You are on page 1of 9

J. Math. Anal. Appl.

472 (2019) 696–704

Contents lists available at ScienceDirect

Journal of Mathematical Analysis and Applications


www.elsevier.com/locate/jmaa

Multiplication in the space of functions of bounded variation


Stanisław Kowalczyk ∗ , Małgorzata Turowska
Institute of Mathematics, Pomeranian University in Słupsk, ul. Kozietulskiego 6-7, 76-200 Słupsk, Poland

a r t i c l e i n f o a b s t r a c t

Article history: Let BV [0, 1] denote the Banach space of all functions on the unit interval that have
Received 23 April 2018 bounded variation. When endowed with the pointwise product, BV [0, 1] becomes
Available online 16 November 2018 a Banach algebra. We will demonstrate that multiplication in BV [0, 1] is an open
Submitted by R.M. Aron
map. Using an observation of F. Nazarov (who has kindly permitted us to include
Keywords: it), we show that multiplication is not uniformly open and we thereby solve in the
Bilinear operation negative the problem of whether open bilinear maps are automatically uniformly
Open mapping open.
Uniformly open mapping © 2018 Published by Elsevier Inc.
Space of functions of bounded
variation
Multiplication in function spaces

1. Preliminaries

Let C[0, 1] be the space of all continuous real-valued functions defined on [0, 1] with the supremum norm

f ∞ = sup |f (t)|.
t∈[0,1]

There are some natural operations on C[0, 1]; for example, addition, multiplication, minimum and maximum.
Such operations were investigated in [4,13,14]. All of these operations are continuous but only addition,
minimum and maximum are open as mappings from C[0, 1] × C[0, 1] to C[0, 1].
David Fremlin observed during the Winter School in Abstract Analysis held in Lhota, Czech Republic
([4]) that multiplication in the real Banach space C[0, 1] is not open. Indeed, let f (x) = x − 12 (x ∈ [0, 1]).
Then f 2 ∈ B 2 (f, 12 ) \ int B 2 (f, 12 ) due to Intermediate Value Theorem. This example gives a relatively easy
example for the failure of Banach’s open theorem in the case of bilinear maps.
According to [4], a map between topological spaces is called weakly open whenever the image of every
non-empty open set has a non-empty interior.

* Corresponding author.
E-mail addresses: stanislaw.kowalczyk@apsl.edu.pl (S. Kowalczyk), malgorzata.turowska@apsl.edu.pl (M. Turowska).

https://doi.org/10.1016/j.jmaa.2018.11.045
0022-247X/© 2018 Published by Elsevier Inc.
S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704 697

In [4] it was shown that multiplication in C[0, 1] is a weakly open operation. This was generalized in [9]
for C(0, 1) and in [2] for C(X), where X is an arbitrary interval.
In [8,10] some properties of multiplication and other operations in the algebra C(X) of real-valued
continuous functions defined on a compact topological space X are considered. The properties of the product
of open balls and of n open sets in the space of continuous functions on [0, 1] are studied in [5] and [6],
respectively. In [12] relationships between properties of the product of closed order intervals in C(X) and
properties of the underlying space X are studied.
Let X, Y be normed spaces. A map T : X → Y is called uniformly open if it is open and for every ε > 0
there exists δ > 0 such that for every x ∈ X we have BY (T (x), δ) ⊂ T (BX (x, ε)), where BZ (z, r) denotes
an open ball in a space Z with the center z and the radius r > 0, [1].
The authors in [11,7] studied the properties of multiplications in the space BV [0, 1] of real functions of
bounded variation on [0, 1] and in the space CBV [0, 1] of continuous real functions of bounded variation
on [0, 1]. Meanwhile, the authors in [3] stated problems of openness and weak openness of multiplication in
(BV [0, 1],  1BV ) and (CBV [0, 1],  1BV ), where  1BV is the norm in BV [0, 1] and CBV [0, 1] defined by
f 1BV = |f (0)| + V01 (f ) (V01 (f ) is a variation of f on [0, 1]). Some answers are presented in [11]; namely,
multiplication in (CBV [0, 1],  BV ) and (BV [0, 1],  BV ) is weakly open. These results were obtained
again in [7], where the authors study properties of multiplication applying theory of invertible elements
in Banach algebras. We shall prove that multiplication in (BV [0, 1],  BV ) is an open map, unlike in
(CBV [0, 1],  BV ) (at least for the real scalars, as proved by Fremlin).

