Siberian Mathematical Journal, Vol. 46, No. 2, pp.

205225, 2005
Original Russian Text Copyright c _ 2005 Astashkin S. V.
EXTRAPOLATION FUNCTORS ON A FAMILY OF SCALES
GENERATED BY THE REAL INTERPOLATION METHOD
S. V. Astashkin UDC 517.982.27
Abstract: A new class of extrapolation functors is dened on a family of scales generated by the real
interpolation method. We prove extrapolation relations for the K - and J-functionals corresponding
to some natural pairs of limit spaces which make it possible to describe the values of these functors.
We can consider these relations as new assertions similar to the classical Yano theorem on estimates
for the norms of operators in interpolation scales of spaces.
Keywords: operator extrapolation, extrapolation functor, rearrangement invariant space, operator
interpolation, real interpolation method
The starting point for extrapolation theory is the classical Yano theorem (see [1] or [2, Chapter 12,
Theorem 4.41]) about the Zygmund spaces L(log L)

and Exp L

with the norms

|f|
L(log L)
=
1
_
0
log

2
_
2
t
_
f

(t) dt and |f|

Exp L
= sup
0<t1
log
1/
2
_
2
t
_
f

(t)
which are important in applications. Here , > 0 and f

(t) is the nonincreasing rearrangement of

a function [f(t)[ on the interval [0, 1]. Suppose that T is a bounded linear operator in the L
p
= L
p
[0, 1]
spaces for all p in some right half-neighborhood of 1 and |T|
LpLp
= O((p 1)

) (p 1+) for some

> 0. Then we can dene T on the Zygmund space L(log L)

so that it becomes a bounded operator

from this space to L
1
. The dual assertion about the dual space Exp L
1/
to L(log L)

is also valid. If
a linear operator T is bounded in L
p
for all suciently large p and |T|
LpLp
= O(p

) (p ) for some
> 0 then T : L

Exp L
1/
.
In the 1990s some general approaches of extrapolation theory began to be developed which are
mainly associated with Jawerth and Milman [35]. The primary goal of this theory consists in studying
the natural limit spaces associated with the interpolation scales of spaces and obtaining estimates for
the norms of operators in these spaces. Jawerth and Milman showed that a source for the Yano-type
theorems is existence of extrapolation constructions (functors) whose values are the limit spaces of these
theorems. In particular, using the functors of intersection and sum , they obtained the following
extrapolation description for the Zygmund spaces of the Yano theorem (for example, see [5, pp. 2223]):

1<p<
_
p

L
p
_
= Exp L
1/
and
1<p<
_
(p 1)

L
p
_
= L(log L)

( > 0).
Jawerth and Milman considered as extrapolation functors only the extreme functors, the sum and
intersection of families of Banach spaces (for example, see [4, 2] or [5, Chapter 2]). This choice leads
only to the generalized Lorentz and Marcinkiewicz spaces (see the denitions below) as the resultant
extrapolation spaces. A broader class of functors (dened however only on the scale of L
p
-spaces) was
introduced in [6]. It is closely connected with the real interpolation method and enables the limit space
to be almost every rearrangement invariant (symmetric) space close to L

and L
1
. This article is
devoted to extending and developing these ideas.
First of all, we extend the denition of functors of [6] to the following family of discrete scales of
Banach spaces. Let be an Orlicz function on [0, ). Denote by A the family of sequences (discrete
scales) of the form

A
K
n
= (A
0
, A
1
)
K
11/(2
n
),(2
n
)
, n = 1, 2, . . . ;
Samara. Translated from Sibirski Matematicheski Zhurnal, Vol. 46, No. 2, pp. 264289, MarchApril, 2005.
Original article submitted August 27, 2004.
0037-4466/05/46020205 c _ 2005 Springer Science+Business Media, Inc. 205
and by B, the family of the form

A
J
n
= (A
0
, A
1
)
J
1/(2
n
),(2
n
)/((2
n
)1)
, n = 1, 2, . . . ,
where

A = (A
0
, A
1
) is an arbitrary Banach pair such that A
1
1
A
0
and (A
0
, A
1
)
K
,p
and (A
0
, A
1
)
J
,p
(0 <
< 1, 1 p ) are the spaces of the real K - and J-methods.
The main results of this article consist in description for the values of some extrapolation functors
(dened below) on the scales of these families. We prove that in the case of A (of B) we precisely
obtain the interpolation spaces with respect to the pair (A
1
, M

A)) (the pair (A

0
,

A)), where M

A)
and

A) are the generalized Marcinkiewicz and Lorentz spaces constructed for the function (u) =
u
1
(log(1 +(e 1)/u)) (0 < u 1). The key point here is the extrapolation relations we obtain below
for K - and J-functionals of the pairs (A
1
, M

A)) and (A
0
,

A)). In the last part of this article we

show that the above-introduced functors can also be dened on some broader families of scales without
changing the values of these functors. This is one of the consequences of the important stability property
of extrapolation constructions.
The results of this article in the particular case of the scale of L
p
-spaces (i.e., for the pair

A =
(L
1
, L

)) were partially announced in [7].

1. Denitions, Notations, and Preliminaries
A detailed exposition of the theory of operator interpolation and the theory of rearrangement invari-
ant spaces can be found in [811].
Henceforth we assume that each embedding of one Banach space into the other is assumed continuous;
i.e., X
1
X
0
means that x X
1
implies that x X
0
and |x|
X
0
C|x|
X
1
for some C > 0. To specify
the constant of the embedding, we sometimes write X
1
C
X
0
.
Let X
0
and X
1
be Banach spaces such that X
1
X
0
. Then we dene the completion of X
1
with
respect to X
0
(or Gagliardo completion) as the set

X
1
of all x X
0
such that there is a sequence
x
n
X
0
with the properties: |x
n
|
X
1
C (n = 1, 2, . . . ) for some C > 0 and x
n
x in X
0
. The
space X
1
is said to be complete with respect to X
0
if

X
1
= X
1
.
If (X
0
, X
1
) is a Banach pair (i.e., the Banach spaces X
0
and X
1
are linearly and continuously
embedded in some Hausdor topological vector space) then we can naturally dene the intersection
X
0
X
1
and the sum X
0
+X
1
with the respective norms
|x|
X
0
X
1
= max
i=0,1
|x|
X
i
,
|x|
X
0
+X
1
= inf|x
0
|
X
0
+|x
1
|
X
1
: x = x
0
+x
1
, x
i
X
i
, i = 0, 1.
Let (X
0
, X
1
) and (Y
0
, Y
1
) be Banach pairs. A triple (X
0
, X
1
, X) of spaces, X
0
X
1
X X
0
+X
1
,
is called an interpolation (exact interpolation) triple with respect to (Y
0
, Y
1
, Y ), Y
0
Y
1
Y Y
0
+ Y
1
,
if every linear operator T on X
0
+ X
1
bounded from X
0
to Y
0
and from X
1
to Y
1
is also bounded from
X to Y (and moreover |T|
XY
max
i=0,1
|T|
X
i
Y
i
). If X
i
= Y
i
(i = 0, 1) and X = Y then we say
that X is an interpolation space with respect to (X
0
, X
1
).
An (exact) interpolation functor is an arbitrary mapping F of the set of Banach pairs to the set of
Banach spaces such that for all Banach pairs

