Annals of the University of Bucharest (mathematical series)

(Analele Universit at ii Bucuresti. Matematic a)

1 (LIX) (2010), 245262
Unicity properties and algebraic properties for the
solutions of the functional equation
f f + af + b1
R
= 0 (II)
Traian Cristian Gdea
Communicated by Ion Chit escu
Abstract - Throughout this paper we shall deal with the functional equa-
tion in the title, for real a, b and b = 0. This functional equation was
completely solved in a previous paper. Namely we found all continuous so-
lutions of the aforementioned functional equation. In this paper we shall
establish bijections between the set of solutions and some special sets of
functions.
Key words and phrases : iterative functional equation, continuous
solution, homeomorphism.
Mathematics Subject Classication (2000) : 39B12, 30D05.
1. Introduction
We consider the functional equation
f f(x) +af(x) +bx = 0, x R (1.1)
where the signs of a and b are taken according to the conventions of [4].
In the rst part of this paper (see [4]) we proved the existence and we
presented the general form of the solutions of the aforementioned equa-
tion. Namely, the aim of the rst part of this paper was to establish what
conditions can guarantee the uniqueness of the continuous solution of this
functional equation. More precisely, if two solutions coincide on an inter-
val under some conditions, then they coincide everywhere. In the rst part
we established what conditions must fulll this interval in each case. Let
= a
2
4b and let r
1
, r
2
be the solutions of the equation x
2
+ax +b = 0.
We proved the following theorem:
Theorem 1.1. (see [4, Theorem 2.1]) Let us consider the functional equa-
tion (1.1) in case > 0. Then the following statements hold.
(i) If 1 < r
1
< r
2
and two continuous solutions coincide on
I = [a

, b

], I (0, ) and
b

r
2
, then they coincide on (0, ).
245
246 Traian Cristian Gdea
A similar result holds if I (, 0).
If I = [0, a

] the solutions coincide on (0, ). A similar result holds if

I = [a

, 0].
(ii) If r
2
< r
1
< 1 and two continuous solutions coincide on
I = [a

, b

], I (0, ) and
b

r
2
2
(also if I (, 0) and I = [a

, b

] and
a

r
2
2
), then they coincide on R.
If 0 I the two solutions coincide on R.
(iii) If r
1
< 1 < r
2
, let us consider two solutions f and g such that
f(0) = 0, g(0) = 0 and there exists a

> 0 such that fand g coincide on

[0, a

]. Then fand g coincide on R. We have a similar result for the case

[a

, 0], a

< 0.
The aim of the paper is to prove the existence of several bijections be-
tween the set of solutions of equation (1.1) and some special sets of functions.
We will consider three cases: 1 < r
1
< r
2
, r
2
< r
1
< 1 and 0 < r
1
< 1 < r
2
.
2. Algebraic properties of the solutions
A. Case 1 < r
1
< r
2
.
The set of continuous solutions of equation (1.1) (see [4, Theorem 1.2])
will be denoted in the sequel by S. From the Calibration Theorem (see [4]),
we have for a solution f of equation (1.1) the inequality
r
1
(x y) f(x) f(y) r
2
(x y), x, y R. (2.1)
We notice that the set of continuous solutions of equation (1.1) on (0, ),
denoted by S
+
, depends upon three parameters: x
0
> 0, x
1
[x
0
r
1
, x
0
r
2
]
and f
0
: [x
0
, x
1
] [x
1
, x
2
]. Here (x
n
)
nZ
is the sequence described in [4,
Theorem 1.2], and f
0
is continuous and bijective which fullls the inequality
r
1
(x y) f
0
(x) f
0
(y) r
2
(x y) for x > y.
Under these assumptions, it follows from Theorem 1.1 that, starting
with (x
0
, x
1
, f
0
) under the previous conditions, they uniquely determine the
solution on (0, ), because two solutions which coincide on [x
0
, x
1
] coincide
on (0, ), according to Theorem 1.1.
Notation 2.1. In the sequel we will use the following notation.
(i) V (x
1
) = {f
0
: [x
0
, x
1
] [x
1
, x
2
] | x
1
[r
1
x
0
, r
2
x
0
], x
2
= ax
1
bx
0
,
f
0
continuous, bijective and fullls (2.1) for x > y, x, y [x
0
, x
1
]}.
(ii) V =

x
1
[x
0
r
1
,x
0
,r
2
]
V (x
1
).
Unicity properties and algebraic properties for the solutions. . . 247
(iii) S
+
= {f : (0, ) (0, ) | f is a continuous solution of equation
(1.1) on (0, )}.
(iv) S