2. Main results

All stated in Preliminaries properties of multiplication in function spaces have been considered for
real valued functions. The only known, for us, results for complex scalar state that multiplication from
Lp (X) × Lq (X), where (X, μ) is a measure space and p1 + 1q = 1, onto L1 (X) is uniformly open [1] and
multiplication in the space of complex valued functions on [0, 1] is open [6] (in unpublished preprint Build-
ing Block: Multiplication on CC [0, 1] E. Behrends proved that multiplication in this space is uniformly
open).
The results presented in this paper are valid both for real and complex valued functions. From now,
BV [0, 1] denotes the space of functions of bounded variation defined on [0, 1] with values in K, where
K ∈ {R, C}. The space BV [0, 1] is endowed with the norm f BV = f ∞ + V01 (f ), where f ∞ is the
standard supremum norm and V01 (f ) is defined by


n−1
Vab (f ) = sup |f (xi+1 ) − f (xi )|.
a=x1 <...<xn =b i=1

(In the previous papers, ([3,11]), BV [0, 1] was usually endowed with the equivalent norm f 1BV =
|f (0)| + V01 (f ).)
Moreover, B(f, r1 ) · B(g, r2 ) = {F · G : F ∈ B(f, r1 ), G ∈ B(g, r2 )}.

Lemma 2.1. Let f ∈ BV [0, 1] and ε > 0. There exists η > 0 such that for every partition

0 ≤ x1 < x2 < . . . < xm ≤ 1

m−1
with |f (xj )| < η for j ≤ m we have j=1 |f (xj+1 ) − f (xj )| < ε.
698 S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704

Proof. Let us choose a partition 0 = y1 < y2 < . . . < yn = 1 such that


n−1
ε
|f (yi+1 ) − f (yi )| > V01 (f ) − .
i=1
2

Set η = 12n ε
. Take any partition 0 ≤ x1 < x2 < . . . < xm ≤ 1 with |f (xj )| < η for j ≤ m and let us consider
the partition that intertwines both x1 < x2 < . . . < xm and 0 = y1 < y2 < . . . < yn = 1. In the case where
[yi , yi+1 ] contains some subset of x1 < x2 < . . . < xm of consecutive elements, xj < xj+1 < . . . < xJ say,
then

|f (xj ) − f (yi )| + |f (xj+1 ) − f (xj )| + . . . + |f (xJ ) − f (xJ−1 )| + |f (yi+1 ) − f (xJ )|



J
> |f (yi )| − η + |f (yi+1 )| − η + |f (xl+1 ) − f (xl )| − 4η
l=j−1


J
≥ |f (yi+1 ) − f (yi )| + |f (xl+1 ) − f (xl )| − 6η
l=j−1

(we take f (x0 ) = f (xm+1 ) = 0). Consequently, the total sum of increments for this partition is bigger than


n−1 
m−1
|f (yi+1 ) − f (yi )| + |f (xj+1 ) − f (xj )| − 6ηn
i=1 j=1

ε 
m−1
ε
> V01 (f ) − + |f (xj+1 ) − f (xj )| − .
2 j=1
2

Meanwhile, this sum is not bigger than V01 (f ). Thus


m−1
|f (xj+1 ) − f (xj )| − ε < 0. 2
j=1

Theorem 2.2. For every F, G ∈ BV [0, 1] and ε > 0 there exist F1 , G1 ∈ BV [0, 1] such that F · G = F1 · G1 ,
F − F1 BV < ε, G − G1 BV < ε and

inf{|F1 (x)| + |G1 (x)| : x ∈ [0, 1]} > 0.