X = (X
0
, X
1
) and

Y = (Y
0
, Y
1
) the triple (X
0
, X
1
, F(

X))
is an (exact) interpolation triple with respect to (Y
0
, Y
1
, F(

Y )). The characteristic function (t) of the

interpolation functor F is dened by the relation F(R, (1/t)R) = (1/(t))R, t > 0. If F is an exact
functor then (t) is a quasiconcave function on (0, ); i.e., (t) increases and (t)/t decreases for t > 0.
An important way for obtaining interpolation spaces is the real interpolation method based on
application of the Peetre K - and J-functionals:
K (t, x; X
0
, X
1
) = inf|x
0
|
X
0
+t|x
1
|
X
1
: x = x
0
+x
1
, x
0
X
0
, x
1
X
1
,
J(t, x; X
0
, X
1
) = max|x|
X
0
, t|x|
X
1
.
206
For a xed t > 0, the rst of them is the norm on the sum X
0
+tX
1
and the second is the norm on the
intersection X
0
tX
1
(if X is a Banach space and > 0 then the space X consists of the same elements
as X but has the norm |x|
X
= |x|
X
). If we take x X
0
+X
1
(x X
0
X
1
) then K (t, x; X
0
, X
1
) is
an increasing concave function of t (J(t, x; X
0
, X
1
) is an increasing convex function).
Let E be a Banach lattice of two-sided real sequences = (
j
)

j=
. If (X
0
, X
1
) is an arbi-
trary Banach pair then the space (X
0
, X
1
)
K
E
of the K -method consists of all x X
0
+ X
1
satisfying
(K (2
j
, x; X
0
, X
1
))
j
E and
|x| = |(K (2
j
, x; X
0
, X
1
))
j
|
E
< .
The space (X
0
, X
1
)
J
E
of the J-method consists of all x X
0
+X
1
representable as
x =

j=
u
j
(w.r.t. the convergence of X
0
+X
1
), where u
j
X
0
X
1
. (1)
The norm in (X
0
, X
1
)
J
E
is dened as inf
{u
j
}
|(J(2
j
, u
j
; X
0
, X
1
))
j
|
E
, where the greatest lower bound is
calculated over all sequences u
j

j=
such that (1) is valid.
If E is a Banach lattice of two-sided sequences and = (
k
)

k=
is a sequence of nonnegative
numbers then the space E(
k
) consists of all a = (a
k
)

k=
such that (a
k

k
)

k=
E and |a|
E(
k
)
=
|(a
k

k
)|
E
. Suppose that E (

) := l

(2
k
) (0 , = E (

l
1
) := l
1
+ l
1
(2
k
)). Then the
mapping (X
0
, X
1
) (X
0
, X
1
)
K
E
((X
0
, X
1
) (X
0
, X
1
)
J
E
) denes an exact interpolation functor. The
collection of all such functors is called the real K -method (J-method). It is important to notice that in
the case of the K -method (J-method) the parameter E can always be considered as an interpolation
space with respect to the pair

l

= (l

, l

(2
k
)) (

l
1
= (l
1
, l
1
(2
k
))) [9, Corollaries 3.3.10 and 3.4.6].
Moreover, if (t) is a quasiconcave function then the functors (, )
J
l
1
(1/(2
k
))
and (, )
K
l(1/(2
k
))
are extreme
in the sense that, for every exact interpolation functor F with the characteristic function (t), we have
(X
0
, X
1
)
J
l
1
(1/(2
k
))
1
F(X
0
, X
1
)
1
(X
0
, X
1
)
K
l(1/(2
k
))
for every Banach pair (X
0
, X
1
).
In particular, for 0 < < 1 and 1 p we obtain the classical interpolation spaces
(X
0
, X
1
)
K
,p
= (X
0
, X
1
)
K
lp(2
k
)
and (X
0
, X
1
)
J
,p
= (X
0
, X
1
)
J
lp(2
k
)
whose properties are studied in detail in [8]. It is well known [8, Theorem 3.3.1] that (X
0
, X
1
)
K
,p
=
(X
0
, X
1
)
J
,p
for arbitrary 0 < < 1 and 1 p . Moreover, it is important to observe that the
constant of the equivalence of their norms depends on and p and may tend to .
Henceforth we speak in particular of the pairs of rearrangement invariant (symmetric) function
spaces on the interval [0, 1]. Recall that a Banach space X of measurable functions on [0, 1] is called
rearrangement invariant provided that
(a) if y = y(t) X and [x(t)[ [y(t)[ then x = x(t) X and |x| |y|;
(b) if y = y(t) X and x

(t) = y

(t) then x X and |x| = |y|.

The simplest and important examples of the rearrangement invariant spaces are the L
p
-spaces
(1 p ) with the usual norms:
|x|
p
=
_
1
_
0
[x(t)[
p
dt
_
1/p
(1 p < ) and |x|

= ess sup
0t1
[x(t)[.
Their natural generalizations are Orlicz spaces. Henceforth by an Orlicz function we mean an increas-
ing continuous convex function N(u) 0 on [0, ) such that N(0) = 0 and N(1) = 1 (thus N always
207
has the inverse N
1
which is concave). The corresponding Orlicz space L
N
consists of all measurable
functions x = x(t) on [0, 1] such that the norm
|x|
L
N
= inf
_
u > 0 :
1
_
0
N
_
[x(t)[
u
_
dt 1
_
is nite. In particular, if N(t) = t
p
(1 p < ) then we obtain the L
p
-space.
Other examples of rearrangement invariant spaces are Lorentz and Marcinkiewicz spaces. If (t) is
a nonnegative increasing concave function on [0, 1] then the Marcinkiewicz space M() consists of all
measurable functions x = x(s) on [0, 1] satisfying
|x|
M()
= sup
0<t1
t
_
0
x

(s) ds
(t)
< ,
and the Lorentz space () consists of all x = x(s) satisfying
|x|
()
=
1
_
0
x

(s) d(s) < .

Henceforth we are interested in the exponential Orlicz spaces Exp L

= L
N

, where N

(u) = (e
(u)

1)/(e 1), and (u) is an Orlicz function. It is well known [12, 13] that the norm in Exp L

is equivalent
to that in the Marcinkiewicz space M() constructed with the concave function (u) = u
1
(log(1 +
(e 1)/u)). Hence, it follows in particular that Exp L

= ()

[10, pp. 152154]. For (u) = u

( 1) we obtain the well-known Zygmund spaces Exp L

and L(log L)
1/
[14] mentioned already in
the introduction in connection with the Yano extrapolation theorem.
In line with [5, p. 8], we recall the denitions of extrapolation space and extrapolation functor
(in the case of discrete scales). A sequence (or discrete scale) A
n

n=1
of Banach spaces is strictly
compatible if there are Banach spaces U
A
and V
A
such that the continuous embeddings V
A
A
n
U
A
(n = 1, 2, . . . ) hold. Suppose that B
n

n=1
is another strictly compatible sequence of Banach spaces
such that V
B
B
n
U
B
(n = 1, 2, . . . ). Then A and B are called extrapolation spaces (with respect
to the scales A
n

n=1
and B
n

n=1
) if V
A
A U
A
and V
B
B U
B
and every linear operator T,
T : U
A
U
B
, sending A
n
to B
n
with |T|
AnBn
1 for all n = 1, 2, . . . , is bounded from A to B.
An extrapolation functor on the family A of strictly compatible scales is a mapping A
n

F
F(A
n
),
A
n
A, such that F(A
n
) and F(B
n
) are extrapolation spaces whenever A
n
and B
n
belong
to the family A.
The simplest extrapolation functors are the sum and intersection functors. Let X
k
(k = 1, 2, . . . )
be Banach spaces embedded linearly and continuously in a Hausdor topological vector space T . Then
their intersection is the Banach space

k=1
X
k
constituted by all x

k=1
X
k
satisfying
|x| = sup
k=1,2,...
|x|
X
k
< .
Suppose additionally that there is a Banach space X
0
embedded into T and such that X
k
1
X
0
(k =
1, 2, . . . ). Then the sum

k=1
X
k
is dened as the set of all x X
0
representable as x =

k=1
x
k
(x
k
X
k
) where

k=1
|x
k
|
X
k
< . This space becomes a Banach space with the norm |x| =
inf

k=1
|x
k
|
X
k
, where the greatest lower bound is calculated over all possible representations of x.
Denote by e
k
(k = 0, 1, 2, . . . ) the standard basis vectors in the space of two-sided real sequences;
i.e., e
k
=
_
e
k
j
_
, e
k
k
= 1 and e
k
j
= 0 (j ,= k). If 1 p then p

is the conjugate of p; i.e., 1/p

+
1/p = 1. The dilation function of a positive function (s), s (0, ), is dened by the relation
M

(t) = sup
s>0
(st)/(s). Finally, the relation F
1
F
2
means henceforth that cF
1
F
2
CF
1
for
some c > 0 and C > 0; moreover, as a rule, the constants c and C are independent of all or a part of the
arguments of F
1
and F
2
.
208
2. Auxiliary Results
In extrapolation theory, it is customary to consider, alongside the usual spaces of the real method,
the modied spaces [5, p. 13]:

(X
0
, X
1
)
K
,p
_
= c
.p
(X
0
, X
1
)
K
,p
, where c
.p
= ((1 )p)
1/p
(1 p < ) and c
,
= 1, (2)
and

