LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE

FUNCTIONS

VINÍCIUS V. FÁVARO, DANIEL PELLEGRINO, ANSELMO RAPOSO JR., AND GEIVISON RIBEIRO

Abstract. We investigate the existence of algebraic structures in the set of continuous, unbounded
and integrable functions in [0, ∞), continuing the work initiated by Calderón-Moreno, Gerlach-Mena
and Prado-Bassas in 2019. We explore the stricter notions of (α, β)-lineability/spaceability and
pointwise lineability/spaceability, which were recently introduced. For example, we prove that the
aforementioned set is pointwise spaceable, in particular, spaceable. On the other hand, negative
results are also obtained. For instance, we prove the non (ℵ0 , c)-spaceability of the family of
unbounded, continuous and integrable functions.

1. Introduction
The investigation of the algebraic structure of the set of unbounded, continuous and integrable
functions on [0, ∞) was initiated by Calderón-Moreno, Gerlach-Mena and Prado-Bassas in [5], where
the authors prove, among other results, that the set
 
A := f ∈ C [0, ∞) ∩ L1 [0, ∞) : lim sup |f (x)| = ∞
x→∞
is lineable. In this paper we continue this investigation by proving, with different techniques, other
results on the topological and algebraic structure of this set.
We begin recalling the notions of lineability and spaceability. Roughly speaking, these concepts
are related to the search of large algebraic structures within nonlinear subsets of a topological vector
space. Their origins date back to the seminal paper [2] by Aron, Gurariy and Seoane-Sepúlveda,
where the concepts of lineability and spaceability were originally presented.
From now on N denotes the set of positive integers and R represents the real scalar field. The
letters α, β will always represent cardinal numbers, card (A) denotes the cardinality of the set A,
ℵ0 := card (N) and c := card (R). If V is a vector space, we say that A ⊂ V is α-lineable if A ∪ {0}
contains an α-dimensional subspace of V . If, in addition, V has a topology, the subset A is α-spaceable
in V whenever there is a closed α-dimensional subspace of V contained in A ∪ {0}. If p > 0, as usual,
ℓp will denote the classical space of absolutely p-summable sequences of real numbers, that is,
∞
 
∞ N P p
ℓp := (an )n=1 ∈ R : |an | < ∞ .
n=1
Nowadays, this theme has been successfully explored in several research branches of mathematics,
with increasingly relevant applications (see, for instance, [1, 3, 4, 6, 7, 8, 9, 10, 11, 13] and the
references therein).
Towards a stricter notion of lineability, the following concepts were recently introduced in [12, 14]:

2020 Mathematics Subject Classification. 15A03, 47B37, 47L05, 46B86.

Key words and phrases. Lineability; spaceability; unbounded functions; continuous functions; integrable functions.
V.V. Fávaro is supported by FAPEMIG Grant PPM-00217-18 and partially supported by FAPEMIG Grant RED-
00133-21.
D. Pellegrino is supported by CNPq Grant 307327/2017-5 .
1
2 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

• If α < β and V is a vector space, a subset A of V is called (α, β)-lineable if it is α-lineable

and for each α-dimensional subspace Wα ⊂ A ∪ {0} there is a β-dimensional subspace Wβ
such that

(1.1) Wα ⊂ Wβ ⊂ A ∪ {0} .

Note that (α, β)-lineability implies β-lineability. When the vector pace V has a topology
we say that A is (α, β)-spaceable, if A is α-lineable and for each α-dimensional subspace
Wα ⊂ A ∪ {0} there is a closed β-dimensional subspace Wβ satisfying (1.1). Observe that
(α, β)-spaceable means α-lineable (not spaceable) plus the existence of a closed β-dimensional
subspace containing any α-dimensional vector space.
• A subset A of a vector space V is called pointwise β-lineable (spaceable) if, for each x ∈ A,
there is a (closed) β-dimensional subspace Wx such that

x ∈ Wx ⊂ A ∪ {0} .

Obviously, pointwise β-lineability/spaceability implies (1, β)-lineability/spaceability. How-

ever, the converse is not true (see [14, Example 2.2]).
We shall examine some results presented in [5] under the perspective of these new concepts. In
particular, we improve [5, Theorem 2.4] by obtaining pointwise spaceability (and so spaceability)
and [5, Theorem 3.2] obtaining pointwise c-lineability. WeSalso adapt our techniques to prove that,
in contrast to the (1, β)-spaceability of the set Lp [0, 1] \ q∈(p,∞) Lq [0, 1], for p > 0, proved in [12,
Theorem 3.2], this set is not (ℵ0 , c)-spaceable in Lp [0, 1].

2. Algebraic and topological structure of the set of unbounded, continuous and

integrable functions
For p > 0, let Lp [0, 1] and Lp [0, ∞) denote the classical spaces of the (equivalence classes of)
measurable functions f : I → R equipped with the norm (p-norm if 0 < p < 1) defined by
Z 1
p
p
∥f ∥p = |f (t)| dt ,
I

with I = [0, 1] and I = [0, ∞), respectively.

Let C [0, ∞) be the space of all continuous functions f : [0, ∞) → R, endowed with the topology
of the uniform convergence on compacta, under which C [0, ∞) becomes a complete metrizable
topological vector space or F -space. Consider, as in [5],

X := C [0, ∞) ∩ L1 [0, ∞)

equipped with the metric dX : X × X → R given by

P∞ 1 ∥f − g∥∞,n
dX (f, g) := ∥f − g∥1 + n
· ,
n=1 2 1 + ∥f − g∥∞,n
where ∥f ∥∞,n = max {|f (x)| : x ∈ [0, n]}. In [5, Theorem 2.4] it was shown that
 
A := f ∈ X : lim sup |f (x)| = ∞
x→∞

is maximal dense-lineable.
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 3

A prototype of a function ω : [0, ∞) → R in A is the following:


1

 2n s2n (x − n) , if n ≤ x ≤ n + n ,
2 sn




  
1 1 1


ω (x) = −2n s2n x − n − n−1 , if n + n ≤ x ≤ n + n−1 ,

 2 sn 2 sn 2 sn

∞
  

 S 1
 0,

 if x ∈
/ n, n + n−1 ,
n=1 2 sn
where sn = 1 + 1/2 + · · · + 1/n.

s5
s4
s3
s2

s1

1 2 3 4 5

Figure 1. The graph of ω.