Proof. By Lemma 2.1, we can find η > 0 such that for every partition

0 ≤ x1 < x2 < . . . < xm ≤ 1,



m−1
ε
if |F (xi )| < η for i≤m then |F (xi+1 ) − F (xi )| < (2.1)
i=1
32

and


m−1
ε
if |G(xi )| < η for i≤m then |G(xi+1 ) − G(xi )| < . (2.2)
i=1
32

Let A be the family of those connected components U of the interior of the set

{x ∈ [0, 1] : |F (x)| + |G(x)| < η}


S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704 699

for which inf{|F (x)| + |G(x)| : x ∈ U } = 0. Since F , G have bounded variation, A must be a finite set, say
A = {(a1 , b1 ), (a2 , b2 ), . . . , (an , bn )}. Let
   
ε ε
ci = max sup |F (x)|, and di = max sup |G(x)|,
x∈(ai ,bi ) 32n x∈(ai ,bi ) 32n
n n
for i ≤ n. By (2.1), we obtain i=1 ci < ε
16 and, by (2.2), i=1 di < ε
16 .
Define f, g : [0, 1] → K, by

 n
F (x) whenever x ∈
/ i=1 (ai , bi ),
f (x) =
ci + di whenever x ∈ (ai , bi ), i = 1, 2, . . . , n

and
 n
G(x) whenever x ∈
/ i=1 (ai , bi ),
g(x) = F (x)G(x)
ci +di whenever x ∈ (ai , bi ), i = 1, 2, . . . , n.

Clearly, f · g = F · G. We claim that F − f BV < 2ε and G − gBV < 2ε .


n
Let us notice that F − f = i=1 fi , where fi is equal to 0 outside (ai , bi ) and fi (x) = F (x) − (ci + di )
for x ∈ (ai , bi ). Obviously, fi ∞ < 2(ci + di ) and

V01 (fi ) = sup Vpq (F − (ci + di )) + lim+ |fi (x)| + lim |fi (x)|
[p,q]⊂(ai ,bi ) x→ai x→b−
i

≤ sup Vpq (F ) + 4(ci + di ).


[p,q]⊂(ai ,bi )

n ε
By (2.1), i=1 sup[p,q]⊂(ai ,bi ) Vpq (F ) < 32 . Hence,


n 
n
F − f BV ≤ F − f ∞ + V01 (fi ) < max 2(ci + di ) + sup Vpq (F )
i≤n
i=1 i=1 [p,q]⊂(ai ,bi )


n n
ε 7ε
+ 4(ci + di ) < 6 (ci + di ) + < .
i=1 i=1
32 8

n F (x)G(x)
Next, observe that G − g = i=1 gi , where gi is equal to 0 outside (ai , bi ) and gi (x) = G(x) − ci +di
for x ∈ (ai , bi ). Then

|F (x)|
gi ∞ ≤ sup |G(x)| + sup |G(x)| ≤ 2di ,
x∈(ai ,bi ) x∈(ai ,bi ) ci + di

|F (x)|
because ci +di < 1 for x ∈ (ai , bi ). Moreover,


1
Vpq (gi ) ≤ Vpq (G)+ Vp (F ) sup |G(x)| + Vp (G) sup |F (x)|
q q
ci + di x∈[p,q] x∈[p,q]


supx∈[p,q] |F (x)| supx∈[p,q] |G(x)|
= Vpq (G) 1 + + Vpq (F ) ≤ 2Vpq (G) + Vpq (F )
ci + di ci + di
700 S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704

for every [p, q] ⊂ (ai , bi ). Hence,





F (x)
V01 (gi ) = sup Vpq (gi )
+ lim+ G(x) 1 − + lim G(x) 1 − F (x)
x→ai ci + di x→bi− ci + di
[p,q]⊂(ai ,bi )

≤ sup (2Vpq (G) + Vpq (F )) + 4di .


[p,q]⊂(ai ,bi )

n ε
By (2.2), i=1 sup[p,q]⊂(ai ,bi ) Vpq (G) < 32 . Hence,


n 
n
G − gBV ≤ G − g∞ + V01 (gi ) < max 2di + sup (2Vpq (G) + Vpq (F ))
i≤n
i=1 i=1 [p,q]⊂(ai ,bi )