= {f : (, 0) (, 0) | f is a continuous solution of equation

(1.1) on (0, )}.
(v) B =
_
h : [0, 1] [0, 1] | h(0) = 0, h is continuous, increasing,
h(1)
_
r
1
r
2
, 1
_
;
h(x) h(y)
x y

_
r
1
r
2
, 1
_
for x, y [0, 1], x > y
_
.
Lemma 2.1. There exists a bijection H : V S
+
.
Proof. Let f
0
V . There exists x
1
[x
0
r
1
, x
0
r
2
] such that f
0
V (x
1
).
We dene H(f
0
) = f, where f is the only function that fullls the conditions
f(x
0
) = x
1
, f S
+
and f(x) = f
0
(x) for all x [x
0
, x
1
] (see Theorem 1.1).
Then it follows that f fullls (2.1), for all x > y 0. Hence H is well
dened.
a) H is injective. Let f
1
, f
2
S
+
be such that H
_
f
1
0
_
= f
1
; H
_
f
2
0
_
= f
2
and f
1
= f
2
. Let x
0
> 0 and x
1
[x
0
r
1
, x
0
r
2
] be such that f
1
(x
0
) =
f
2
(x
0
) = x
1
, f
1
0
= f
1
|
[x
0
,x
1
]
; f
2
0
= f
2
|
[x
2
,x
1
]
. Because f
1
|
[x
0
,x
1
]
= f
2
|
[x
0
,x
1
]
, it
follows that f
1
0
= f
2
0
, so H is injective.
b) H is surjective. Let f S
+
. We look for f
0
V which fullls the
condition H(f
0
) = f. Consider x
0
> 0; x
1
= f(x
0
); f
0
= f|
[x
0
,x
1
]
; x
2
=
f(x
1
); f
0
: [x
0
, x
1
] [x
1
, x
2
]. Let us prove that f
0
V (x
1
). Because f S
we have r
1
x
0
f(x
0
) r
2
x
0
(see [4, Lemma 1.1]). So x
1
[r
1
x
0
, r
2
x
0
].
Since f(f(x
0
)) = a f(x
0
) bx
0
, we have x
2
= ax
1
bx
0
. Because f
fullls (2.1), it follows that f
0
fullls (2.1) (this results from the fact that
f
0
= f|
[x
0
,x
1
]
). Hence H is surjective and it follows that H is bijective. 2
Theorem 2.1. There exists a bijection F : B S
+
.
Proof. Firstly we shall prove that there exists T : B V, T bijective. Let
x
0
> 0 and x
1
[r
1
x
0
, r
2
x
0
], x
2
= ax
1
bx
0
.
Let h B. If x [x
0
, x
1
], there exists t [0, 1] such that t =
x x
0
x
1
x
2
and conversely. We dene now with the aid of h:
: [x
0
, x
1
] [x
1
, x
2
] , (x) =
x
2
x
1
h(1)
h
_
x x
0
x
1
x
0
_
x
1
.
We can take x
1
= x
0
h(1)r
2
b
h(1)r
2
a + 1
(namely, we shall prove that for this
x
1
one has x
1
[r
1
x
0
, r
2
x
0
]). According to the denition of the set B,
h(1)
_
r
1
r
2
, 1
_
.
248 Traian Cristian Gdea
Let us consider g :
_
r
1
r
2
, 1
_
R, given by g(x) = x
0
(xr
2
b)
xr
2
a + 1
. Then
g

(x) = x
0
(r
2
(xr
2
a + 1) (xr
2
b) r
2
)
(xr
2
a + 1)
2
= x
0
ar
2
+r
2
+br
2
(xr
2
a + 1)
2
=
= x
0
r
2
b a + 1
(xr
2
a + 1)
2
=
x
0
r
2
(r
1
1) (r
2
1)
(xr
2
a + 1)
2
.
It follows that g is increasing, so g(x)
_
g
_
r
1
r
2
_
, g(1)
_
. We have
g
_
r
1
r
2
_
=
x
0
(r
1
b)
r
1
r
1
r
2
+ 1
=
x
0
r
1
(1 r
2
)
1 r
2
= x
0
r
1
.
Similarly we can prove that g(1) = x
0
r
2
. Hence x
1
[x
0
r
1
, x
0
r
2
].
We dene T : B V by T(h) = .
a) We shall prove that T is well dened; more precisely, starting with
h B and dening as above, let us prove that V (x
1
). Obviously
(x
0
) = x
1
and (x
1
) = x
2
; because h is increasing it follows that is
increasing. Hence is surjective and increasing, therefore is bijective.
Let us prove that
(y) (x)
y x
[r
1
, r
2
], for y > x. By using the relations
_
x = x
0
+t
1
(x
1
x
0
)
y = x
0
+t
2
(x
1
x
0
) ,
we deduce that y x = (t
2
t
1
) (x
1
x
0
). Therefore y > x if and only if
t
2
> t
1
. According to the denition of x
1
it follows that
x
1
h(1)r
2
+x
1
(a + 1) = x
0
h(1)r
2
x
0
b.
One easily deduces the relation
h(1) =
1
r
2
_
x
2
x
1
x
1
x
0
_
. (2.2)
But
h(t
1
) =
(x) h(1)
x
2
x
1
x
1

