Linear Algebra and its Applications 578 (2019) 153–158

Contents lists available at ScienceDirect

Linear Algebra and its Applications

www.elsevier.com/locate/laa

Continuity and discontinuity of seminorms on

infinite-dimensional vector spaces
Jacek Chmieliński a , Moshe Goldberg b,∗
a
Department of Mathematics, Pedagogical University of Kraków, Kraków, Poland
b
Department of Mathematics, Technion – Israel Institute of Technology, Haifa,
Israel

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

Article history: Let S be a seminorm on an infinite-dimensional real or

Received 28 March 2019 complex vector space X. Our purpose in this note is to study
Accepted 6 May 2019 the continuity and discontinuity properties of S with respect
Available online 9 May 2019
to certain norm-topologies on X.
Submitted by T.J. Laffey
© 2019 Elsevier Inc. All rights reserved.
MSC:
15A03
47A30
54A10
54C05

Keywords:
Infinite-dimensional vector spaces
Norms
Seminorms
Norm-topologies
Continuity

Throughout this note let X be a vector space over a field F , either R or C. As usual,
a real-valued function N : X → R is said to be a norm on X if, for all x, y ∈ X and
α ∈ F,

* Corresponding author.
E-mail addresses: [email protected] (J. Chmieliński), [email protected] (M. Goldberg).

https://doi.org/10.1016/j.laa.2019.05.003
0024-3795/© 2019 Elsevier Inc. All rights reserved.
154 J. Chmieliński, M. Goldberg / Linear Algebra and its Applications 578 (2019) 153–158

N (x) > 0, x = 0,
N (αx) = |α|N (x),
N (x + y) ≤ N (x) + N (y).

Further, a real-valued function S: X → R is a seminorm on X if, for all x, y ∈ X and

α ∈ F,

S(x) ≥ 0,
S(αx) = |α|S(x),
S(x + y) ≤ S(x) + S(y);

hence, a seminorm is a norm if and only if it is positive-definite.

We recall that if X is finite-dimensional, then all norms on X are equivalent, thus
inducing on X a unique norm-topology. This well-known fact leads to the following result.

Theorem 1 ([1]). Let S be a seminorm on a finite-dimensional vector space X over F ,

either R or C. Then S is continuous with respect to the unique norm-topology on X.

Contrary to the finite-dimensional case, if X is infinite-dimensional then not all norms

on X are equivalent, and consequently we no longer have a unique norm-topology. Indeed,
our main purpose in this note is to explore the questions of continuity and discontinuity
of seminorms when the assumption of finite-dimensionality is removed.
We begin our journey by posting the following lemma.

Lemma. Let X, an infinite-dimensional vector space over F , either R or C, be equipped

with a seminorm S and a norm N . Then:

(a) Continuity of S at the origin implies ubiquitous continuity with respect to the topology
induced by N .
(b) Discontinuity of S at the origin implies ubiquitous discontinuity with respect to the
above mentioned topology.

Proof. The assertion in (a) is well known to many, old and young. Yet, for the reader’s
convenience, we provide a short proof.
Let S be continuous at x = 0 with respect to the topology induced by the given norm;
hence, if {xn }∞
n=1 is any sequence in X, then

N (xn ) → 0 implies S(xn ) → 0 as n → ∞. (1)

Choosing x ∈ X, and taking an arbitrary sequence {yn }∞

n=1 for which N (x − yn ) → 0,
(1) implies that,

S(x − yn ) → 0 as n → ∞.
 J. Chmieliński, M. Goldberg / Linear Algebra and its Applications 578 (2019) 153–158 155

Thus, aided by the familiar inequality

|S(x) − S(y)| ≤ S(x − y), x, y ∈ X, (2)

which follows from the subadditivity of S, we get,

|S(x) − S(yn )| → 0,

and part (a) of the lemma is in the bag.

As for part (b), the assumption that S is discontinuous at x = 0 implies that there
exists a sequence {xn }∞
n=1 in X such that

N (xn ) → 0 but S(xn ) → 0 as n → ∞.

Since S(xn ) → 0 and S is nonnegative, we can provide, by passing if necessary to a

subsequence, a positive constant κ such that

S(xn ) ≥ κ, n = 1, 2, 3, . . . .

Moreover, since N (xn ) → 0, by passing to yet another subsequence if necessary, we may

assume that
1
N (xn ) ≤ , n = 1, 2, 3, . . . .
n2
Now, put

yn = nxn , n = 1, 2, 3, . . . .

Since N (yn ) ≤ 1
n and S(yn ) ≥ nκ, we see that

N (yn ) → 0 but S(yn ) → ∞ as n → ∞.

It follows that at any point x ∈ X, the sequence {zn }∞

n=1 defined by

zn = x + yn , n = 1, 2, 3, . . . ,

satisfies

N (x − zn ) = N (yn ) → 0.