n 
n
3ε ε
+ 4di < 6 di + < .
i=1 i=1
32 2

By the definition of the sets (ai , bi ), we obtain

lim (|f (t)| + |g(t)|) > 0 for x ∈ [0, 1) and


t→x+

lim (|f (t)| + |g(t)|) > 0 for x ∈ (0, 1],


t→x−

because if limt→x+ (|f (t)| + |g(t)|) = 0 or limt→x− (|f (t)| + |g(t)|) = 0, then we can find an open interval I
n
such that I ∩ i=1 (ai , bi ) = ∅, |f (t)| + |g(t)| = |F (t)| + |G(t)| < η for t ∈ I and inf t∈I (|F (t)| + |G(t)|) = 0.
n
Let A = {x ∈ [0, 1] : |f (x)| +|g(x)| = 0}. Then A ∩ i=1 (ai , bi ) = ∅ and A ⊂ {x ∈ [0, 1] : |f (x)| + |g(x)| ≥ η}.
This implies that A is a finite set. Let k be the cardinality of A. Finally, we may define F1 : [0, 1] → K by


f (x) for x ∈
/ A,
F1 (x) = ε
32k for x ∈ A.

Then, F1 · g = F · G and F1 − f BV ≤ (2k + 1) 32k


ε
< ε
8. Hence, F1 and G1 = g meet all the required
conditions. 2

Lemma 2.3. Suppose that f ∈ BV [0, 1]. If inf{|f (x)| : x ∈ [0, 1]} ≥ γ > 0 and f BV ≤ Γ, then 1
∈ BV [0, 1]
f

and f1 ≤ γΓ2 .
BV

 
V01 (f )
Proof. Clearly, V01 1
f ≤ γ2 and f1 ≤ γ1 . Then,

V01 (f ) 1 V01 (f ) f ∞ f BV


1 Γ
f ≤ 2
+ ≤ 2
+ 2
= 2
≤ 2. 2
BV γ γ γ γ γ γ

Theorem 2.4. Suppose that F, G ∈ BV [0, 1] are such that

inf{|F (x)| + |G(x)| : x ∈ [0, 1]} > 0.

Then multiplication in BV [0, 1] is open at (F, G), that is, for every ε > 0 there exists η > 0 for which
B(F · G, η) ⊂ B(F, ε) · B(G, ε).
S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704 701

Proof. Set δ = min{1, 12 inf x∈[0,1] (|F (x)| + |G(x)|)}, K = 2 max{F BV , GBV , 1} and choose any
δ8 ∞
ε ∈ (0, 1). Let h ∈ BV [0, 1] be such that hBV < ε 128K 6 . Inductively, we can define sequences (Fn )n=0 ,

(Gn )∞ ∞
n=0 and (hn )n=0 in BV [0, 1] setting F0 = F , G0 = G, h0 = h and

Gn
Fn+1 = Fn + hn · ,
|Fn |2 + |Gn |2
Fn
Gn+1 = Gn + hn · ,
|Fn |2 + |Gn |2
F n · Gn
hn+1 = −h2n ·
(|Fn |2 + |Gn |2 )2

for n ≥ 0. We will show that for all n ≥ 0 we have

(i) Fn Gn + hn = F G + h,
(ii) Fn BV ≤ 12 K + 1 − 2−n and Gn BV ≤ 12 K + 1 − 2−n ,
(iii) inf x∈[0,1] (|Fn (x)| + |Gn (x)|) ≥ δ + δ · 2−n ,
δ8
(iv) hn BV ≤ ε2−n 128K 6.

Verifying that (i) is satisfied is a matter of rote calculation, because F0 G0 + h0 = F G + h and

|Fn |2 |Gn |2
Fn+1 Gn+1 + hn+1 = Fn Gn + hn · + hn · = Fn Gn + hn .
|Fn |2 + |Gn |2 |Fn |2 + |Gn |2

For n = 0, we obtain F0 BV = F BV ≤ 12 K, G0 BV = GBV ≤ 12 K,

inf (|F0 (x)| + |G0 (x)|) = inf (|F (x)| + |G(x)|) ≥ 2δ


x∈[0,1] x∈[0,1]

8
and h0 BV = hBV < ε 128K δ
6 . Let n ≥ 0 and assume that (ii)–(iv) hold for n. In particular, Fn BV < K,