h(1)
x
2
x
1
and
h(t
2
) =
(y) h(1)
x
2
x
1
x
1

h(1)
x
2
x
1
.
It follows that
h(t
2
) h(t
1
)
t
2
t
1
=
h(1)
(x
2
x
1
)

_
(y) (x)
y x
_

_
y x
t
2
t
1
_
=
Unicity properties and algebraic properties for the solutions. . . 249
=
(y) (x)
y x

h(1)
x
2
x
1
(x
1
x
0
) =
=
_
(y) (x)
y x
_

1
r
2

x
2
x
1
x
1
x
0

x
1
x
0
x
2
x
1
=
1
r
2
_
(y) (x)
y x
_
. (2.3)
Hence
(y) (x)
y x
= r
2

_
h(t
2
) h(t
1
)
t
2
t
1
_
.
Since
h(t
2
) h(t
1
)
t
2
t
1

_
r
1
r
2
, 1
_
, one has
(y) (x)
y x
[r
1
, r
2
] and therefore
V (x
1
). Hence T is well dened. We shall prove that C = [r
1
, r
2
], where
C =
_
x
1
[x
0
r
1
,x
0
r
2
]
_
V (x
1
)
_
(y) (x)
y x

y > x, x, y [x
0
, x
1
]
_
.
Obviously C [r
1
, r
2
]. On the other hand, one has C
1
C, where
C
1
=
_
x
1
[x
0
r
1
, x
0
r
2
]
_
W (x
1
)
D(x
1
, ) with
D(x
1
, ) = {
(y) (x)
y x
| y > x, x, y [x
0
, x
1
]}
and W (x
1
) = V (x
1
)

{ : [0, 1] R | (x) = px +q}.
We shall prove that C
1
= [r
1
, r
2
] . Let [x
0
r
1
, x
0
r
2
] be arbitrarily
taken. We look for linear having the form (x) = px +q such that p = .
Hence (x
0
) = x
1
, (x
1
) = x
2
if and only if px
0
+q = x
1
and px
1
+q = x
2
.
Consequently p (x
1
x
0
) = x
2
x
1
, that is
p =
x
2
x
1
x
1
x
0
=
ax
1
bx
0
x
1
x
1
x
0
=
(a 1)x
1
bx
0
x
1
x
0
,
with x
1
[x
0
r
1
, x
0
r
2
].
We write g
1
(x) =
(a 1)x bx
0
x x
0
, g
1
: [x
0
r
1
, x
0
r
2
] R. Then
g

1
(x) =
(a 1) (x x
0
) ((a 1)x bx
0
)
(x x
0
)
2
=
x
0
(b a + 1)
(x x
0
)
2
0. (2.4)
It follows that g
1
is increasing (g
1
). We have
g
1
(x
0
r
1
) =
x
0
r
2
1
x
0
r
1
x
0
r
1
x
0
= r
1
and similarly g
1
(x
0
r
2
) = r
2
. Because g
1
is continuous and increasing, it
follows that g
1
([x
0
r
1
, x
0
r
2
]) = [r
1
, r
2
]. Hence, for all [x
0
r
1
, x
0
r
2
] there
exist p and (x) = px +q, such that p = .
250 Traian Cristian Gdea
Consequently C
1
= [r
1
, r
2
] and C = [r
1
, r
2
] .
b) We shall prove that T is injective. Let h
1
, h
2
B, h
1
= h
2
. Let us
prove that T (h
1
) = T (h
2
), T (h
1
) =
1
and T (h
2
) =
2
, where
1
,
2
V .
There exist x
1
and x

1
such that
1
V (x
1
) and
2
V (x

1
).
1) If x
1
= x

1
we have
1
(x
0
) = x
1
and
2
(x
0
) = x

1
with x
1
= x

1
, hence
we have
1
=
2
.
2) If
1
,
2
V (x
1
) one has x
1
= x

1
and so x
2
= x

2
. On the other hand,
h
1
(1) =
1
r
2
_
x
2
x
1
x
1
x
0
_
=
1
r
2
_
x

2
x

1
x

1
x
0
_
= h
2
(1). Because h
1
= h
2
, there
exists (0, 1) such that h
1
() = h
2
(). We choose x

