GAZETA MATEMATICĂ

SERIA B
PUBLICAT
, IE LUNARĂ PENTRU TINERET

Fondată ı̂n anul 1895

Anul CXXVII nr. 6-7-8 iunie-iulie-august 2022

ARTICOLE S, I NOTE MATEMATICE

A CONNECTEDNESS-TYPE PROPERTY OF CLOSED
SETS
Radu-Andrei Lecoiu1)
Abstract. In this article we characterize closed sets with a connectedness
property and apply their properties in olympiad problems. Most of the
problems also have different solutions, but the following results lead to
more straightforward ones, using similar ideas throughout their proofs.
Keywords: closed set, connected set, intermediate value property
MSC : 26A03, 26A15

Introduction
For a subset A ⊆ R denote by A the set of all its real limit points,
i.e. the set of real numbers x for which there exists a sequence (xn )n1 of
numbers from A such that lim xn = x. The set A is called closed if A ⊆ A.
n→∞
For instance, any interval [a, b] is closed.
Whenever dealing with continuous functions, it is natural to consider
the set A , where A is a set that is easy to work with. A well-known and very
important result that we will use throughout this paper is that the intersec-
tion of two closed sets is also closed. Moreover, we will use the infimum and
supremum of a set. It is easy to prove that inf A and sup A belong to A ,
provided that they are real numbers.
Next, we will present the statements of the main results and prove them,
followed by some examples of olympiad problems.
The two main properties
Lemma 1 (bounded version). Consider a nonempty bounded set
M ⊆ R with the following two properties:
i) M is closed;
1) Elev, Liceul Teoretic Internat, ional de Informatică, Bucures, ti.
282 Articole s, i note matematice

ii) For each two distinct numbers x, y ∈ M there exists λ ∈ M such

that x < λ < y.
Then M is a closed interval [a, b] of R.
Proof. Consider a = inf M and b = sup M . Since M is bounded, then
a and b are real numbers and a  b. We will show that M = [a, b].
First, since M is closed then a, b ∈ M , then M ⊆ [a, b].
Next, consider an arbitrary c from [a, b] and the following two sets:
U = [a, c] ∩ M, V = [c, b] ∩ M.
We will show that c ∈ M . Note that a ∈ U and b ∈ V , hence U and V
are not empty sets.
If c = a or c = b, then c ∈ M , hence we can assume the contrary.
Consider sup U =: α  c  β := inf V . Since M is closed, then both U
and V are closed, as each of them is the intersection of two closed sets. Then
α ∈ U and β ∈ V .
If c = α, then, since U is closed, c ∈ U , hence c ∈ M . Similarly, if
c = β, then c ∈ M .
Finally, assume that α < c < β. Using property (ii) of M , there exists
λ ∈ M such that α < λ < β.
If c < λ, then c < λ < β, hence λ belongs to V and is smaller than
inf V , which is a contradiction. Similarly, if c > λ we get a contradiction
with the definition of α = sup U . Thus, the only possible case is when λ = c,
so c ∈ M .
Since c was randomly chosen from [a, b], [a, b] ⊆ M ⊆ [a, b] and the
lemma is proved. 
Lemma 2 (unbounded version). Consider an unbounded set M ⊆ R
with the following two properties:
i) M is closed;
ii) For each two distinct numbers x, y ∈ M there exists λ ∈ M such
that x < λ < y.
Then M is of the form [a, ∞), or (−∞, b], or M = R.
Proof. Assume that M is unbounded from above and bounded from
below. Consider inf M = a ∈ M to infer that M ⊆ [a, ∞). Since M is
unbounded from above, there exist a sequence (xn )n1 of numbers from M
such that xn  n.
Consider the sets Mn = [a, xn ] ∩ M , where n  a.
Since Mn ⊆ M and Mn is the intersection of two closed sets, then Mn
has both properties (i) and (ii), hence we can use Lemma 1 to infer that
Mn = [a, xn ], for all n  a. Consider an arbitrary c greater than a and
p = c.
Thus, since xp  p  c, [a, c] ⊆ [a, p] ⊆ [a, xp ] = Mp ⊆ M , hence c ∈ M .
Since c can be any number greater than a, then [a, ∞) ⊆ M ⊆ [a, ∞) and
the conclusion follows.
 R.-A. Lecoiu, A connectedness-type property of closed sets 283

In a similar fashion we can handle the case when M is bounded from

above and unbounded from below.
The third case, when M is unbounded in both directions, follows by a
similar reasoning as above by choosing two sequences of numbers from M ,
say (xn )n1 and (yn )n1 such that xn  n and yn  −n and using again
Lemma 1 to infer that Mn = [yn , xn ] ∩ M = [yn , xn ]. Since xn and yn can
be simultaneously arbitrarily large and arbitrarily small, respectively, and
Mn ⊆ M , the lemma is now proved.