Gn BV < K and inf x∈[0,1] (|Fn (x)| + |Gn (x)|) > δ. Moreover,

|Fn |2 + |Gn |2 ≤ Fn 2BV + Gn 2BV < 2K 2
BV

and

δ2
inf (|Fn (x)|2 + |Gn (x)|2 ) ≥ .
x∈[0,1] 2

Then, by Lemma 2.3,

2K 2
Fn+1 BV ≤ Fn BV + hn BV Gn BV 2
( δ2 )2
1 δ 8 8K 3 1
≤ K + 1 − 2−n + 2−n 6 4
< K + 1 − 2−n−1 ,
2 128K δ 2
and similarly,

2K 2
Gn+1 BV ≤ Gn BV + hn BV Fn BV 2
( δ2 )2
1 δ 8 8K 3 1
≤ K + 1 − 2−n + 2−n < K + 1 − 2−n−1 ,
2 128K 6 δ 4 2
which verifies (ii).
702 S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704

In order to verify (iii), let us observe that

2K 2
|Fn+1 (x)| ≥ |Fn (x)| − hn BV Gn BV 2
( δ2 )2
δ 8 8K 3
≥ |Fn (x)| − 2−n > |Fn (x)| − 2−n−2 δ,
128K 6 δ 4

and

2K 2
|Gn+1 (x)| ≥ |Gn (x)| − hn BV Fn BV 2
( δ2 )2
δ 8 8K 3
≥ |Gn (x)| − 2−n > |Gn (x)| − 2−n−2 δ
128K 6 δ 4

for every x ∈ [0, 1]. Consequently,

inf (|Fn+1 (x)| + |Gn+1 (x)|) ≥ inf (|Fn (x)| + |Gn (x)|) − 2−n−1 δ > δ + δ2−n−1 .
x∈[0,1] x∈[0,1]

Finally, for (iv), let us note that

(2K 2 )2
hn+1 BV ≤ hn 2BV · Fn BV · Gn BV  2 2
( δ2 )2
δ 16 16(2K 2 )2 δ8 δ8
< ε2−2n 6 2
K2 8
= ε2−2n−1 6
≤ ε2−n−1 .
(128K ) δ 128K 128K 6

Thus, we have verified properties (i)–(iv). By (ii), applying Helly’s selection principle, we can find sub-
sequences (Fnk )k≥1 and (Gnk )k≥1 pointwise convergent to some functions of bounded variation f and g,
respectively. (Actually, the sequences (Fn )n≥0 and (Gn )n≥0 are convergent, because they are Cauchy se-
quences and (BV [0, 1],  BV ) is complete.)
By (iv) the sequence (hn )n≥0 is uniformly convergent to 0. Therefore, by (i) we obtain

f · g = lim (Fnk · Gnk ) = lim (Fnk · Gnk + hnk ) = F G + h.


k→∞ k→∞

Finally,


 ∞
 2K 2
f − F BV ≤ Fn+1 − Fn BV ≤ hn BV Gn BV 2
n=0 n=0
( δ2 )2

 ∞
δ8 8K 2 1  −n ε
≤ ε2−n K ≤ ε 2 <
n=0
128K 6 δ4 16K n=0 4

and, similarly, we may check that g − GBV < 4ε . Thus we have proved that B(F · G, η) ⊂ B(F, ε) · B(G, ε)
δ8
for η = ε 128K 6 , which completes the proof. 2

Theorem 2.5. Multiplication in BV [0, 1] is open.


S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704 703

Proof. Let us take an open subset U of BV [0, 1] × BV [0, 1] and (F, G) ∈ U . By Theorem 2.2, we can find
F1 , G1 ∈ BV [0, 1] such that F · G = F1 · G1 , (F1 , G1 ) ∈ U and inf{|F1 (x)| + |G1 (x)| : x ∈ [0, 1]} > 0. Then,
by Theorem 2.4,

B(F · G, δ) ⊂ {f · g : (f, g) ∈ U }

for sufficiently small δ. Therefore, {f · g : (f, g) ∈ U } is open. 2

By [7, Proposition 4.1], openness of multiplication in a Banach algebra implies that the invertible elements
are dense. Thus, we recover [11, Lemma 3.1]:

Corollary 2.6. The set of functions f ∈ BV [0, 1] such that f (x)


= 0 (x ∈ [0, 1]) is dense in BV [0, 1].