= x
0
+(x
1
x
0
),
x

(x
0
, x
1
). It follows that

1
(x

) =
x
2
x
1
h
1
(1)
h
1
_
x

x
0
x
1
x
0
_
+x
1
=
x
2
x
1
h
1
(1)
h
1
() +x
1
.
Similarly

2
(x

) =
_
x

2
x

1
h
2
(1)
_
h
2
() +x

1
.
Since x
1
= x

1
, x
2
= x

2
and h
1
() = h
2
(), it follows that
1
(x

) =
2
(x

).
Hence
1
=
2
, which implies that T is injective.
c) We shall prove that T is surjective. Let V ; there exists x
1
such
that V (x
1
). Let us prove that there exists h B such that T(h) = .
Because V (x
1
), it follows that
x x
0
x
1
x
0
[0, 1], for all x [x
0
, x
1
]. We
denote t =
x x
0
x
1
x
0
; t [0, 1]. Let us dene
h(t) =
_
(x
0
+t(x
1
x
0
)) x
1
x
2
x
1
__
x
2
x
1
x
1
x
0
_

1
r
2
for any t [0, 1].
Let us prove that h B. Obviously h is continuous, h is increasing,
h(0) = 0 and h(1) =
1
r
2
_
x
2
x
1
x
1
x
0
_
. Since x
2
= ax
1
bx
0
one has
h(1) =
1
r
2
_
(a 1)x
1
bx
0
x
1
x
0
_
. Due to (2.2), if x
1
[r
1
x
0
, r
2
x
0
], one has
1
r
2
_
x
2
x
1
x
1
x
0
_

_
r
1
r
2
, 1
_
, for all x
1
[r
1
x
0
, r
2
x
0
]. Hence h(1)
_
r
1
r
2
, 1
_
.
Thus it follows that h(1) is well dened. Let 0 t
1
< t
2
1 and write
_
x = x
0
+t
1
(x
1
x
0
)
y = x
0
+t
2
(x
1
x
0
) .
Due to (2.3) we have:
(y) (x)
y x
= r
2
_
h(t
2
) h(t
1
)
t
2
t
1
_
.
Unicity properties and algebraic properties for the solutions. . . 251
Consequently
(y) (x)
y x
[r
1
, r
2
], which implies
h(t
2
) h(t
1
)
t
2
t
1

_
r
1
r
2
, 1
_
and thus it follows that h B. Let us prove now that T(h) = . We have
T(h)(t) =
_
x
2
x
1
h(1)
__
(x) x
1
x
2
x
1
__
x
2
x
1
x
1
x
0
_

1
r
2
+x
1
=
=
_
(x) x
1
h(1)
__
x
2
x
1
x
1
x
0
_

1
r
2
+x
1
.
Since h(1) =
x
2
x
1
(x
1
x
0
) r
2
, it follows that T(h)(t) = (x). We deduce that
T(h) = , which means that T is surjective. In conclusion, T is bijective.
We dene F = H T; H : V S
+
, T : B V and so H T : B S
+
.
Since H and T are bijective, it follows that F is also bijective. 2
Remark 2.1. In the same way it can be proved that there exists a bijection
F
2
: B S

.
Corollary 2.1. There exists a bijection F : B B S.
Proof. According to Theorem 2.1, there exists F
1
: B S
+
bijective and
there exists F
2
: B S

bijective. We now dene F as follows:

F(h)(x) =
_
F
1
(h
1
)(x), x 0
F
2
(h
2
)(x), x < 0,
where h B B, h = (h
1
, h
2
).
a) We shall prove that F is injective. Since h = h

, it follows that h
1
= h

1
or h
2
= h

2
. We deduce that F
1
(h
1
) = F
1
(h

1
) or F
2
(h
2
) = F
2
(h

2
), because
F
1
, F
2
are injective. Hence F(h) = F(h

).
b) We shall prove that F is surjective.
Let f S. Then f(x) =
_
f
1
(x), x 0
f
2
(x), x < 0.
According to Theorem 2.1, there exists F
1
: B S
+
which satises
the relation F
1
(h
1
) = f
1
and F
2
: B S