(X
0
, X
1
)
J
,p
_
= c

.p
(X
0
, X
1
)
J
,p
, where c

.p
= ((1 )p

)
1/p

(1 < p ) and c

,1
= 1.
The characteristic function of the functors

(, )
K
,p
_
and

(, )
J
,p
_
is exactly t

[5]; moreover [15, Exam-

ple 7],

(L
1
, L

)
K
1/q

,q
) = L
q
(1 q ) (3)
and the norms of the spaces in (3) are equivalent with a constant independent of q. Note that for = 1/q

the constant c
.q
is equal to (q

)
1/q
, whence 1/

2 c
.q
< 1 if q 2. Thereby, by (3),
(L
1
, L

)
K
1/q

,q
= L
q
(4)
with a constant of the equivalence of norms independent of q 2.
Moreover, we need some auxiliary assertions. The rst contains a formula for calculation of the
K -functional of the pair (L

, M()) (M() is the Marcinkiewicz space on [0, 1]). In the particular case
of (t) = t log
1/2
2
(2/t), this was proved in [16] and applied therein to studying the space of multipliers
generated by the Rademacher system.
Lemma 1. Let be an increasing concave function on [0, 1] such that its dilation function satises
the condition M

(1/2) < 1. Then the relation

K (t, f; L

, M()) t sup
u:(u)tu
f

(u)u
(u)
is valid with constants independent of f M() and t > 0.
Proof. First of all, since L

= M(
0
) where
0
(u) = u, we have
K (t, f; L

, M()) |f|
M(t)
, where
t
(u) = max(u, (u)/t), (5)
with constants independent of f M() and t > 0 [17].
Verify that the dilation function M
t
(u) of
t
satises
M
t
(1/2) M

(1/2) < 1 (t > 0). (6)

To this end, represent
M
t
(1/2) =
1
2
sup
0<u1
F
t
(u), where F
t
(u) =
max
_
1,
2(u/2)
tu
_
max
_
1,
(u)
tu
_
. (7)
Since is concave, the following three cases are possible:
(a) (u) tu;
(b) (u) < tu 2(u/2);
(c) 2(u/2) < tu.
In the rst two cases we have:
F
t
(u) =
2(u/2)
(u)
2M

(1/2)
and
F
t
(u) =
2(u/2)
tu
2M

(1/2)
(u)
tu
< 2M

(1/2),
respectively. In the last case F
t
(u) = 1 2M

(1/2). Thereby, (6) is a consequence of (7).

209
Using (6) and arguing as in the proof of Lemma 1.4 of [10], we obtain
s
_
0

t
(u)
u
du C
t
(s),
where C > 0 is independent of s [0, 1] and t > 0. Hence, by the denition of the norm of the
Marcinkiewicz space,
|f|
M(t)
= sup
0<s1
1

t
(s)
s
_
0
f

(u) du
sup
0<s1
1

t
(s)
s
_
0

t
(u)
u
du sup
0<u1
uf

(u)

t
(u)
C sup
0<u1
uf

(u)

t
(u)
.
Since the reverse inequality |f|
M(t)
sup
0<u1
uf

(u)
t(u)
is obvious, it follows from (5) that
K (t, f; L

, M()) sup
0<u1
uf

(u)

t
(u)
= sup
0<u1
f

(u)
max(1, (u)/(ut))
= t sup
u:(u)ut
uf

(u)
(u)
.
Lemma 2. Let Exp L

be the exponential Orlicz space corresponding to an Orlicz function . Then

the inequality
|f|
Exp L
C sup
q1
|f|
q

1
(q)
is valid with some C > 0 independent of f.
Proof. It is easy to verify that log(1 + 2t) c
0
log(1 + t) (t 1), where c
0
:= log 3/ log 2 < 2.
Therefore, by concavity of
1
, the inequality M

(1/2) c
0
/2 < 1 is valid for the function (u) =
u
1
(log(1 + (e 1)/u)). Consequently, by [13] and [10, Theorem 2.5.3],
|f|
Exp L
|f|
M()
sup
0<u1
f

(u)

1
(log(1 + (e 1)/u))
, (8)
whence
|f|
Exp L
C
1
sup
q1
f

((e 1)/(e
q
1))

1
(q)
C
1
sup
q1
f

(e
q
)

1
(q)
.
The assertion of the lemma now follows from the fact that
|f|
q

_
e
q
_
0
f

(t)
q
dt
_
1/q

(e
q
)
e
.
In line with [18], we now consider an approach in which the spaces of the real K -method are
parametrized by rearrangement invariant spaces on the interval [0, 1]. Namely, with each rearrangement
invariant space X on [0, 1] and an arbitrary Banach pair

A = (A
0
, A
1
) we associate the space
X(

A) := a A
0
+A
1
: lim
t+0
K (t, a;

A) = 0 and K

(t, a;

A)
[0,1]
(t) X
with the norm |a|
X(

A)
:= |K

(, a;

A)
[0,1]
|
X
.
As usual, we denote by c
0
the space of all sequences (
k
)

k=
such that lim
k

k
= 0.
210
Lemma 3. Let F be a Banach lattice of two-sided real sequences such that F c
0
+l

(2
k
) and
_
_
_
_

k=1
e
k
_
_
_
_
F
K|e
0
|
F
. (9)
Then for every Banach pair

A = (A
0
, A
1
) such that A
1
A
0
we have
(A
0
, A
1
)
K
F
= X(

A), where X := (L
1
, L

)
K
F
;
moreover, the norms of these spaces satisfy the inequality
1
K + 1
|a|
X(

A)
|a|
(A
0
,A
1
)
K
F
(K + 1)|a|
X(

A)
. (10)
Proof. If a (A
0
, A
1
)
K
F
then, by hypothesis,
lim
t+0
K (t, a;

A) = 0, |a|
(A
0
,A
1
)
K
F
:= |(K (2
k
, a;

A))
k
|
F
< .
Since A
1
1
A
0
, we have K (t, a;

A) = |a|
A
0
(t 1) and, by (9),
_
_
_
_

k=1
K (2
k
, a;

A)e
k
_
_
_
_
F
= |a|
A
0
_
_
_
_

k=1
e
k
_
_
_
_
F
K|a|
A
0
|e
0
|
F
K
_
_
_
_

k=0
K (2
k
, a;

A)e
k
_
_
_
_
F
,
whence
_
_
_
_

k=0
K (2
k
, a;

A)e
k
_
_
_
_
F
|a|
(A
0
,A
1
)
K
F
(K + 1)
_
_
_
_

k=0
K (2
k
, a;

A)e
k
_
_
_
_
F
. (11)
In particular, using the well-known equality K (t, f; L
1
, L

) =
_
t
0
f

(s) ds [8, Theorem 5.2.1], we obtain

_
_
_
_

k=0
2
k
_
0
f

(s) ds e
k
_
_
_
_
F
|f|
X
(K + 1)
_
_
_
_

k=0
2
k
_
0
f

(s) ds e
k
_
_
_
_
F
. (12)
Since the function K (t, a;

A) is absolutely continuous in t [10, p. 67], we now have K (t, a;

A) =
_
t
0
K

(s, a;

A) ds. Using the fact that the derivative K

(s, a;

A) of a concave function decreases, from (11)

and (12) we therefore obtain
|a|
X(

A)
= |K

(, a;

A)
[0,1]
|
X
(K + 1)
_
_
_
_

k=0
2
k
_
0
K

(s, a;

A) ds e
k
_
_
_
_
F
(K + 1)|a|
(A
0
,A
1
)
K
F
.
Thus, the embedding (A
0
, A
1
)
K
F
X(

A) and the left inequality in (10) are proved. The reverse embed-
ding and the right inequality in (10) can be veried similarly by means of (11) and (12).
3. Extrapolation Functors on a Family of
Scales Generated by the Real K -Method
Suppose that (u) is an Orlicz function on [0, ) and

A = (A
0
, A
1
) is an arbitrary Banach pair such
that A
1
1
A
0
. Introduce the discrete scale of spaces constructed for the pair

A by the real K -method:

A
K
n
= (A
0
, A
1
)
K
11/(2
n
),(2
n
)
, n = 1, 2, . . . .
Denote by A the family of all such scales.
211
Definition 1. Let F be an intermediate Banach lattice with respect to the Banach pair

l

=
(l

, l

(2
k
)); i.e., (

) F (

). Dene L
K
,F
(

A
K
n
) as the set of all a A
0
such that the
sequence u
a
=

n=1
|a|

A
K
n
e
n
belongs to F.
It is easy to verify that L
K
,F
__

A
K
n
__
is a Banach space with the norm |a|
L
K
,F
({

A
K
n
})
:= |u
a
|
F
and
the mapping
_

A
K
n
_
L
K
,F
__

A
K
n
__
is an extrapolation functor on the family A in the sense of the
denition of [5, p. 8] (see 1).
Our main goal is to describe the corresponding limit (or extrapolation) spaces of these scales. We
will show that, under some conditions, all of them are precisely interpolation spaces with respect to the
pair (A
1
, M

A)) where M

A) is the generalized Marcinkiewicz space constructed with the increasing

concave function
(u) = u
1
(log(1 + (e 1)/u)) (0 < u 1).
Recall [19, p. 422] that
M

A) = (A
0
, A
1
)
K
l(1/(2
k
))
.
Note that, by (11) and the fact that A
1
1
A
0
, the last space is well-dened in spite of the fact that the
function (u) is dened only on the interval [0, 1].
First of all, we nd an extrapolation description of the K -functional for the pair (A
1
, M

A)),
whence, in particular, it will follow that M

A) is a limit space for the scale

_

A
K
n
_
as n . Recall
that

A
1
is the completion of A
1
with respect to A
0
(see 1).
Theorem 1. Let (u) be an Orlicz function on [0, ). For a Banach pair

A = (A
0
, A
1
) such that
A
1
1
A
0
and

A
1
= A
1
, the relation
K (2
k
, a; A
1
, M

A)) sup
nk
2
kn
|a|

A
K
n
is valid with constants independent of

A, a M

A), and k = 1, 2, . . . .
First, consider the particular case in which

A is the pair (L
1
, L

) of spaces of functions on [0, 1].

Then M

A) coincides with the usual Marcinkiewicz space M() or the exponential Orlicz space Exp L

(with equivalence of norms) [12, 13].

Proposition 1. If is an arbitrary Orlicz function then the following holds:
K (p, f; L

, Exp L

) p sup
qp
|f|
(q)
q
with constants independent of f Exp L

and p 1.
Proof. By Lemma 1,
K (
1
(t), f; L

, Exp L

)
1
(t) sup
0<u(e1)/(e
t
1)
f

(u)

1
(log(1 + (e 1)/u))
(t > 0). (13)
Given q 1, we estimate
|f|
q

_
e
2q
e
2q
_
0
(f

(s))
q
ds
_
1/q
e
2
_
e
2q
_
0
(
1
(log(1 + (e 1)/u)))
q
du
_
1/q
sup
0<ue
q
f

(u)

1
(log(1 + (e 1)/u))
. (14)
212
By change of variables, we obtain
e
2q
_
0
(
1
(log(1 + (e 1)/u)))
q
du = (e 1)

_
log(1+e
2q
)
(
1
(v))
q
e
v
(e
v
1)
2
dv 8I(, q), (15)
where I(, q) :=
_

2q
(
1
(v))
q
e
v
dv.
Since
1
(v) is concave and
1
(v)

(
1
(v)) is increasing; therefore, integrating by parts, we infer
the estimate
I(, q) = e
2q
(
1
(2q))
q
+q

_
2q
e
v
(
1
(v))
q1

(
1
(v))
dv e
2q
2
q
(
1
(q))
q
+
q

1
(2q)

(
1
(2q))
I(, q).
Moreover, since v

(v) (v), we have

1
(2q)

(
1
(2q)) 2q and
I(, q) e
2q
2
q
(
1
(q))
q
+
1
2
I(, q),
whence
I(, q) e
2q
2
1+q
(
1
(q))
q
.
The last inequality together with (13)(15) yields
|f|
q
C
1
K (
1
(q), f; L

, Exp L

) (q 1).
Since K (t, x; X
0
, X
1
) is a concave function of t, we have
|f|
q

1
(q)
C
1
K (
1
(q), f; L

, Exp L

1
(q)
C
1
K (
1
(p), f; L

, Exp L

1
(p)
for all q p 1. Changing variables and using the properties of , we arrive eventually at the estimate
p sup
qp
|f|
(q)
q
C
1
K (p, f; L

, Exp L

).
We have to prove the reverse inequality. To this end, by (13), it is sucient to show that the
inequality
f

(u) C
2
sup
qp
|f|
q

1
(q)

1
(log(1 + (e 1)/u)) (16)
is valid for all 0 < u (e 1)/(e
p
1).
From Lemma 2 and (8) we obtain
f

(u) C
3
sup
q1
|f|
q

1
(q)

1
(log(1 + (e 1)/u)) (0 < u 1). (17)
In the proof of (16) we can assume that f = f

and s [0, 1] : f(s) ,= 0 [0, 2

1p
] (since (e1)/(e
p
1)
2
1p
). If 1 q p then, by Holders inequality,
|f|
q
2
(1p)(pq)
pq
|f|
p
2
2
p
q
|f|
p
.
Moreover,
1
(p)
p
q

1
(q) 2
p/q

1
(q) by concavity of
1
. Hence, for 1 q p
|f|
q

1
(q)

4|f|
p

1
(p)
,
whence
sup
1qp
|f|
q

1
(q)

4|f|
p

1
(p)
.
Thereby (16) follows from (17) and the proposition is proved.
As a consequence, we obtain an extrapolation description for the exponential Orlicz spaces.
213
Corollary 1. For every Orlicz function we have
|f|
Exp L
sup
p1
|f|
(p)
p
.
Proof. Since L

1
Exp L

, by concavity of the K -functional, we have

|f|
Exp L
= K (1, f; L

, Exp L

) K (2, f; L

, Exp L

) 2|f|
Exp L
,
and the claim follows from Proposition 1.
Proof of Theorem 1. Using the denition of generalized Marcinkiewicz space, [19, p. 384], and
the condition

A
1
= A
1
, we nd that
M

A) = (A
0
, A
1
)
K
l(1/(2
k
))
and A
1
= (A
0
, A
1
)
K
l(2
k
)
(18)
isometrically. In particular,
Exp L

= (L
1
, L

)
K
l(1/(2
k
))
and L

= (L
1
, L

)
K
l(2
k
)
. (19)
Since is a concave function and lim
s+0
(s) = 0, the conditions of Lemma 3 are satised for the
spaces l

(2
k
) and l

(1/(2
k
)); hence,
M

A) = Exp L

A), A
1
= L

A), (20)
and the norms of these spaces are equivalent with a constant independent of

A. Moreover, l

(2
k
)
and l

(1/(2
k
)) are interpolation spaces with respect to

l

= (l

, l

(2
k
)) (for example, see [19,
p. 422]). Therefore, putting f
a
(s) := K (s, a;

A) (a A
0
) and applying the reiteration arguments [19,
Corollary 7.1.1], from (18), (19), and Proposition 1 we obtain
K (t, a; A
1
, M

A)) = K
_
t, a; (A
0
, A
1
)
K
l(2
k
)
, (A
0
, A
1
)
K
l(1/(2
k
))
_
K (t, (f
a
(2
k
))
k
; l

(2
k
), l

(1/(2
k
)))
= K
_
t,
_
2
k
_
0
f

a
(s) ds
_
k
; l

(2
k
), l

(1/(2
k
))
_
K (t, f

a
; L

, Exp L

) t sup
qt
|f

a
|
(q)
q
.
Rewrite the last relation in discrete form:
K (2
k
, a; A
1
, M

A)) sup
nk
2
kn
|f

a
|
(2
n
)
(k = 1, 2, . . . ). (21)
It is easy to verify that the parameter l
q
(2
(1/q1)k
) satises (9) with the constant K = 3 for every q 2.
Therefore, by Lemma 3,
(A
0
, A
1
)
K
1/q

,q
= (L
1
, L

)
K
1/q

,q
(

A)
and the norms of these spaces are equivalent with the same constant for all q 2. Using this together
with (4) and (21), we obtain the claim of the theorem.
In particular, the generalized Marcinkiewicz space M

A) is an extrapolation space with respect to

the scale
_

A
K
n
_
.
214
Corollary 2. The following equivalence holds for every Orlicz function and an arbitrary Banach
pair

