2
By Theorem 3.2, we have CZ (R2 , k · kψβ ) < M22 .
√
ψ (1/2)
Assume that 1/ 2 < β ≤ 1. Since M1 = ψβ2 (1/2) , we have, by Proposition 2.3,
CN0 J (R2 , k · kψβ ) = M22 =
CN0 J (R2 , k · kψβ ) = M12 M22 = 2(β 2 + (1 − β)2 ).
On the other hand, we take s = β and t = 1 − β in Theorem 3.4. Then we have
= 1/2. By Theorem 3.4, we have
r = ψ(β)(1−β)+ψ(1−β)β
ψ(β)+ψ(1−β)
CZ (R2 , k · kψβ ) = M12 M22 = 2(β 2 + (1 − β)2 ).
√
Example 4.5. We consider ψβ in Example 4.4 in case of β = 1/ 2. Then we
have
√ √
CN0 J (R2 , k · kψβ ) = CN J (R2 , k · kψβ ) = M22 = 2 2( 2 − 1).
√ √
On the other hand, we have CZ (R2 , k · kψβ ) = M22 = 2 2( 2 − 1).
For this ψβ , define a convex function ϕ ∈ Ψ2 by
(
ψβ (t) (0 ≤ t ≤ 1/2),
ϕ(t) =
ψ2 (t) (1/2 ≤ t ≤ 1).
GEOMETRIC CONSTANTS OF R2
33
As in Example 4.2, we similarly have
√ √
CZ (R2 , k · kϕ ) < CN0 J (R2 , k · kϕ ) = CN J (R2 , k · kϕ ) = M22 = 2 2
2−1 .
Acknowledgement. The second author is supported in part by Grants-inAid for Scientific Research, Japan Society for the Promotion of Science (No.
23540189). The authors would also like to thank the referees for some helpful
comments.
References
1. J. Alonso, P. Martin and P.L. Papini, Wheeling around von Neumann-Jordan constant in
Banach Spaces, Studia Math. 188 (2008), no. 2, 135–150.
2. J. Alonso and P. Martin, A counterexample for a conjecture of G. Zbâganu about the
Neumann-Jordan constant, Rev. Roumaine Math. Pures Appl. 51 (2006), 135–141.
3. J.A. Clarkson, The von Neumann-Jordan constant for the Lebesgue spaces, Ann. of Math.
(2) 38 (1937), no. 1, 114–115.
4. J. Gao, A Pythagorean approach in Banach spaces, J. Inequal. Appl. (2006), Art. ID 94982,
1–11.
5. J. Gao and K. Lau, On the geometry of spheres in normed linear spaces, J. Austral. Math.
Soc., 48(1990), 101–112.
6. J. Gao and S. Saejung, Normal structure and the generalized James and Zbăganu constant,
Nonlinear Anal. 71 (2009), no. 7-8, 3047–3052.
7. J. Gao and S. Saejung, Some geometric measures of spheres in Banach spaces, Appl. Math.
Comput. 214 (2009), no. 1, 102–107.
8. P. Jordan and J. von Neumann, On inner products in linear metric spaces, Ann. of Math.
(2) 36 (1935), no. 3, 719–723.
9. E. Llorens-Fuster, E.M. Mazcuñán-Navarro and S. Reich, The Ptolemy and Zbăganu constants of normed spaces, Nonlinear Anal. 72 (2010), no. 11, 3984–3993.
10. M. Kato and Y. Takahashi, On sharp estimates concerning von Neumann-Jordan and James
constants for a Banach space, Rend. Circ. Mat. Palermo, Serie II, Suppl. 82 (2010), 1–17.
11. K.-S. Saito, M. Kato and Y. Takahashi, Von Neumann-Jordan constant of absolute normalized norms on C2 , J. Math. Anal. Appl. 244 (2000), no. 2, 515–532.
12. Y. Takahashi and M. Kato, A simple inequality for the von Neumann-Jordan and James
constants of a Banach space, J. Math. Anal. Appl. 359 (2009), no. 2, 602–609.
13. Y. Takahashi, M. Kato and K. -S. Saito, Strict convexity of absolute norms on C2 and direct
sums of Banach spaces, J. Inequal. Appl., 7(2002), 179–186.
14. C. Yang and H. Li, An inequality between Jordan-con Neumann constant and James constant, Appl. Math. Lett. 23 (2010), no. 3, 277–281.
15. G. Zbăganu, An inequality of M. Rǎdulescu and S. Rǎdulescu which characterizes inner
product spaces, Rev. Roumaine Math. Puers Appl. 47 (2001), 253–257.
Department of Mathematical Sciences, Graduate School of Science and Technology, Niigata University, Niigata, 950-2181 Japan.
E-mail address: mizuguchi@m.sc.niigata-u.ac.jp
E-mail address: saito@math.sc.niigata-u.ac.jp
Download