In this section we shall show a positive and a negative result on the topological structure of A.
Our first result shows that A is pointwise c-spaceable.
Theorem 2.1. The set A is pointwise c-spaceable in (X, dX ).
Proof. Let f ∈ A and (xn )∞
n=1 be an increasing sequence in [0, ∞) such that
 n→∞
 xn −→ ∞,


(2.1) f (xn ) ̸= 0 for each n = 1, 2, . . . ,

 |f (x )| n→∞

−→ ∞. n
(k)
Given n, k ∈ N, we define tn as follows:
(k)
xn + xn+1 xn + tn
t(1)
n = and tn(k+1) = .
2 2

xn. . . (4) (3) (2) (1) xn+1

tn tn tn tn

(k)
Figure 2. A geometric view of tn .

Now, given k, n ∈ N, let

(k) (k+1)
tn − tn
0 < εn(k) ≤
2
4 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

be such that
1
|f (xn )| ≤ (k)
.
2 n εn
For each positive integer k, let fk : [0, ∞) → R be the function
f (xn ) 
 
(k) (k) (k) (k) (k)


 (k)
x − tn + ε n , if tn − εn ≤ x ≤ tn ,



 εn

 f (x )  
n (k) (k) (k) (k) (k)
(2.2) fk (x) = − (k) x − tn − εn , if tn ≤ x ≤ tn + εn ,


 εn

 ∞ h i
(k) (k) (k) (k)
 S
if x ∈ tn − εn , tn + εn .

 0,
 /
n=1

f (xn )
f3 f2 f1

xn xn+1

Figure 3. Overlaid graphs of the functions fk on [xn , xn+1 ].

Roughly speaking, fk takes the k-th triangle on each interval [xn , xn+1 ]. Observe that:
(i) fk is continuous, for each k ∈ N;
(ii) fk ∈ L1 [0, ∞), for each k ∈ N, because
Z ∞ ∞ ∞ 1
ε(k)
P P
|fk | dx = n |f (xn )| ≤ n
= 1;
0 n=1 n=1 2
(iii) For each k ∈ N,
  
n→∞ n→∞
t(k) −→ ∞ and fk tn(k)  = |f (xn )| −→ ∞.
 
n

Therefore, lim sup |fk (x)| = ∞;

 x→∞ 
(k) (k) (k) (k)
(iv) If x ∈ tn − εn , tn + εn for some positive integers k and n, then, for each j ̸= k, we
have fk (x) ̸= 0 = fj (x).
 
S (k) (k) (k) (k)
Thus, if x ∈/ k,n∈N tn − εn , tn + εn , then fk (x) = 0, for each k ∈ N. In particular,
fk (xn ) = 0 for all positive integers n, k. Fixing a non zero sequence (an )∞
n=0 ∈ ℓ1 , let
g : [0, ∞) → R be the function
∞
P
g (x) = ak fk (x) ,
k=0
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 5
 
(k ) (k ) (k ) (k )
where f0 = f . Let us verify that g is well-defined. If x ∈ tn00 − εn00 , tn00 + εn00 for some
pair k0 , n0 of positive integers, then
g (x) = a0 f (x) + ak0 fk0 (x)
 
S (k) (k) (k) (k)
and, if x ∈
/ k,n∈N tn − εn , tn + εn , then
g (x) = a0 f (x) .
 
(k) (k) (k) (k)
Now we will show that g is continuous. If x0 lies in some interval tn − εn , tn + εn , then
lim g (x) = lim [a0 f (x) + ak fk (x)] = a0 f (x0 ) + ak fk (x0 ) = g (x0 ) .
x→x0 x→x0
 
S (k) (k) (k) (k)
If x0 is an interior point of R \ k,n∈N tn − εn , tn + εn , then
lim g (x) = lim a0 f (x) = a0 f (x0 ) = g (x0 ) .
x→x0 x→x0
(k) (k)
If x0 = tn − εn for some pair k, n of positive integers, then
lim g (x) = lim a0 f (x) = a0 f (x0 ) = g (x0 )
x→x−
0 x→x−
0

and
lim g (x) = lim [a0 f (x) + ak fk (x)] = a0 f (x0 ) + ak fk (x0 ) = a0 f (x0 ) = g (x0 ) .
x→x+
0 x→x+
0

Analogously, we verify that

lim g (x) = g (x0 ) = lim g (x)
x→x−
0 x→x+
0
(k) (k)
if x0 = tn + εn for some positive integers k and n. This assures that g is continuous.
Now, since
∞
P ∞
P
∥ak fk ∥1 ≤ |a0 | ∥f ∥1 + |ak | < ∞,
k=0 k=1
∞
P
then g = ak fk belongs to L1 [0, ∞).
k=0
Finally, notice that, if a0 ̸= 0, then
n→∞
|g (xn )| = |a0 f (xn )| = |a0 | |f (xn )| −→ ∞
and, therefore, lim sup |g (x)| = ∞. If a0 = 0, let j be such that aj ̸= 0. Then,
x→∞
     
n→∞
g t(j)  = aj fj t(j)  = |aj | |f (xn )| −→ ∞
   
n n

and, consequently, lim sup |g (x)| = ∞. Hence, g ∈ A; in particular, we have proved that the operator
x→∞
T : ℓ1 → C [0, ∞) ∩ L1 [0, ∞)
∞
(ak )∞
P
k=0 →
7 ak fk
k=0
is well-defined and T (ℓ1 ) ⊂ A ∪ {0}. It is easy to see that T is linear and injective. Since
c = dim (ℓ1 ) = dim (T (ℓ1 ))
(X,dX ) (X,dX )
and f = T (1, 0, 0, . . .) ∈ T (ℓ1 ), we just need to show that T (ℓ1 ) ⊂ A ∪ {0}, where T (ℓ1 ) ,
denotes the closure of T (ℓ1 ) in (X, dX ).
6 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