A = (A
0
, A
1
) such that A
1
1
A
0
and

A
1
= A
1
:
|a|
M(

A)
sup
n1
2
n
|a|

A
K
n
.
Proof. It is easy to verify that A
1
1
M

A). Therefore, the arguments are exactly the same as in

the proof of Corollary 1.
Observe that A
1
(if

A
1
= A
1
) and the Marcinkiewicz space M

A) are the least and greatest extrap-

olation spaces of the scale
_

A
K
n
_
which correspond to the family L
K
,F
of extrapolation functors (with
a xed function ). Namely, we have
L
K
,l
__

A
K
n
__
= A
1
and L
K
,l(2
k
)
__

A
K
n
__
= M

A). (22)
Indeed, by (4), Lemma 3, and Corollary 1,
|u
a
|
l
= sup
n=1,2,...
|a|

A
K
n
sup
k=1,2,...
|K

(, a;

A)
[0,1]
|
(2
k
)
= |K

(, a;

A)
[0,1]
|
L
and
|u
a
|
l(2
k
)
= sup
n=1,2,...
2
n
|a|

A
K
n
sup
n=1,2,...
2
n
|K

(, a;

A)
[0,1]
|
(2
n
)
= |K

(, a;

A)
[0,1]
|
Exp L
.
We are done on applying relations (20).
From (22) we infer that the following embeddings hold for every Banach pair

A = (A
0
, A
1
) such that
A
1
1
A
0
and every Banach lattice F intermediate between l

and l

(2
k
):
A
1
L
K
,F
__

A
K
n
__
M

A).
Moreover, Theorem 2, the main result of this section, implies that L
K
,F
_

A
K
n
_
is an interpolation
space with respect to (A
1
, M

A)), provided that F is an interpolation lattice with respect to

.
Theorem 2. Suppose that is an Orlicz function and

A = (A
0
, A
1
) is a Banach pair such that
A
1
1
A
0
and A
1
is complete with respect to A
0
. Then for every interpolation Banach lattice F with
respect to

= (l

, l

(2
k
)), we have
L
K
,F
__

A
K
n
__
= (A
1
, M

A))
K
F
(with equivalence of norms).
Proof. It is sucient to show that the relation
|u
a
|
F
|v
a
|
F
(23)
holds with some constants independent of a A
0
, where
u
a
=

n=1
|a|

A
K
n
e
n
and v
a
=

n=
K (2
n
, a; A
1
, M

A))e
n
.
Since A
1
1
M

A), we have (v
a
)
n
= 2
n
(v
a
)
0
= 2
n
|a|
M(

A)
for n = 0, 1, 2, . . . . Therefore, by
Lemma 2 of [6], we obtain
|v
a
|
F
|P
+
v
a
|
F
(24)
with a constant independent of a, where P
+
w =

j=1
w
j
e
j
and w = (w
j
)

j=
.
215
It is well known [9, Remark 3.3.8] that
|w|
F
| w|
F
, where w = (w
k
)

k=
, w
k
= sup
n=0,1,...
min[1, 2
kn
][w
n
[,
provided that F is an interpolation space with respect to

. Consequently, |u
a
|
F
| u
a
|
F
, whence, by
the inequalities u
a
P
+
u
a
u
a
, we obtain
|u
a
|
F
|P
+
u
a
|
F
. (25)
Since the norm of a function in L
q
[0, 1] increases in q, application of Lemma 3 to the parameter
l
q
(2
(1/q1)k
) (k 2) using (4) demonstrates that the sequence u
a
increases in the generalized sense:
|a|

A
K
n
C|a|

A
K
m
for some C > 0 and arbitrary 1 n m. Therefore,
P
+
u
a

k=1
2
k
sup
nk
|a|

A
K
n
2
n
e
k
.
Hence, by Theorem 1, |P
+
u
a
|
F
|P
+
v
a
|
F
with universal constants. Thereby relation (23) follows
from (24) and (25).
Consider the important particular case:

A =

L := (L
1
, L

). By (4),

L
K
n
= L
(2
n
)
with a constant
independent of n = 1, 2, . . . . Consequently, L
K
,F
__

L
K
n
__
is a rearrangement invariant space on [0, 1]
consisting of all measurable functions f on [0, 1] such that the sequence u
f
=

k=1
|f|
(2
k
)
e
k
belongs
to F and |f|
L
K
,F
({

L
K
n
})
= |u
f
|
F
.
Corollary 3. The following is valid for an arbitrary Orlicz function and every interpolation
Banach lattice F with respect to (l

, l

(2
k
)):
L
K
,F
(L
(2
n
)
) = (L

, Exp L

)
K
F
(with equivalence of norms).
Interpolation for the pair (L

, Exp L

) is described by the real K -method [17] (also see [20, Propo-

sition 1]). This means that every interpolation space X with respect to this pair is representable in
the form X = (L

, Exp L

)
K
F
, where F is a Banach lattice of two-sided real sequences. Moreover, as
observed in 1, we can assume that F is an interpolation lattice with respect to

l

= (l

, l

(2
k
)).
Therefore, from Theorem 2 we obtain the following extrapolation description for interpolation spaces
with respect to (L

, Exp L

).
Corollary 4. Let be an arbitrary Orlicz function. For every interpolation space X with respect to
(L

, Exp L

), there is a Banach lattice F such that X = L

K
,F
(L
(2
n
)
) (with equivalence of the norms).
4. Extrapolation Functors on a Family of
Scales Generated by the Real J-Method
Suppose that is an Orlicz function and

A = (A
0
, A
1
) is a Banach pair, A
1
1
A
0
. Introduce the scale
of spaces constructed for the pair

A by the real J-method:

A
J
n
= (A
0
, A
1
)
J
1/(2
n
),(2
n
)/((2
n
)1)
, n = 1, 2, . . . .
Denote the family of all such scales by B.
Definition 2. Let a Banach lattice G be intermediate with respect to the Banach pair

l
1
=
(l
1
, l
1
(2
k
)); i.e., (

l
1
) G (

l
1
). Dene L
J
,G
(

A
J
n
) as the set of all b A
0
for which there
is a representation
b =

n=1
b
n
(w.r.t. the convergence of A
0
), b
n


A
J
n
and

n=1
|b
n
|

A
J
n
e
n
G. (26)
216
It is easy to verify that L
J
,G
__

A
J
n
__
is a Banach space with the norm
|b|
L
J
,G
({

A
J
n
})
= inf
_
_
_
_

n=1
|b
n
|

A
J
n
e
n
_
_
_
_
G
,
where the greatest lower bound is taken over all representations (26) and the mapping
_

A
J
n
_

L
J
,G
__

A
J
n
__
is the extrapolation functor on B.
Show that the corresponding extrapolation spaces of the scale
_

A
J
n
_
are interpolation spaces with
respect to (A
0
,

A)), where

A) is the generalized Lorentz space [19, p. 430],

A) = (A
0
, A
1
)
J
l
1
(2
k
(2
k
))
, where (u) = u
1
(log(1 + (e 1)/u)) (0 < u 1).
Since A
1
1
A
0
, in the denition of this space we can conne ourselves to representations of the form
x =

k=0
u
k
(u
k
A
1
) (see the proof of Theorem 4 below).
We start with a theorem on description of the J-functional for the pair (A
0
,

A)). In its proof

we need the following well-known assertion on duality of the J- and K -methods [8, Theorem 3.7.1].
Suppose that

A = (A
0
, A
1
) is a Banach pair such that A
0
A
1
is everywhere dense in A
0
and A
1
. Then
the dual spaces A

0
and A

1
also constitute a Banach pair and the following holds:
_
(A
0
, A
1
)
J
,r
_

=
_
A

0
, A

1
_
K
,r

,
1
r
+
1
r

= 1 (1 r < ). (27)
It is important to observe that (27) holds isometrically for all 0 < < 1 and 1 r < . Indeed,
the embedding
_
(A
0
, A
1
)
J
,r
_

1
(A

0
, A

1
)
K
,r

is proved in [8, Theorem 3.7.1, relation (3)]. The reverse

embedding (also with constant 1) can be veried in exactly the same way as (4) in [8].
Theorem 3. Let (u) be an Orlicz function on [0, ). Given a Banach pair