Olympiad Problems
Now, we are ready to show how to use these two properties for contest-
type problems. The main trick for each problem is to choose a set for which
one of the two lemmas from above can be applied.
1. (Mihai Piticari, Romanian Mathematical Olympiad 2005, District
Round) Let f : R → R be a continuous function such that for any a, b ∈ R,
with a < b and f (a) = f (b), there exists some c ∈ (a, b) such that f (a) =
f (b) = f (c). Prove that f is monotonic over R.
Solution. Assume that f is not monotonic. Then it is easy to prove
that there exist a < b < c such that
f (a) < f (b) and f (b) > f (c)
or
f (a) > f (b) and f (b) < f (c).
Assume we are in the first case, as the second one can be treated in a
similar way. Choose λ such that
f (b) < λ < max{f (a), f (c)}
Since f is continuous, it follows from the intermediate value theorem that
there exist x, y such that a < x < b < y < c and f (x) = f (y) = λ. Hence, if
we consider the set
M = {t | t ∈ [x, y] and f (t) = λ},
then x, y ∈ M and M is clearly bounded. From the continuity of f it follows
that M is closed. Moreover, from the problem hypothesis, it is clear that for
each u < v from M there exists w ∈ (u, v) such that w ∈ M .
So we can apply Lemma 1, hence M is a closed interval. Since x, y ∈ M
and x < b < y, then b ∈ M , which is a contradiction since f (b) < λ. The
conclusion follows. 
2. (Marius Cavachi, Romanian Mathematical Olympiad 2020, District
Round, 2020) Determine the continuous functions f : R → R having the
property that, for all x, y ∈ R, there exists t ∈ (0, 1) such that
f ((1 − t)x + ty) = (1 − t)f (x) + tf (y).
284 Articole s, i note matematice

Solution. We will prove that only affine functions work.

First of all, it is easy to see that they do satisfy the condition.
Conversely, consider two real numbers a < b. We will take M as the set
of points t ∈ [a, b] such that the point (t, f (t)) lies on the line determined by
(a, f (a)) and (b, f (b)). Formally speaking, consider the set
     
M = x ∈ [a, b] | f (x) − f (a) (x − b) = f (x) − f (b) (x − a) .
It is clear that M is bounded, a, b ∈ M and M is closed since f is continuous.
Moreover, the hypothesis implies that between any two numbers u and v from
M there exists w = (1−t)u+tv, t ∈ (0, 1), so that f (w) = (1−t)f (u)+tf (v),
which can be rewritten as
f (w) − f (u) f (w) − f (v)
=
w−u w−v
or, equivalently, w ∈ M . Thus M has also property (ii) so, from Lemma 1,
M = [a, b], hence f is affine on the whole interval [a, b]. This shows that f is
affine on any closed interval.
In order to finish the solution, assume that the graph of f is not a line.
Hence, there exist three real numbers x < y < z such that the points of coor-
dinates (x, f (x)), (y, f (y)), and (z, f (z)) are not collinear. This contradicts
the fact that f is affine on [x, z] and concludes the solution. 
3. (Nicolae Bourbăcut, ) Let f : R → R be a function such that for every
x, y ∈ R, x < y, there exists z ∈ (x, y) so that
(y − x)f (z)  (y − z)f (x) + (z − x)f (y).
a) Give an example of a non-convex function f with the given property.
b) Prove that a continuous function with the given property is convex.
Solution. We will say that a function which fulfills the condition from
the statement has property P.
a) Consider the real function