(X,dX )
∞  ∞
(j)
Let h ∈ T (ℓ1 ) \ {0}. Then, there exists a sequence c(j) j=1
in ℓ1 , with c(j) = ck such
k=0
that
∞
P (j)
lim ck fk = h
j→∞ k=0
∞
P (j)
in (X, dX ). In particular, ck fk converges uniformly to h over each compact interval [a, b] ⊂ [0, ∞).
k=0
Therefore, fixing n ∈ N and recalling that fk (xn ) = 0 for all positive integers n, k, we have
∞
(j) P (j)
lim c0 f (xn ) = lim ck fk (xn ) = h (xn ) ,
j→∞ j→∞ k=0

which implies
(j) h (xn )
lim c0 = := c0 .
j→∞ f (xn )
The uniqueness of the limit of a sequence and the arbitrariness of n ∈ N assure that h (xn ) = c0 f (xn ),
for each n ∈ N.
If c0 ̸= 0, then
lim |h (xn )| = lim |c0 | |f (xn )| = ∞
n→∞ n→∞
and we have h ∈ A.  
S (k) (k) (k) (k)
If c0 = 0, it follows from (iv) that, if x ∈
/ k,n∈N tn − εn , tn + εn , then fk (x) = 0 for each
k ∈ N and, hence,
∞
P (j)
h (x) = lim ck fk (x) = 0.
j→∞ k=1
 
(r) (r) (r) (r)
If x ∈ tn − εn , tn + εn for some pair of positive integers (n, r), it follows from (iv) that
fr (x) ̸= 0 = fk (x) whenever k ̸= r and, therefore,
∞
(j)
ck fk (x) = lim c(j)
P
h (x) = lim r fr (x) .
j→∞ k=1 j→∞

Thus
h (x)
lim c(j)
r = := cr
j→∞ fr (x)
and
(2.3) h (x) = cr fr (x) .
Since h ̸= 0, there is a pair (n, k) of positive integers such that h (x0 ) ̸= 0 for a certain

(k) (k) (k) (k) (k) (k) (k) (k)
x0 ∈ tn − εn , tn + εn . Thus ck ̸= 0 and, since (2.3) is valid for all x ∈ tn − εn , tn + εn ,
we conclude that      
lim h tn  = lim |ck | fk t(k)
(k) 
 = ∞.
  
n→∞ n n→∞
Thus, h ∈ A and the proof is done. □

As an immediate consequence we get:

Corollary 2.2. The set A is (1, c)-spaceable in (X, dX ).
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 7

Remark 2.3. Let C n [0, ∞), n = 1, 2, . . ., be the linear space of the functions f : [0, ∞) → R whose
derivative of order n is continuous and let C ∞ [0, ∞) be the linear space of the functions f : [0, ∞) → R
having derivatives of any order. Using bump functions, a slight adaptation of the above proof assures
that the sets
 
An = f ∈ C n [0, ∞) ∩ L1 [0, ∞) : lim sup |f (x)| = ∞ , n = 1, 2, . . .
x→∞
and  
∞ ∞
A = f ∈C [0, ∞) ∩ L1 [0, ∞) : lim sup |f (x)| = ∞
x→∞
are pointwise c-lineable.
Our second result shows that the set A is not (ℵ0 , c)-spaceable:
Theorem 2.4. A is not (ℵ0 , c)-spaceable in (X, dX ).
(1) ∞
 
(1)
Proof. Let f ∈ A and let xn be an increasing sequence in [0, ∞), with x1 > 0, such that
n=1
    
(1) n→∞ (1)  n→∞
xn −→ ∞, f xn ̸= 0 for each n = 1, 2, . . . and f x(1)  −→ ∞.

n
h i
(1) (1) (1) (1)
For each n ∈ N, let tn be the midpoint of the interval xn , xn+1 and let εn > 0 be such that
(1) (1)
xn+1 − xn    1
ε(1)
n ≤ and ε(1) f x (1) 
 ≤ n.

n n
2 2

(1)
Let us define xn = xn for each n ∈ N and f1 : [0, ∞) → R by
f (xn ) 
 
(1) (1) (1) (1) (1)


 (1)
x − t n + ε n , if tn − εn ≤ x ≤ tn ,



 εn

 f (x )  
n (1) (1) (1) (1) (1)
f1 (x) = − (1) x − tn − εn , if tn ≤ x ≤ tn + εn ,


 εn

 ∞ h i
(1) (1) (1) (1)
 S
if x ∈ tn − εn , tn + εn .

 0,
 /
n=1
 ∞ n o
(2) (1)
Let xn be the unique increasing ordering of {xn : n ∈ N} ∪ tn : n ∈ N . For each n ∈ N,
n=1 h i
(2) (2) (2) (2)
let tn be the midpoint of the interval xn , xn+1 and let εn > 0 be such that
h i h
(2) (2)
i 1
t(2)
n − ε (2) (2)
n , tn + ε (2)
n ⊂ x n , xn+1 and ε(2)
n |f (xn )| ≤ n .
2
Let f2 : [0, ∞) → R be the function defined by


 f1 (x), if x1 ≤ x ≤ x2 ,


f (xn ) 
 
(2) (2) (2) (2) (2)

x − t + ε , if tn − εn ≤ x ≤ tn and x ∈ / [x1 , x2 ] ,



 (2) n n
 εn


f2 (x) = f (xn )  (2) (2)

(2) (2) (2)

 − (2)
x − tn − ε n , if tn ≤ x ≤ tn + εn and x ∈ / [x1 , x2 ] ,
εn




 ∞ h

 S (2) i
 (2) (2) (2)


 0, if x ∈
/ tn − ε n , tn + ε n ∪ [x1 , x2 ].
n=1
8 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

f1

f (x3 )
f (x1 )
...
x1 x2 x3 x4

f2

f (x6 )
f (x3 )
f (x1 )
...
(2) (2) (2) (2) (2) (2) (2)
x1 x2 x3 x4 x5 x6 x7

f (x12 )
f3

f (x5 )

f (x1 )
...
(3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3)
x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13
..
.