A = (A
0
, A
1
) such that
A
1
1
A
0
and A
1
is everywhere dense in A
0
, we have
J(2
k
, b; A
0
,

A)) |b|
U
k
,
where U
k
=

nk
2
nk

A
J
n
, with constants independent of

A, b A
0
, and k = 1, 2, . . . .
Proof. For the duality reasons, it is sucient to verify the equivalence of the norms of the corre-
sponding dual spaces. In other words [8, p. 74], it is necessary to show that
K (2
k
, a; A

0
, (

A))

) |a|
(U
k
)
(k = 1, 2, . . . ). (28)
Since, by hypothesis, A
1
is everywhere dense in A
0
and l
1
(2
k
(2
k
)) is an interpolation space with
respect to

l
1
[19, p. 430], by the duality theorem for the real method [19, Theorem 7.5.1],
(

A))

= (A

0
, A

1
)
K
F
,
where F is the dual Banach lattice to l
1
(2
k
(2
k
)) with respect to the bilinear form, )=

k=

k
.
It is easy to verify that F = l

(2
k
/(2
k
)); i.e.,
(

A))

= M

(A

0
, A

1
),
where (u) = u(1/u). Using the denition of generalized Marcinkiewicz space and the equality
K (t, b; A
0
, A
1
) = tK (1/t, b; A
1
, A
0
) (t > 0), (29)
we rewrite the last relation as
(

A))

= M

), where

A

:= (A

1
, A

0
). (30)
217
Since A
1
is everywhere dense in

A
J
n
(n = 1, 2, . . . ) [8, Theorem 3.4.2]; therefore, applying the duality
theorem for innite sums and intersections of Banach spaces [3] and relation (27), we obtain
(U
k
)

=
nk
2
kn
_

A
J
n
_

=
nk
2
kn
(A

0
, A

1
)
K
1/(2
n
),(2
n
)
or, using (29) again,
(U
k
)

=
nk
2
kn
(A

1
, A

0
)
K
11/(2
n
),(2
n
)
=
nk
2
kn

A
K
n
.
By (30), we see that (28) becomes the equivalence
K (2
k
, a; A

0
, M

)) sup
nk
2
kn
|a|

A
K
n
.
The space A

0
is complete with respect to A

1
[10, Theorem 1.2.2] and so applying Theorem 1 to

A

, we
conclude that the last relation is valid with universal constants. Consequently, Theorem 3 is proved.
Corollary 5. Suppose that

A = (A
0
, A
1
) is a Banach pair, A
1
1
A
0
, and A
1
is everywhere dense
in A
0
. If is an Orlicz function and (u) = u
1
(log(1 + (e 1)/u)) then

A) =

n=1
2
n

A
J
n
(with equivalence of norms).
Proof. Since

A)
1
A
0
, we have J(1, b; A
0
,

A))=|b|
(

A)
. By the denition of J-functional,
1
2
|b|
(

A)
J(1/2, b; , A
0
,

A)) |b|
(

A)
,
and the claim follows from Theorem 3.
In the particular case

A = (L
1
, L

) the space

A) coincides with the rearrangement invariant

Lorentz space (). From (4) and (27) we nd that
(L
1
, L

)
J
1/q,q

= L
q

with a constant independent of q 2. Therefore, Theorem 3 yields

Corollary 6. If is an arbitrary Orlicz function then the relations
J(2
k
, g; L
1
, ()) |g|
U
k
,
where U
k
=

jk
_
2
jk
L
r
j
_
and r
j
= (2
j
)/((2
j
) 1) (j = 1, 2, . . . ), are valid with a constant
independent of g () and k = 1, 2, . . . .
We now prove the main result of this section by using Theorem 3.
Theorem 4. Suppose that G is an interpolation Banach lattice with respect to

l
1
= (l
1
, l
1
(2
k
)) and

A = (A
0
, A
1
) is a Banach pair such that A
1
1
A
0
and A
1
is everywhere dense in A
0
. Then L
J
,G
__

A
J
n
__
=
(A
0
,

A))
J
G
(with equivalence of norms).
Proof. Denote Y = (A
0
,

A))
J
G
and put J(t, b) = J(t, b; A
0
,

A)) for arbitrary b

A)
and t > 0. By the denition of spaces of the J-method,
|b|
Y
= inf |(J(2
k
, b
k
))
k
|
G
,
where the greatest lower bound is taken over all representations
b =

k=
b
k
(w.r.t. the convergence of A
0
), b
k

A). (31)
218
If b L
J
,G
__

A
J
n
__
then, by denition,
b =

n=1
c
n
, c
n


A
J
n
and
_
_
_
_

n=1
|c
n
|

A
J
n
e
n
_
_
_
_
G
< .
Then c
n
U
n
and, by Theorem 3, J(2
n
, c
n
) C
1
|c
n
|

A
J
n
for some C
1
> 0. Thereby (31) is valid for the
sequence b
n
such that b
n
= c
n
, n = 1, 2, . . . , and b
n
= 0, n = 0, 1, 2, . . . ; moreover, b
n

A) and
|(J(2
n
, b
n
))
n
|
G
C
1
_
_
_
_

n=1
|c
n
|

A
J
n
e
n
_
_
_
_
G
.
Hence, L
J
,G
__

A
J
n
__
Y and |b|
Y
C
1
|b|
L
J
,G
({

A
J
n
})
.
To prove the reverse embedding, we show rst that the norm of Y can be calculated (to within
equivalence) using only representations (31) with b
k
= 0 for k = 0, 1, 2, . . . .
Since

A)
1
A
0
, we obtain J(2
k
, b) = 2
k
|b|
(

A)
(k = 0, 1, 2, . . . ). Suppose that b
k

A)
satises (31). If C
2
is the constant of the embedding G (

l
1
) then
_
_
_
_

k=0
J(2
k
, b
k
)e
k
_
_
_
_
G
=
_
_
_
_

k=0
2
k
|b
k
|
(

A)
e
k
_
_
_
_
G
C
1
2
_
_
_
_

k=0
2
k
|b
k
|
(

A)
e
k
_
_
_
_
l
1
(2
k
)
= C
1
2

k=0
|b
k
|
(

A)
. (32)
We now consider a new representation:
b =

k=
b

k
, b

k
= b
k
for k = 2, 3, . . . ,
b

1
=

i=1
b
i
and b

k
= 0 for k = 0, 1, . . . .
Then from (32) and the fact that G is a Banach lattice we obtain
|(J(2
k
, b

k
))
k
|
G
|(J(2
k
, b
k
))
k
|
G
+

k=0
|b
k
|
(

A)
|e
1
|
G
(1 + 2C
2
C
3
)|(J(2
k
, b
k
))
k
|
G
,
where C
3
is the constant of the embedding (

l
1
) G.
Thus, if b Y then there is some representation of the form (31) such that b
k
= 0 for k = 0, 1, . . .
and
|b|
Y
C
1
4
_
_
_
_

k=1
J(2
k
, b
k
)e
k
_
_
_
_
G
. (33)
By Theorem 3, for each k = 1, 2, . . . , we can nd a representation
b
k
=

n=k
b
k,n
, where b
k,n


A
J
n
, (34)
such that
J(2
k
, b
k
) C
1
5

n=k
2
nk
|b
k,n
|

A
J
n
. (35)
219
Since G (

l
1
), from (33) and (35) we derive

k=1

n=k
|b
k,n
|

A
J
n

k=1

n=k
2
nk
|b
k,n
|

A
J
n
C
5

k=1
J(2
k
, b
k
)
C
5
C
2
_
_
_
_

k=1
J(2
k
, b
k
)e
k
_
_
_
_
G
C
5
C
4
C
2
|b|
Y
< . (36)
In particular, this demonstrates that the sum of the series

k=1

n=k
b
k,n
is independent of the sum-
mation order. Therefore,
b =

n=1
c
n
(w.r.t. the convergence of A
0
), where c
n
=
n

k=1
b
k,n
.
Moreover, by (34), we have c
n


A
J
n
and
|c
n
|

A
J
n

k=1
|b
k,n
|

A
J
n
. (37)
Now, consider the sequence

b
=

n=1
n

k=1
|b
k,n
|

A
J
n
e
n
. (38)
By (36),

b
=

k=1

k
(w.r.t. the convergence of (

l
1
)), where
k
=

n=k
|b
k,n
|

A
J
n
e
n
.
It follows from (35) that
k
(

l
1
), k = 1, 2, . . . . Since G is an interpolation space with respect to
the pair

l
1
, applying Lemma 4 of [6], we obtain
|
b
|
G
C
6
_
_
_
_

k=1

n=k
max(1, 2
nk
)|b
k,n
|

A
J
n
e
k
_
_
_
_
G
= C
6
_
_
_
_

k=1

n=k
2
nk
|b
k,n
|

A
J
n
e
k
_
_
_
_
G
.
Therefore, (37), (38), (35), and (33) imply eventually that
_
_
_
_

n=1
|c
n
|

A
J
n
e
n
_
_
_
_
G
|
b
|
G
C
4
C
5
C
6
|b|
Y
,
whence b L
J
,G
__

A
J
n
__
and |b|
L
J
,G
({

A
J
n
})
C
4
C
5
C
6
|b|
Y
.
In the particular case

A =

L := (L
1
, L

), relations (4) and (27) imply the equality

L
J
n
= L
rn
,
where r
n
= (2
n
)/((2
n
) 1), with constants independent of n = 1, 2, . . . . Therefore, the value of the
above-introduced functor at this pair is the rearrangement invariant space L
J
,G
__