On the other hand,

S(x − zn ) = S(yn ) → ∞,

which forces the discontinuity of S at x. 2

156 J. Chmieliński, M. Goldberg / Linear Algebra and its Applications 578 (2019) 153–158

From the above lemma we may deduce the following result which hardly requires a
proof.

Theorem 2. Let X, an infinite-dimensional vector space over F , either R or C, be

equipped with a seminorm S and a norm N . Then:

(a) S is ubiquitously continuous in X with respect to the topology induced by N if and

only if there exists some point of X at which S is continuous.
(b) Similarly, S is ubiquitously discontinuous in X with respect to the above mentioned
topology if and only if there exists some point of X at which S is discontinuous.

We turn now to the main purpose of this note, which is to show that given a nontrivial
seminorm S on X, there exist two different norms, such that S is ubiquitously continuous
in X with respect to one of them, and ubiquitously discontinuous with respect to the
other.

Theorem 3. Let S = 0 be a seminorm on an infinite-dimensional vector space X over F ,

either R or C. Then:

(a) There exists a norm with respect to which S is ubiquitously continuous in X.

(b) There exists a norm with respect to which S is ubiquitously discontinuous in X.

Proof. (a) Let H be a Hamel basis for X. Then,

 
h
B= : h ∈ H, S(h) > 0 ∪ {h ∈ H : S(h) = 0}
S(h)

is a basis for X as well. So every x in X assumes a unique representation of the form


x= αb (x)b, αb (x) ∈ F , (3)
b∈B

where {b ∈ B : αb (x) = 0} is a finite set. With this representation at hand, we can easily
confirm that the real-valued function

N (x) = |αb (x)|, x ∈ X, (4)
b∈B

is a norm on X. Furthermore, by the construction of B, we infer that

S(b) ∈ {0, 1} for all b ∈ B.

Consequently, for every x in X,

 J. Chmieliński, M. Goldberg / Linear Algebra and its Applications 578 (2019) 153–158 157

 
  
S(x) = S αb (x)b ≤ |αb (x)|S(b) ≤ |αb (x)| = N (x). (5)
b∈B b∈B b∈B

By (2) and (5), therefore,

|S(x) − S(y)| ≤ N (x − y), x, y ∈ X,

and (a) follows.

(b) Consider the nonempty set

A = {x ∈ X : S(x) > 0},

and let us show that

span A = X. (6)

Indeed, assume on the contrary that span A  X. Then there exists an element y ∈ X
such that y ∈
/ span A and S(y) = 0. Now, for every z in A, we have

S(y + z) = S(y + z) + S(y) ≥ S(y + z − y) = S(z) > 0,

whence, both z and y + z belong to A, so y = (y + z) − z ∈ span A, a contradiction.

In view of (6), we may appeal to Corollary 4.2.2 in [3] (compare [2]), which tells us
that A contains a Hamel basis H for X. So by the definition of A, we have S(h) > 0 for
all h ∈ H. With this in mind, we fix a sequence {hn }∞n=1 of distinct elements in H, and
construct a new basis B  for X by replacing hn by

nhn
bn = , n = 1, 2, 3, . . . ,
S(hn )

and leaving all other elements in H unchanged.

Surely,

S(bn ) = n, n = 1, 2, 3, . . . .

Further, modifying the representation in (3) and the norm N in (4) by replacing the
underlying basis B by the new basis B  , we get,

N (bn ) = 1, n = 1, 2, 3, . . . .

Hence, setting

bn
xn = , n = 1, 2, 3, . . . ,
n
158 J. Chmieliński, M. Goldberg / Linear Algebra and its Applications 578 (2019) 153–158

we obtain
1
N (xn ) = → 0,
n
whereas

S(xn ) = 1, n = 1, 2, 3, . . . .

This establishes the discontinuity of S at x = 0; so by Theorem 2(b), S is discontinuous

everywhere, and we are done. 2

Falling back on Theorems 1 and 3(b), we may record now the following simple obser-
vation.

Corollary. Let S = 0 be a seminorm on a vector space X over F , either R or C. Then

S is continuous with respect to every norm on X if and only if X is finite-dimensional.

We conclude this note by recalling that a seminorm S is left-equivalent to a norm N

on X if there exists a constant τ > 0 such that

S(x) ≤ τ N (x) for all x ∈ X.

Therefore, by (5), every seminorm S on X is left-equivalent to some norm on X; for

example, to the norm in (4) (which, of course, depends on S). As indicated in [1], if X
is finite-dimensional, then S is left-equivalent to every norm on X.

References

[1] Moshe Goldberg, Continuity of seminorms on finite-dimensional vector spaces, Linear Algebra Appl.
515 (2017) 175–179.
[2] James D. Halpern, Bases in vector spaces and the axiom of choice, Proc. Amer. Math. Soc. 17 (3)
(1966) 670–673.
[3] Marek Kuczma, An Introduction to the Theory of Functional Equations and Inequalities. Cauchy’s
Equation and Jensen’s Inequality, 2nd edition, Birkhäuser Verlag, Basel, 2009, edited by Attila
Gilányi.