Figure 4. The sequence (fk )∞

k=1 .

(k) ∞
 
In general, for each integer k > 1, let xn be the unique increasing ordering of
n o n o n=1 h i
(k−1) (k−1) (k) (k) (k) (k)
xn : n ∈ N ∪ tn : n ∈ N , tn be the midpoint of the interval xn , xn+1 and εn > 0 be
such that
(k) (k)
xn+1 − xn 1
ε(k)
n ≤ and ε(k)
n |f (xn )| ≤ n .
2 2
Then fk : [0, ∞) → R is the function defined by
if x1 ≤ x ≤ x2 ,

 f1 (x),



 f (xn ) 
(k) (k)

(k) (k) (k)


 (k)
x − t n + εn , if tn − εn ≤ x ≤ tn and x ∈ / [x1 , x2 ] ,
 εn


fk (x) = f (xn )  (k) (k)

(k) (k) (k)

 − (k)
x − t n − ε n , if tn ≤ x ≤ tn + εn and x ∈ / [x1 , x2 ] ,



 ε n
 ∞ h i
 S (k) (k) (k) (k)
if x ∈ tn − εn , tn + εn ∪ [x1 , x2 ].

 0,
 /
n=1
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 9

Given k ∈ N, let nk be the smallest index such that xnk > max {x2 , k}. Let gk : [0, ∞) → R be
defined by
(
fk (x), x ∈ [0, x2 ] ∪ [xnk , ∞) ,
gk (x) =
0, if x ∈ [x2 , xnk ] .

Let
(
f1 (x), if 0 ≤ x ≤ x2 ,
g (x) =
0, if x ≥ x2 .

It is simple to check that {g1 , g2 , . . .} is linearly independent and that

span {gn : n ∈ N} \ {0} ⊂ A.

k→∞
We will prove that gk −→ g in (X, dX ) and, to do this, we just need to prove that

k→∞
∥gk − g∥1 −→ 0

and

P∞ 1 ∥gk − g∥∞,n k→∞

n
−→ 0.
n=1 2 1 + ∥gk − g∥∞,n

It is easy to see that

(
0, if x ≤ xnk ,
(gk − g) (x) =
gk (x) = fk (x) , if x ≥ xnk .

Since xnk → ∞, we obtain

Z ∞
k→∞
∥gk − g∥1 = |gk | −→ 0.
xnk

Since xnk > k, we get ∥gk − g∥∞,n = 0, for every n = 1, . . . , k, and then

P∞ 1 ∥gk − g∥∞,n P∞ 1 ∥gk − g∥∞,n P∞ 1 1

n 1 + ∥g − g∥
= n 1 + ∥g − g∥
≤ n
= k
n=1 2 k ∞,n n=k+1 2 k ∞,n n=k+1 2 2

for each k ∈ N. Therefore

k→∞
dX (gk , g) −→ 0.
10 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

g1

f (x3 )
f (x1 )
1 ...
x1 x2 x3 x4

g2

f (x6 )
f (x4 )
f (x1 )
1 2 ...
(2) (2) (2) (2) (2) (2) (2)
x1 x2 x3 x4 x5 x6 x7

f (x12 ) g3
f (x9 )

f (x1 )
1 2 3 ...
(3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3)
x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13

..
.

f (x1 )
...
x1 x2 x3 x4

Figure 5. The sequence (gk )∞

k=1 and its limit g.

Since g ∈ X \ A, it follows that

span {gn : n ∈ N} \ {0} ⊈ A.
Thus, there is no infinite dimensional closed subspace containing span {gn : n ∈ N} and contained in
A ∪ {0}. Hence A is not (ℵ0 , c)-spaceable in (X, dX ). □
Corollary 2.5. A is not (ℵ0 , c)-spaceable in L1 [0, ∞).
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 11

Corollary 2.6. A is not (ℵ0 , c)-spaceable in C [0, ∞) endowed with the topology of the uniform
convergence on compacta.

We finish this section by proving, by means of a similar argument, the result we mentioned at the
end of the introduction.
S
Theorem 2.7. Let p > 0. The set Lp [0, 1] \ q∈(p,∞) Lq [0, 1] is not (ℵ0 , c)-spaceable.

Proof. Let
∞
S
[1/2, 1) = In ,
n=1

where In = [an , bn ), an = 1−1/2n and bn = 1−1/2n+1 . For each x ∈ In , there is a unique tx,n ∈ [0, 1)
such that
x = (1 − tx,n ) an + tx,n bn .
S
Let f ∈ Lp [0, 1] \ q∈(p,∞) Lq [0, 1].For each n ∈ N, let fn : [0, 1] → R be the function given by

 1,
 if x ∈ [0, 1/2),
fn (x) = f (tx,n ), if x ∈ In ,

0, if x ∈
/ In ∪ [0, 1/2).


It is easy to see that fn ∈ Lp [0, 1], for every n ∈ N. Furthermore, if q > p, denoting the length of
the interval In by |In |, n ∈ N, and making a change of variables, we conclude that
Z 1 Z Z Z
1
|fn |q = |fn |q + |fn |q = + |fn |q
0 [0,1/2) [1/2,1] 2 In
Z 1
1
= + |In | · |f |q = ∞.
2 0
S
Thus, fn ∈ Lp [0, 1] \ q∈(p,∞) Lq [0, 1] for each n ∈ N.
Note that {f1 , f2 , . . .} is linearly independent. In fact, let a1 , a2 , . . . , am be scalars such that

a1 f1 + a2 f2 + · · · + am fm = 0.

For each k = 1, . . . , m, let xk ∈ Ik be such that fk (xk ) ̸= 0. Hence

ak fk (xk ) = (a1 f1 + a2 f2 + · · · + am fm ) (xk ) = 0

and thus ak = 0. It is also simple to check that

[
W \ {0} ⊂ Lp [0, 1] \ Lq [0, 1],
q∈(p,∞)

where W = span {fn : n ∈ N}.