L
J
n
__
consisting of
all measurable functions g on [0, 1] with a representation g =

k=1
g
k
(w.r.t. the convergence of L
1
),
g
k
L
r
k
and

k=1
|g
k
|
r
k
e
k
G. The norm in L
J
,G
__

L
J
n
__
is given by the relation
|g|
L
J
,G
({

L
J
n
})
= inf
_
_
_
_

k=1
|g
k
|
r
k
e
k
_
_
_
_
G
,
where the greatest lower bound is taken over all possible representations. Eventually we obtain
220
Corollary 7. The equality L
J
,G
(L
rn
) = (L
1
, ())
J
G
(with equivalence of norms) is valid for
an arbitrary Orlicz function and every interpolation Banach lattice G with respect to (l
1
, l
1
(2
k
)).
Since the space () is simultaneously an Orlicz space [12], the interpolation in the pair (L
1
, ())
is described by the real K -method [21, Corollary 5.10]. By the theorems on connection between the
K - and J-functors [9], it is also described by the J-method. In other words, if Y is an interpolation
space with respect to the pair (L
1
, ()) then Y = (L
1
, ())
J
G
for some Banach lattice G interpolation
with respect to the pair

l
1
= (l
1
, l
1
(2
k
)). Therefore, Theorem 4 gives an extrapolation description for
such spaces.
Corollary 8. Let be an arbitrary Orlicz function. For every space Y that is an interpolation space
with respect to (L
1
, ()), there is a Banach lattice G satisfying Y = L
J
,G
(L
rn
) (with equivalence of
norms).
5. Further Generalizations and Stability of Extrapolation Functors
Here we show that the results of two previous sections are valid for a more general family of scales
generated by the real method of interpolation. The key role in the proof is played by a certain stability
of extrapolation intersections of families of Banach spaces. Namely, if the latter are the values of inter-
polation functors at the same Banach pair then, under some conditions, the intersection depends only on
the characteristic functions of these functors.
From a technical point of view, in this section it is more convenient to use functors of the real
interpolation method with functional parameters. For example, in this case the space

A
K
,q
:= (A
0
, A
1
)
K
,q
of the K -method consists of all a A
0
+A
1
such that
|a|
K
,q
=
_

_
0
(t

K (t, a;

A))
q
dt
t
_
1/q
< (0 < < 1, 1 q < )
and
|a|
K
,
= sup
t>0
(t

K (t, a;

A)) < (0 < < 1).

Note that this norm is equivalent to the norm of the corresponding space with the discrete parameter
l
q
(2
k
) (see 1). Moreover, the equivalence constant can be chosen to be the same for all 0 < < 1 and
1 q [8, Lemmas 3.1.3 and 3.2.3]. Thereby we can identify these spaces from both interpolation
and extrapolation points of view.
In the proof of the following theorem we will use some arguments of the proof of Theorem 21 of [5]:
Theorem 5. Suppose that is an Orlicz function on [0, ),
n
= 1 1/(2
n
), and F
n
is an
exact interpolation functor with the characteristic function t
n
(n = 1, 2, . . . ). Then for a Banach pair

A = (A
0
, A
1
) such that A
1
1
A
0
and

A
1
= A
1
,

nk
2
n
F
n
(

A)
nk
2
n

A
K
n,1/(1n)
with constants independent of

A and k = 1, 2, . . . .
Proof. It suces to show (see 1) that

nk
2
n

A
K
n,

nk
2
n

A
J
n,1
(39)
with a constant independent of

A and k = 1, 2, . . . . First, prove that

A
K
,1
_


A
J
,1
(40)
with a universal constant, where

A
K
,p
_
stands, as above, for the modied spaces of the K -method
(see (2)).
221
Let a

A
K
,1
. Then by a strengthened version of the fundamental lemma of interpolation theory [5,
Lemma 1], there is a representation a =
_

0
u(s) ds/s such that

_
0
min(1, t/s)J(s, u(s);

A)
ds
s
K (t, a;

A)
with a universal constant . Therefore, using (2), we obtain
|a|
J
,1

_
0
s

J(s, u(s);

A)
ds
s
= (1 )

_
0
J(s, u(s);

A)

_
0
min(1, t/s)t

dt
t
ds
s
= (1 )

_
0

_
0
J(s, u(s);

A) min(1, t/s)
ds
s
t

dt
t
(1 )

_
0
K (t, a;

A)t

dt
t
= |a|

A
K
,1

.
Thereby (40) is proved. Therefore, to prove (39) it is sucient to verify that

nk
2
n

A
K
n,

nk
2
n

A
K
n,1
) (41)
with a constant independent of

A and k = 1, 2, . . . .
Show that in the case when A
1
1
A
0
the inequality
|a|
K
,q
9
_
1
_
0
(t

K (t, a;

A))
q
dt
t
_
1/q
(42)
holds for all [1/2, 1) and 1 q (we used similar arguments in the proof of Lemma 3).
Indeed, since K (t, a;

A) = |a|
A
0
(t 1) and [1/2, 1), we obtain
_
1
_
0
(t

K (t, a;

A))
q
dt
t
_
1/q

_
1
_
1/4
(t

K (t, a;

A))
q
dt
t
_
1/q
K (1/4, a;

A)
_
1
_
1/4
(t

)
q
dt
t
_
1/q

1
4
|a|
A
0
.
Therefore,
_

_
1
(t

K (t, a;

A))
q
dt
t
_
1/q
|a|
A
0
_

_
1
t
q/2
dt
t
_
1/q
2|a|
A
0
8
_
1
_
0
(t

K (t, a;

A))
q
dt
t
_
1/q
and (42) follows from the denition of |a|
K
,q
.
For each k = 1, 2, . . . , we dene the functions
k
(t) = inf
nk
2
n
t
n
and
k
(t) = inf

k
u<1
M(u)t
u
on
the interval (0, 1], where M(u) =
1
(1/(1 u)).
First of all,

k
(t)
k
(t) 2
k
(t). (43)
Since M(
n
) = 2
n
, the left inequality is obvious. To prove the right, we take u [
k
, 1). Then
n
u <

n+1
for some n k and, since the function M(u) is increasing,
M(u)t
u
M(
n
)t

n+1
= 2
n
t

n+1
=
1
2
M(
n+1
)t

n+1
.
Consequently, (43) is proved.
222
Show now that

k
(t
2
) 2t
k
(t), 0 t 1, k = 1, 2, . . . . (44)
Indeed, since
k
1/2, we obtain

k
(t
2
) = inf

k
u<1
M(u)t
2u
= t inf

k
u<1
M(u)t
2u1
= t inf
2
k
1u<1
M
_
u + 1
2
_
t
u
2t
k
(t),
where in the last inequality we used the fact that 2
k
1 <
k
and
M((u + 1)/2) =
1
(2/(1 u)) 2M(u).
From (43) and (44) we nd that

k
(t
2
) 4t
k
(t), 0 t 1, k = 1, 2, 3, . . . .
This and the denition of the function
k
(t) imply that
1
_
0

k
(t)t

k
dt
t
4
1
_
0
t
1/2

k
(t
1/2
)t

k
dt
t
4 sup
0<t1
(t

k
/2

k
(t
1/2
))
1
_
0
t
(1
k
)/2
dt
t
=
8
1
k
sup
0<t1
(t

k
(t))
8
1
k
2
k
.
Using the last inequality and (42), we nally obtain
sup
nk
2
n
|a|