1, if x = 0
f (x) =
0, otherwise
Clearly this function works, since any z ∈ (x, y), with z = 0 satisfies
(y − x)f (z) = (y − x) · 0 = 0  (y − z)f (x) + (z − x)f (y).
Finally, note than f is not convex since f (0) > f (1)+f
2
(−1)
.
b) Suppose that f is not a convex function. Then, there exist three real
numbers x, y, λ, with x < y and λ ∈ (0, 1), such that
f (λx + (1 − λ)y) > λf (x) + (1 − λ)f (y).
Consider the function g : R → R defined as g(t) = f (t) − mt − n, where the
coefficients m and n are taken so that g(x) = g(y) = 0, that is m = f (y)−f
y−x
(x)
 R.-A. Lecoiu, A connectedness-type property of closed sets 285

yf (x)−xf (y)
and n = y−x . Define the set
M = {t ∈ [x, y] | g(t)  0}
and note that x, y ∈ M , since g(x) = g(y) = 0.
Since f is continuous, then g is also continuous. Moreover, it is easy to
check that g has property P, since the affine function is canceled out from
both sides. From the continuity of g, it follows that M is closed. Choose
u < v from M . Since g has property P, then there exists w ∈ (u, v) such
that
(v − u)g(w)  (v − w)g(u) + (w − u)g(v).
Since v > u and the right hand side term is nonpositive, then w ∈ M , which
shows that M has property (ii). As M is clearly bounded, we can apply
Lemma 1 to infer that M = [x, y].
So, for every z ∈ (x, y), f (z)  mz + n. Substituting the values of m, n
in this inequality easily leads to
(y − x)f (z)  (y − z)f (x) + (z − x)f (y).
In particular, for z = λx + (1 − λ)y this yields
f (λx + (1 − λ)y)  λf (x) + (1 − λ)f (y),
which is a contradiction with the initial assumption on x, y and λ. This
proves that f must be convex on R. 
4. (after a problem of Dan Marinescu) A function f : R → R is said
to have property P if for every a, b ∈ R there exists c ∈ (a, b) such that
f (c) ∈ {f (a), f (b)}, where the brackets can also denote a multiset.
a) Give an example of a non-constant function possessing P.
b) If f has P and is continuous, prove that it is constant.
Solution. a) The Dirichlet function

1, if x ∈ Q
f (x) =
0, if x ∈ R \ Q
is not constant and property P follows from the density of both Q and R \ Q
in R.
b) Assume that f is not constant and pick real numbers a < b, such
that f (a) = f (b). Define the set
M = {x ∈ [a, b] | f (x) = f (a) or f (x) = f (b)}.
It is clear that a, b ∈ M . Choose u < v from M . Since f has property P,
then there exists w ∈ (u, v) such that f (w) = f (u) or f (w) = f (v). Since
u, v ∈ M then
f (u), f (v) ∈ {f (a), f (b)} ⇒ f (w) ∈ {f (a), f (b)}
hence w ∈ M .
286 Articole s, i note matematice

Next, we proceed to show that M is closed. Consider a sequence (xn )n1

of real numbers from M such that lim xn = x. Since f is continuous,
n→∞
then lim f (xn ) = f (x). Since f (xn ) can take only two values, that is f (a)
n→∞
and f (b), and the sequence is convergent, then it must become eventually
stationary. Thus, the limit f (x) is either f (a) or f (b), therefore x ∈ M .
Hence, we can apply Lemma 1 to infer that M = [a, b] i.e.
 
f [a, b] ⊆ {f (a), f (b)}.
Since f is continuous, the image of f is an interval. From the last relation
it is clear that the interval must be reduced to one point, which means that
f (a) = f (b), a contradiction.
Further study
We invite the reader to solve the following problems with the help of
the aforementioned properties and ideas.
Problem 1. (Nicolae Bourbăcut, , Romanian Mathematical Olympiad
2012, District Round) We will say that a function f : R → R has property
F if for each real number a there exists b < a such that f (x)  f (a), for all
x ∈ (b, a).
a) Give an example of a function with property F that is not monotonic
on R.
b) Prove that a continuous function that has property F is nondecrea-
sing.
Problem 2. (Radu Gologan) Let f : R → R be a continuous function
such that, on each non degenerated interval I, the function reaches its ma-
ximum or its minimum in an interior point of I. Prove that f is a constant.

Problem 3. (Dorin Andrica and Mihai Piticari, Romanian Mathemat-

ical Olympiad 2007) A P −function is a differentiable function f : R → R
with a continuous derivative on R, such that f (x + f  (x)) = f (x) for all x in
R.
a) Prove that the derivative of a P −function has at least one zero.
b) Provide an example of a non-constant P −function.
c) Prove that a P −function whose derivative has at least two distinct
zeros is constant.

References
[1] The Romanian Mathematical Competitions Collection 2004-2020.