Finally, denoting by χ[0,1/2) the characteristic function at [0, 1/2), we have
Z Z Z 1
fn − χ[0,1/2) p = fn χ[1/2,1) p = p n→∞
|f |p −→ 0,
 
p
|fn | = |In | ·
[0,1] In 0
S
and χ[0,1/2) ∈ W . Since χ[0,1/2) ∈
/ Lp [0, 1] \ q∈(p,∞) Lq [0, 1], the proof is done. □
12 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

3. Sequences of unbounded, continuous and integrable functions

The following remark will be useful in this section:
Remark 3.1. Let f ∈ A and let (xn )∞ n=1 be an increasing sequence in [0, ∞) satisfying (2.1). For
each k ∈ N, let fk as in (2.2). Fixing k ∈ N, let jk be the smallest index such that xjk ≥ k. Let us
define g0 = f and gk : [0, ∞) → R by
(
0, if 0 ≤ x ≤ xjk ,
gk (x) =
fk (x) , if x ≥ xjk .
It is clear that the proof of Theorem 2.1 holds if we replace the sequence (fk )∞
k=0 by the sequence
(gk )∞
k=0 .
Let c0 (X) be the sequence space defined by
n o
r→∞
c0 (X) := (fr )∞
r=1 : fr ∈ X for each r and dX (fr , 0) −→ 0 .

In this space we shall consider the distance dc0 (X) : c0 (X) × c0 (X) → R given by
dc0 (X) ((fr )∞ ∞
r=1 , (gr )r=1 ) = sup dX (fr , gr ) .
r∈N

Now, let A0 ⊂ c0 (X) be the set defined by

n o
r→∞
A0 := (fr )∞r=1 : fr ∈ A for each r and dX (fr , 0) −→ 0 .


In [5, Theorem 3.2] it was proved that A0 is maximal dense-lineable in c0 (X) , dc0 (X) , that is, A0
contains, except by the null vector, a dense vector space with the same dimension of c0 (X). Here we
establish the following result:


Theorem 3.2. A0 is pointwise c-spaceable in c0 (X) , dc0 (X) .
(r) ∞
 
Proof. Let f = (fr )∞
r=1 ∈ A0 . For each r ∈ N, let x j be an increasing sequence in [0, ∞) such
j=1
that     
(r) j→∞ (r) (r)  j→∞
xj −→ ∞, fr xj ̸= 0 for each j = 1, 2, . . . and fr xj  −→ ∞.

 ∞
(r)
Let us consider the function fk,r defined from fr and xj as in (2.2), i.e.
j=1
  
(r)
 fr x n  
(k,r) (k,r) (k,r) (k,r) (k,r)

− if tn − εn ≤ x ≤ tn ,



 (k,r)
x tn + ε n ,



 εn
  
 (r)
fk,r (x) = fr x n  
(k,r) (k,r) (k,r) (k,r) (k,r)

 − (k,r)
x − tn − ε n , if tn ≤ x ≤ tn + εn ,



 εn
∞ h

 i
 S (k,r) (k,r) (k,r) (k,r)
0, if x ∈/ t − ε , t + ε ,

n n n n


n=1
(k,r)
where tn is recursively defined by
(r) (r) (r) (k,r)
xn + xn+1 xn + t n
tn(1,r) = and t(k+1,r)
n = ,
2 2
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 13

(k,r)
εn is such that
(k,r) (k+1,r)
tn − tn
0< ε(k,r)
n <
2
and    1
εn(k,r) fr x(r)  ≤ n.
 
n
2
(r)
For each r ∈ N let jr be the smallest index such that xjr ≥ r. For each k ∈ N, let gk,r : [0, ∞) → R
be defined by
 (r)
 0, if 0 ≤ x ≤ xjr ,
gk,r (x) =
(r)
fk,r (x) , if x ≥ xjr .

∞
Denoting gj = (2−r gj,r )r=1 , we need to show that gj ∈ A0 , for each j. (In fact, the factor 2−r in
the definition of gj is not really necessary. However, as we will see, its use simplifies the arguments.)
r→∞
Fixing j ∈ N, since gj,r ∈ A for each r ∈ N, we only need to prove that dX (2−r gj,r , 0) −→ 0. If
j ≥ 1, then Z ∞
−r
2 gj,r = 2−r r→∞
1
|gj,r | dx ≤ 2−r −→ 0.
0
Given n ∈ N, it is clear that
−r
= max 2−r gj,r (x) : x ∈ [0, n] = 0
 
2 gj,r
∞,n

whenever r ≥ n. Consequently,
−r r→∞
2 gj,r
∞,n
−→ 0
for each n ∈ N. Hence, for r ≥ 1,
∞ 1 ∥2−r gj,r ∥∞,n
dX 2−r gj,r , 0 = ∥2−r gj,r ∥1 +

P
·
n 1 + ∥2−r g ∥
n=1 2 j,r ∞,n
∞
P 1 −r
∥2 gj,r ∥∞,n
≤ 2−r + ·
n=r+1 2
n 1 + ∥2−r gj,r ∥∞,n
∞ 1 r→∞
≤ 2−r + = 2−r+1 −→ 0.
P
2 n
n=r+1

Now, let (an )∞

r=0 ∈ ℓ1 be non null and define
∞ ∞ ∞
 
P −1 P −2 P
h = a0 f + an gn = a0 f1 + 2 an gn,1 , a0 f2 + 2 an gn,2 , . . . .
n=1 n=1 r=1
∞
Following the same lines of the proof of Theorem 2.1, we get hr = a0 fr + 2−r
P
an gn,r ∈ A for each
n=1
r. We also notice that, if r ≥ n, then
∞ ∞ ∞