A
K
n,1

9 sup
nk
2
n
(1
n
)
n
1
_
0
K (t, a;

A)t
n
dt
t
9 sup
nk
2
n
(1
n
)
n
1
_
0

n
(t)t
n
dt
t
sup
0<s1
K (s, a;

A)

n
(s)
72 sup
0<s1
K (s, a;

A)

n
(s)
= 72 sup
0<s1
sup
nk
K (s, a;

A)2
n
s
n
= 72 sup
nk
2
n
sup
0<s1
K (s, a;

A)s
n
= 72 sup
nk
2
n
|a|

A
K
n,
,
which proves (41) and the theorem.
For an arbitrary sequence q = q
n

n=1
, 1 q
n
, and every Banach pair

A = (A
0
, A
1
), A
1
1
A
0
,
we introduce the discrete scale of spaces

A
K
n
(q) = (A
0
, A
1
)
K
11/(2
n
),qn
, n = 1, 2, . . . .
In particular, if q
n
= (2
n
) then we obtain the scale

A
K
n
of 3.
Theorems 1 and 4 yield
Corollary 9. For an arbitrary Orlicz function and a Banach pair

A = (A
0
, A
1
) such that A
1
1
A
0
and

A
1
= A
1
, the following relation is valid:
K (2
k
, a; A
1
, M

A)) sup
nk
2
kn
|a|

A
K
n
( q)
.
with a constant independent of

A, a M

A), a sequence q, and k = 1, 2, . . . .

Under some conditions on q, a similar assertion is valid for the ordinary (nonmodied) K -method.
223
Corollary 10. If (2
n
) q
n
(n = 1, 2, . . . ) then for every Banach pair

A = (A
0
, A
1
) such that
A
1
1
A
0
and

A
1
= A
1
, we have
K (2
k
, a; A
1
, M

A)) sup
nk
2
kn
|a|

A
K
n
( q)
with universal constants.
Proof. Since q
n
2 and
n
:= 1 1/(2
n
) 1/2, it is easy to verify that 1/

2 c
n,qn
e
1/e
(see (2)). Thereby we can replace the modied spaces of the K -method in the previous corollary with
the ordinary spaces.
In particular, under the same conditions on the Banach pair

A, we obtain
K (2
k
, a; A
1
, M

A)) sup
nk
2
kn
|a|
(A
0
,A
1
)
K
11/(2
n
),
.
Theorem 6. Suppose that is an Orlicz function and

A = (A
0
, A
1
) is a Banach pair such that
A
1
1
A
0
and

A
1
= A
1
. If (2
n
) q
n
(n = 1, 2, . . . ) then, for every interpolation Banach lat-
tice F with respect to

= (l

, l

(2
k
)), the equality L
K
,F
__

A
K
n
(q)
__
= (A
1
, M

A))
K
F
is valid (with
equivalence of norms).
Proof. Corollary 10 and inspection of the proof of Theorem 2 demonstrate that it suces to validate
the inequality
|a|

A
K
n
( q)
C|a|

A
K
m
( q)
(1 n m).
To this end, by hypothesis and Theorem 3.4.1 of [8], it suces to show that
|a|
K
n,q
0
n
18|a|
K
m,
(1 n < m),
where, as above,
n
= 1 1/(2
n
) and q
0
n
= (2
n
). The last inequality is obtained from the following
estimates in which we use (42) and the inequality (2
m
) 2 (2
n
):
|a|
K
n,q
0
n
9
_
1
_
0
(t
n
K (t, a;

A))
q
0
n
dt
t
_
1/q
0
n
9 sup
0<t1
(t
m
K (t, a;

A))
_
1
_
0
t
1(2
n
)/(2
m
)
dt
t
_
1/(2
n
)
18|a|
K
m,
.
In conclusion, we dene the dual scale of spaces of the J-method: if s = s
n

n=1
, 1 s
n
, and

A = (A
0
, A
1
) is a Banach pair such that A
1
1
A
0
then

A
J
n
(s) = (A
0
, A
1
)
J
1/(2
n
),sn
, n = 1, 2, . . . .
By analogy with Theorems 3 and 4, we can prove the following
Theorem 7. Suppose that is an Orlicz function and 1 s
n
(2
n
)/((2
n
) 1), n = 1, 2, . . . .
Given a Banach pair

A = (A
0
, A
1
) such that A
1
1
A
0
and A
1
is everywhere dense in A
0
, we then have
J(2
k
, b; A
0
,

A)) |b|
U
k
,
where U
k
=

nk
2
nk

A
J
n
(s), with constants independent of

A, b A
0
, a sequence s, and k = 1, 2, . . . .
Theorem 8. Suppose that is an Orlicz function and 1 s
n
(2
n
)/((2
n
) 1), n = 1, 2, . . . .
Moreover, suppose that G is an interpolation Banach lattice with respect to the pair

l
1
= (l
1
, l
1
(2
k
)),
and

A = (A
0
, A
1
) is a Banach pair such that A
1
1
A
0
and A
1
is everywhere dense in A
0
. Then
L
J
,G
(

A
J
n
(s)) = (A
0
,

A))
J
G
(with equivalence of norms).
224
References
1. Yano S., An extrapolation theorem, J. Math. Soc. Japan, 3, 296305 (1951).
2. Zygmund A., Trigonometric Series [Russian translation], Mir, Moscow (1965).
3. Jawerth B. and Milman M., Extrapolation Spaces with Applications, Amer. Math. Soc., Providence RI (1991). (Mem.
Amer. Math. Soc.; 440.)
4. Jawerth B. and Milman M., New results and applications of extrapolation theory, Israel Math. Conf. Proc., 5, 81105
(1992).
5. Milman M., Extrapolation and Optimal Decompositions with Applications to Analysis, Springer-Verlag, Berlin (1994).
(Lecture Notes in Math.; 1580.)
6. Astashkin S. V., On extrapolation properties of the scale of Lp-spaces, Mat. Sb., 194, No. 6, 2342 (2003).
7. Astashkin S. V., Some new extrapolation estimates for the scale of Lp-spaces, Funktsional. Anal. i Prilozhen., 37,
No. 3, 7377 (2003).
8. Bergh J. and Lofstrom J., Interpolation Spaces. An Introduction [Russian translation], Mir, Moscow (1980).
9. Brudnyi Yu. A. and Krugliak N. Ya., Interpolation Functors and Interpolation Spaces, North-Holland, Amsterdam
(1991).
10. Kren S. G., Petunin Yu. I., and Semenov E. M., Interpolation of Linear Operators [in Russian], Nauka, Moscow (1978).
11. Lindenstrauss J. and Tzafriri L., Classical Banach Spaces. 2. Function Spaces, Springer-Verlag, Berlin (1979).
12. Lorentz G. G., Relations between function spaces, Proc. Amer. Math. Soc., 12, 127132 (1961).
13. Rutitski Ya. B., On some classes of measurable functions, Uspekhi Mat. Nauk, 20, No. 4, 205208 (1965).
14. Bennett C. and Rudnick K., On LorentzZygmund spaces, Dissertationes Math., 175, 567 (1980).
15. Milman M., A note on extrapolation theory, J. Math. Anal. Appl., 282, 2647 (2003).
16. Astashkin S. V., A space of multipliers generated by Rademacher system, Mat. Zametki, 75, No. 2, 173181 (2004).
17. Cwikel M. and Nilsson P., Interpolation of Marcinkiewicz spaces, Math. Scand., 56, 2942 (1985).
18. Bennett C., Banach function spaces and interpolation methods, J. Funct. Anal., 17, 409440 (1974).
19. Ovchinnikov V. I., The method of orbits in interpolation theory, Math. Rep. Ser. 2, 1, 349515 (1984).
20. Astashkin S. V., On interpolation of subspaces of rearrangement invariant spaces generated by Rademacher system,
Izv. Ross. Akad. Estestv. Nauk Ser. MMMIU, 1, No. 1, 1835 (1997).
21. Kalton N. J., Calderon couples of rearrangement invariant spaces, Studia Math., 106, No. 3, 233277 (1993).
225