The problem of openness of multiplication in BV [0, 1] was again stated in [1, Problem 2, p. 492]. The
same paper posed another problem: whether the notion of uniform openness is essentially stronger than
the notion of openness for the class of continuous bilinear surjective operators between Banach spaces,
[1, Problem 1, p. 492]? We (by applying observation of F. Nazarov) positively answer this question.

Theorem 2.7. Multiplication in BV [0, 1] is not uniformly open.

Proof. For a ∈ K by a · 1[0,1] we denote a constant function a · 1[0,1] (x) = a for x ∈ [0, 1]. Choose any
ε > 0 and k ∈ N such that k1 < ε. Define f = 4 · 1A , where A is the union of k pairwise disjoint open
intervals in (0, 1). Obviously, f ∈ BV [0, 1] and f BV = 8k + 4. Consider the pair (f, 0 · 1[0,1] ). Then
f · (0 · 1[0,1] ) = 0 · 1[0,1] . Let us take any (F, G) ∈ B(f, 1) × B(0 · 1[0,1] , 1). Then, |F (x)| < 1 for x ∈ [0, 1] \ A
and |F (x)| > 3 for x ∈ A. Suppose that F · G = k1 · 1[0,1] . Then |G(x)| > k1 for x ∈ [0, 1] \ A and |G(x)| < 3k 1

for x ∈ A. Therefore,

2 4
GBV > V01 (G) > 2k = > 1,
3k 3

which is a contradiction, because G ∈ B(0 · 1[0,1] , 1). Thus



1
k / B(f, 1) · B(0 · 1[0,1] , 1) and k1 · 1[0,1] BV =
· 1[0,1] ∈ 1
k < ε.

This means that multiplication in BV [0, 1] is not uniformly open. 2

Acknowledgment

The authors would like to thank F. Nazarov for his kind suggestions and observations that simplified
proofs and enriched the article.
The authors would like also to thank the anonymous reviewer for his or her valuable advice and remarks.

References

[1] M. Balcerzak, E. Behrends, F. Strobin, On certain uniformly open multilinear mappings, Banach J. Math. Anal. 10 (3)
(2016) 482–494.
[2] M. Balcerzak, A. Maliszewski, On multiplication in spaces of continuous functions, Colloq. Math. 122 (2011) 247–253.
[3] M. Balcerzak, F. Strobin, A. Wachowicz, Bilinear mappings – selected properties and problems, in: Traditional and
Present-Day Topics in Real Analysis, Łódź University Press, 2013, pp. 281–306 (Chapter 18).
[4] M. Balcerzak, A. Wachowicz, W. Wilczyński, Multiplying balls in the space of continuous functions on [0, 1], Studia Math.
170 (2005) 203–209.
704 S. Kowalczyk, M. Turowska / J. Math. Anal. Appl. 472 (2019) 696–704

[5] E. Behrends, Walk the dog, or: products of open balls in the space of continuous functions, Funct. Approx. Comment.
Math. 44 (2011) 153–164.
[6] E. Behrends, Products of n open subsets in the space of continuous functions on [0, 1], Studia Math. 204 (2011) 73–95.
[7] S. Draga, T. Kania, When is multiplication in a Banach algebra open?, Linear Algebra Appl. 538 (2018) 149–165.
[8] A. Komisarski, A connection between multiplication in C(X) and the dimension of X, Fund. Math. 189 (2006) 149–154.
[9] S. Kowalczyk, Weak openness of multiplication in C(0, 1), Real Anal. Exchange 35 (2010) 235–241.
[10] S. Kowalczyk, On operations in C(X) determined by continuous functions, Acta Math. Hungar. 142 (2013) 56–71.
[11] S. Kowalczyk, M. Turowska, Openness and weak openness of multiplication in the space of functions of bounded variation,
J. Math. Anal. Appl. 434 (2017) 1290–1301.
[12] J. Moreno, R. Schneider, Multiplication of convex sets in C(K) spaces, Studia Math. 232 (2016) 173–187.
[13] A. Wachowicz, Baire category and standard operations on pairs of continuous functions, Tatra Mt. Math. Publ. 24 (2002)
141–146.
[14] A. Wachowicz, On Some Residual Set, PhD dissertation, Łódź Technical University, Łódź, 2004 (in Polish).

You might also like