−r P −r P
|an | ∥gn,r ∥1 ≤ |a0 | ∥fr ∥1 + 2−r
P
∥hr ∥1 = a0 fr + 2

≤ |a0 | ∥fr ∥1 + 2
an gn,r |an | .
n=1 1 n=1 n=1

Since
∥hr ∥∞,n = max {|a0 fr (x)| : x ∈ [0, n]} = |a0 | ∥fr ∥∞,n
whenever r ≥ n, then it follows that
r→∞
∥hr ∥∞,n −→ 0
14 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

for each n ∈ N. Since dX (fr , 0) → 0, we conclude that

∞ 1 ∥fr ∥∞,n r→∞
r→∞ P
∥fr ∥1 −→ 0 and n
· −→ 0.
n=1 2 1 + ∥fr ∥∞,n
If a0 = 0, then
P∞ 1 |a0 | ∥fr ∥∞,n
n
· = 0.
n=1 2 1 + |a0 | ∥fr ∥∞,n
Notice that, if 0 < |a0 | ≤ 1, then
∞ 1
P |a0 | ∥fr ∥∞,n ∞ 1
P |a0 | ∥fr ∥∞,n ∞ 1
P ∥fr ∥∞,n r→∞
n
· ≤ n
· = n
· −→ 0
n=1 2 1 + |a0 | ∥fr ∥∞,n n=1 2 |a0 | + |a0 | ∥fr ∥∞,n n=1 2 1 + ∥fr ∥∞,n
and, if |a0 | > 1, then
P∞ 1 |a0 | ∥fr ∥∞,n ∞ 1
P |a0 | ∥fr ∥∞,n P∞ 1 ∥fr ∥∞,n r→∞
n
· ≤ n
· = |a0 | n
· −→ 0.
n=1 2 1 + |a0 | ∥fr ∥∞,n n=1 2 1 + ∥fr ∥∞,n n=1 2 1 + ∥fr ∥∞,n
Therefore,
P∞ 1 ∥hr ∥∞,n
dX (hr , 0) = ∥hr ∥1 + n
·
n=1 2 1 + ∥hr ∥∞,n
Pr 1 |a0 | ∥fr ∥∞,n P∞ 1 ∥hr ∥∞,n
= ∥hr ∥1 + n
· + n
·
n=1 2 1 + |a0 | ∥fr ∥∞,n n=r+1 2 1 + ∥hr ∥∞,n
∞ ∞ 1 |a0 | ∥fr ∥∞,n ∞ 1
≤ |a0 | ∥fr ∥1 + 2−r
P P P
|an | + n
· +
n=1 n=1 2 1 + |a0 | ∥fr ∥∞,n n=r+1 2n
∞  ∞ 1 |a0 | ∥fr ∥∞,n r→∞
= |a0 | ∥fr ∥1 + 2−r
P P
|an | + 1 + n
· −→ 0.
n=1 n=1 2 1 + |a0 | ∥fr ∥∞,n
Hence h ∈ A0 . In particular, denoting g0 = f , we have that the operator
T : ℓ1 → c0 (X)
∞
(an )∞
P
n=0 7→ an gn ,
n=0
is well-defined, linear, injective and satisfies
T (ℓ1 ) ⊂ A0 ∪ {0} .
Since
c = dim (ℓ1 ) = dim (T (ℓ1 ))
and f = T (1, 0, 0, . . .) ∈ T (ℓ1 ), at this moment, we shown that A0 is pointwise c-lineable. The result
(c0 (X),dc0 (X))
will be fully proved if we show that T (ℓ1 ) ⊂ A0 ∪ {0}.
∞ (c0 (X),dc0 (X) )
Let w = (wr )r=1 ∈ T (ℓ1 ) \ {0}. Notice that all we have to do to complete this proof
is to show that wr ∈ A for each r ∈ N.

∞ (j) ∞
 
Let a(j) j=1 be a sequence in ℓ1 , with a(j) = an , such that
n=0
∞ j→∞
v (j) = a(j)
P
n gn −→ w
n=0


in c0 (X) , dc0 (X) . Hence, denoting
∞
(j)
vr(j) = a0 fr + 2−r a(j)
P
n gn,r ,
n=1
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 15

we have     j→∞
dc0 (X) v (j) , w = sup dX vr(j) , wr −→ 0.
r∈N
In particular,
  j→∞
dX vr(j) , wr −→ 0
(j) j→∞
for each r ∈ N. Therefore, for each r ∈ N, vr −→ wr uniformly over each non-degenerate compact
interval [a, b] ⊂ [0, ∞). Proceeding as in Theorem 2.1, we conclude that wr ∈ A. □


Corollary 3.3. A0 is (1, c)-spaceable in c0 (X) , dc0 (X) .


Theorem 3.4. A0 is not (ℵ0 , c)-spaceable in c0 (X) , dc0 (X) .
Proof. Let us take a sequence f = (fr )∞
r=1 ∈ A0 and consider the functions fk,r defined as in (2.2)
(r) ∞
from fr and a suitable sequence xk in [0, ∞) as in (2.1). For each pair (n, r) of positive
k=1
(r)
integers, let nr be the smallest index such that xnr ≥ rn. Let gn,r : [0, ∞) → R defined by

(r)
 0, if 0 ≤ x ≤ xnr ,
gn,r (x) =
(r)
fn,r (x) , if x ≥ xnr .


Defining ϕn : [0, ∞) → R by


 1 − x, if 0 ≤ x ≤ 1,


(1)
ϕn (x) = 0 if 1 ≤ x ≤ xn1 ,


 −1
 (1)
n gn,1 (x) , if x ≥ xn1 ,
denote gn = ϕn , 2−2n gn,2 , 2−3n gn,3 , . . . , 2−rn gn,r , . . . .


We need to show that gn ∈ A0 , for each n. Since gn,r ∈ A for each n, r ∈ N, we only need to prove
r→∞
that dX (2−rn gn,r , 0) −→ 0. If r > 1, then
Z ∞
−rn
2 gn,r = 2−rn r→∞
|gn,r | ≤ 2−rn −→ 0.

1
0
(r)
Since xnr ≥ rn, we have
−rn
= max 2−rn gn,r (x) : x ∈ [0, m] = 0,
 
2 gn,r
∞,m

whenever m ≤ rn. Consequently,

−rn r→∞
2 gn,r
∞,m
−→ 0
for each m ∈ N. Hence, given r > 1,
∞ 1 ∥2−rn gn,r ∥∞,m
dX 2−rn gn,r , 0 = ∥2−rn gn,r ∥1 +

P
(3.1) ·
m 1 + ∥2−rn g
m=1 2 n,r ∥∞,m
∞ −rn
∥2 gn,r ∥∞,m
1
≤ 2−rn +
P
·
m=rn+1 2
m 1 + ∥2−rn gn,r ∥∞,m
∞ 1 r→∞
≤ 2−rn + = 2−rn+1 −→ 0.
P
2m
m=rn+1
16 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

Now, let us prove that {gn : n ∈ N} is linearly independent. Suppose that

N
P
an gn = 0.
n=1

Since
N N N N N
 
P P P −2n P −3n P −4n
an gn = an ϕn , 2 an gn,2 , 2 an gn,3 , 2 an gn,4 , . . . ,
n=1 n=1 n=1 n=1 n=1
we have
N
2−2n an gn,2 = 0.
P
n=1
Since {gn,2 : n ∈ N} is linearly independent, we obtain an = 0, for each n = 1, . . . , N . Consider
W := span {gn : n ∈ N}. Let us prove that W \ {0} ⊂ A0 . Let
N
N N N N

P P P −2n P −3n P −4n
an gn = an ϕn , 2 an gn,2 , 2 an gn,3 , 2 an gn,4 , . . . ∈ W \ {0} .
n=1 n=1 n=1 n=1 n=1
N N
2−rn an gn,r ∈ A
P P
Following the same lines of the proof of Theorem 2.1, we get an ϕn ∈ A and
n=1 n=1
for each r > 1. We also notice that
N N  N  N
P −rn
2−rn an  ∥gn,r ∥ ≤ 2−rn an  ≤ 2−r
P  P  P
2 an gn,r ≤ |an | .
1
n=1 1 n=1 n=1 n=1

It is obvious that, given m ∈ N,

N  N  
P −rn  P −rn 
2 an gn,r
= max  2 an gn,r (x) : x ∈ [0, m] = 0
n=1 ∞,m n=1

whenever r > m. Consequently,

N
P −rn r→∞
2 an gn,r
−→ 0
n=1 ∞,m
for each m ∈ N. Therefore,

N
−rn
P
2 an gn,r
∞ 1
N  N
P −rn P −rn P n=1 ∞,m
dX 2 an gn,r , 0 = 2 an gn,r +

m
·
N

n=1 n=1 1 m=1 2 P −rn

1 + 2 an gn,r


n=1 ∞,m
N
P −rn
2 an gn,r
∞
N
1

P −rn P n=1 ∞,m
= 2 a g
n n,r
+ m
· N


n=1 1 m=r+1 2 P −rn

1+ 2 an gn,r

n=1 ∞,m
N ∞ 1 r→∞
≤ 2−r
P P
|an | + m
−→ 0.
n=1 m=r+1 2

N
P
Hence, an gn ∈ A0 and
n=1
W \ {0} ⊂ A0 .
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 17

Let us define v := (vr )∞

r=1 , where
(
1 − x, if 0 ≤ x ≤ 1,
v1 (x) :=
0, if x ≥ 1.

and vr ≡ 0 for each r ≥ 2. We have:

r→∞
(i) dX (vr , 0) −→ 0;
(ii) vr ∈ X for each r ∈ N;
(iii) vr ∈
/ A for each r ∈ N.
It follows from (i), (ii) and (iii) that v ∈ c0 (X) \ A0 .
Since, for each n ∈ N,

P∞ 1 ∥ϕn − v1 ∥∞,m
dX (ϕn , v1 ) = ∥ϕn − v1 ∥1 + m
m=1 2 1 + ∥ϕn − v1 ∥∞,m
∞ ∞ 1 ∥gn,1 ∥∞,m
Z
−1 P
=n |gn,1 | + m
m=n+1 2 1 + ∥gn,1 ∥∞,m
(1)
x n1
∞
≤ n−1 + 2−m
P
m=n+1
−1 −n
=n +2
−1
≤ 2n

it follows from (3.1) that

dc0 (X) (gn , v) = sup dX (ϕn , v1 ) , dX 2−2n gn,2 , v2 , dX 2−3n gn,3 , v3 , . . .



≤ sup 2n−1 , dX 2−2n gn,2 , v2 , dX 2−3n gn,3 , v3 , . . .



= sup 2n−1 , dX 2−2n gn,2 , 0 , dX 2−3n gn,3 , 0 , . . .



n→∞
≤ sup 2n−1 , 2−2n+1 , 2−3n+1 , . . . = 2n−1 −→ 0.


This means that v ∈ W . Since v
∈ / A0 ∪ {0}, it follows that W \ {0} ̸⊂ A0 and, hence, A0 is not
(ℵ0 , c)-spaceable in c0 (X) , dc0 (X) . □

For p > 0, consider the sequence space

∞
ℓp (X) := {(fr )∞
r=1 : fr ∈ X for each r and (dX (fr , 0))r=1 ∈ ℓp }

endowed with the following distance:

 P∞
[d (f , g )]p , if 0 < p < 1
 r=1 X r r



dℓp (X) ((fr )∞ ∞
r=1 , (gr )r=1 ) =   1/p
∞
P p
, if 1 ≤ p < ∞.


 [dX (fr , gr )]
r=1


Observe that ℓp (X) , dℓp (X) is a metric vector space. Let Ap ⊂ ℓp (X) be the set defined by
∞
Ap := {(fr )∞
r=1 : fr ∈ A for each r and (dX (fr , 0))r=1 ∈ ℓp } .
18 V.V. FÁVARO, D. PELLEGRINO, A. RAPOSO JR., AND G. RIBEIRO

Let us prove that Ap is non-void. In fact, for each r ∈ N, let fr : [0, ∞) → R given by
 1
 n2 2n (x − n) , if n ≤ x ≤ n + n for some integer n ≥ r,
n2




  
1 1 1


fr (x) = −n2 2n x − n − n , if n + n ≤ x ≤ n + n−1 for some integer n ≥ r,

 n2 n2 n2

∞
 
1


 S
 0,
 if x ∈
/ n, n + n−1 .
n=r n2
 
−r+1 1
Obviously, each fr is continuous and it is simple to check that ∥fr ∥1 = 2 and fr n + n = n
n2
∞
if n ≥ r. Hence fr ∈ A for each r ∈ N. Fixing p > 0, let (ar )r=1 ∈ ℓp such that ar ̸= 0 for each r ∈ N.
If we take

∞
(gr )∞
r=1 = 2
r−1
ar fr r=1 ,
then
P∞ 1 ∥gr ∥∞,n P∞ 1 ∥gr ∥∞,n
dX (gr , 0) = ∥gr ∥1 + n
= |ar | + n
.
n=1 2 1 + ∥gr ∥∞,n n=r+1 2 1 + ∥gr ∥∞,n

Note that
!p p
∞ 1 ∥gr ∥∞,n ∞

P P 1 1
n
≤ n
= .
n=r+1 2 1 + ∥gr ∥∞,n n=r+1 2 2rp

Hence,
!p
∞
P ∞
P 1 ∥gr ∥∞,n P∞ 1 1
n
≤ rp
= p
r=1 n=r+1 2 1 + ∥gr ∥∞,n r=1 2 2 −1

and so
!∞
∞
P 1 ∥gr ∥∞,n
n
∈ ℓp .
n=r+1 2 1 + ∥gr ∥∞,n
r=1

Consequently,
!∞
∞ 1 ∥gr ∥∞,n
(dX (gr , 0))∞ (|ar |)∞
P
r=1 = r=1 + n
∈ ℓp
n=r+1 2 1 + ∥gr ∥∞,n
r=1

and then (gr )∞∈ Ap .

r=1
The proofs of Theorems 3.2 and 3.4 can be adapted, mutatis mutandis, to prove the following:

Theorem 3.5. For each p > 0, the set Ap

(a) is pointwise c-spaceable in ℓp (X) , dℓp (X) , for each p > 0;


(b) is (1, c)-spaceable in ℓp (X) , dℓp (X) ;


(c) is not (ℵ0 , c)-spaceable in ℓp (X) , dℓp (X) .

Acknowledgments. We thank the anonymous reviewers for their insightful comments and
suggestions.
 LINEABILITY AND UNBOUNDED, CONTINUOUS AND INTEGRABLE FUNCTIONS 19

References
[1] R.M. Aron, L. Bernal-González, D. Pellegrino, J.B. Seoane-Sepúlveda, Lineability: the search for linearity in
mathematics. Monographs and Research Notes in Mathematics. CRC Press, Boca Raton, FL, 2016. xix+308 pp.
[2] R.M. Aron, V.I. Gurariy, J.B. Seoane-Sepúlveda, Lineability and spaceability of sets of functions on R, Proc. Amer.
Math. Soc. 133 (2004), 795–803.
[3] G. Botelho, D. Diniz, V. V. Fávaro and D. Pellegrino, Spaceability in Banach and quasi-Banach sequence spaces,
Linear Algebra Appl. 434 (2011), 1255–1260.
[4] G. Botelho, V. V. Fávaro, D. Pellegrino, J. B. Seoane-Sepúlveda, Lp [0, 1] − ∪q>p Lq [0, 1] is spaceable for every
p > 0, Linear Algebra Appl. 436 (2012), 2963–2965.
[5] M.C. Calderón-Moreno, P.J. Gerlach-Mena, J.A. Prado-Bassas, Algebraic structure of continuous, unbounded and
integrable functions, J. Math. Anal. Appl. 470 (2019), 348–359.
[6] D. Cariello, V.V. Fávaro, J. B. Seoane-Sepúlveda, Self-similar functions, fractals and algebraic genericity, Proc.
Amer. Math. Soc. 145 (2017), 4151–4159.
[7] D. Cariello and J. B. Seoane-Sepúlveda, Basic sequences and spaceability in ℓp spaces, J. Funct. Anal. 266 (2014),
3797–3814.
[8] J. Carmona Tapia, J. Fernández-Sánchez, J.B. Seoane-Sepúlveda, W. Trutschnig, Lineability, spaceability, and
latticeability of subsets of C([0, 1]) and Sobolev spaces. Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat.
RACSAM 116 (2022), no. 3, Paper No. 113.
[9] K.C. Ciesielski, J.B. Seoane-Sepúlveda, A century of Sierpiński-Zygmund functions. Rev. R. Acad. Cienc. Exactas
Fís. Nat. Ser. A Mat. RACSAM 113 (2019), no. 4, 3863–3901.
[10] J. A. Conejero, M. Fenoy, M. Murillo-Arcila, J. B. Seoane-Sepúlveda, Lineability within probability theory settings.
Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. RACSAM 111 (2017), 673–684.
[11] J. Falcó, Javier, Karl-G. Grosse-Erdmann, Algebrability of the set of hypercyclic vectors for backward shift
operators. Adv. Math. 366 (2020), 107082, 25 pp.
[12] V.V. Fávaro, D. Pellegrino, D. Tomáz, Lineability and spaceability: a new approach. Bull. Braz. Math. Soc. New
Ser. 51 (2019), 27–46.
[13] D. Papathanasiou, Dense lineability and algebrability of ℓ∞ \ c0 . Proc. Amer. Math. Soc. 150 (2022), no. 3,
991–996.
[14] D. Pellegrino, A. Raposo Jr., Pointwise lineability in sequence spaces, Indag. Math. 32 (2021), 536–546.

Faculdade de Matemática, Universidade Federal de Uberlândia, 38400-902 - Uberlândia, Brazil.

Email address: [email protected]

Departamento de Matemática, Universidade Federal da Paraíba, 58.051-900 - João Pessoa, Brazil.

Email address: [email protected]

Departamento de Matemática, Universidade Federal do Maranhão, 65085-580 - São Luís, Brazil.

Email address: [email protected]

Departamento de Matemática, Universidade Federal da Paraíba, 58.051-900 - João Pessoa, Brazil.

Email address: [email protected]