which satises the relation

F
2
(h
2
) = f
2
(recall that F
1
, F
2
are surjective). Then
F(h)(x) =
_
F
1
(h
1
)(x), x 0
F
2
(h
2
)(x), x < 0,
which shows that F(h) = f. 2
B. Case r
2
< r
1
< 1.
The set of solutions of equation (1.1) is given by [4, Theorem 1.3]. From
the Calibration Theorem, we have for a solution f of equation (1.1) the
inequality
r
1
(x y) f(x) f(y) r
2
(x y), x, y R. (2.5)
252 Traian Cristian Gdea
Let us observe that the set of solutions on R depends upon three param-
eters: x
0
> 0, x
1
[x
0
r
2
, x
0
r
1
], f
0
: [x
0
, x
2
] [x
3
, x
1
], where (x
n
)
nZ
is the one of [4, Theorem 1.3] and f
0
is continuous, bijective and satises
(2.5) for x, y [x
0
, x
2
]. From these statements it follows from Theorem 1.1
that starting with (x
0
, x
1
, f
0
) under previous conditions, these parameters
uniquely determine the solution on R, because two solutions which coincide
on [x
0
, x
2
] coincide on R, due to Theorem 1.1.
Notation 2.2. In the sequel we will use the following notation.
(i) V (x
1
) is the set of functions f
0
: [x
0
, x
2
] [x
3
, x
1
] with x
1

[r
2
x
0
, r
1
x
0
], where x
n+2
= ax
n+1
bx
n
for n = 0, 1 and f
0
is continu-
ous, bijective and satises (2.5).
(ii) V =

x
1
[x
0
r
2
,x
0
r
1
]
V (x
1
).
(iii) S = {f : R R | f is a solution of equation (1.1)}.
(iv) B has the same meaning like in the case A.
Lemma 2.2. There exists a bijection H : V S.
Proof. Let f
0
V . There exists x
1
[x
0
r
2
, x
0
r
1
], such that f
0
V (x
1
).
We dene H(f
0
) = f, where f is the unique function which fullls the
conditions f(x
0
) = x
1
, f S, f(x) = f
0
(x) for all x [x
0
, x
2
]. Then it will
follow that f satises (2.5), for all x > y ; x, y R, so H is well dened.
a) We prove that H is injective.
Indeed, let f
1
0
, f
2
0
S be such that H(f
1
0
) = H(f
2
0
) on R. Hence
H
_
f
1
0
_
= f
1
; H
_
f
2
0
_
= f
2
. Let x
0
> 0, x
1
[x
0
r
2
, x
0
r
1
], x
2
= ax
1
bx
0
.
Then f
1
(x
0
) = f
2
(x
0
) = x
1
; f
1
0
= f
1
|
[x
0
,x
2
]
, f
2
0
= f
2
|
[x
0
,x
2
]
. Since f
1
|
[x
0
,x
2
]
=
f
2
|
[x
0
,x
2
]
, we have f
1
0
= f
2
0
and it follows that H is injective.
b) We prove that H is surjective.
Indeed, let f S. We look for f
0
V with H(f
0
) = f. Let x
0
> 0. Write
x
1
= f(x
0
); x
2
= f(x
1
); x
3
= f(x
2
). Dene f
0
: [x
0
, x
2
] [x
3
, x
1
], given
via f
0
(x) = f(x). Let us prove that it holds f
0
V (x
1
). Since f S, it
follows that r
2
x
0
f(x
0
) r
1
x
0
(see [4, Lemma 1.2]), so x
1
[r
2
x
0
, r
1
x
0
].
Because f(f(x
0
)) = a f(x
0
) bx
0
, we have x
2
= ax
1
bx
0
and because
f(f(x
1
)) = a f(x
1
) bx
1
, we have x
3
= ax
2
bx
1
. Since f satises
(2.5) and f
0
= f|
[x
0
,x
2
]
, it follows that f
0
satises (2.5). Consequently H is
surjective, therefore it is bijective. 2
Theorem 2.2. There exists a bijection F : B S.
Proof. Firstly we shall prove that there exists a bijection T : B V ,
T. Let x
0
> 0 and x
1
[r
2
x
0
, r
1
x
0
]. Consequently x
2
, x
3
are given by the
relations x
2
+ax
1
+bx
0
= 0; x
3
+ax
2
+bx
1
= 0.
Unicity properties and algebraic properties for the solutions. . . 253
Let h B. We now try to dene : [x
0
, x
2
] [x
3
, x
1
] . We notice that
x [x
0
, x
2
] if and only if
x x
0
x
2
x
0
[0, 1]. Therefore there exists t [0, 1]
such that t =
x x
0
x
2
x
0
. Hence t [0, 1] if and only if x [x
0
, x
2
]. We can
dene now as follows:
: [x
0
, x
2
] [x
3
, x
1
] , (x) =
x
3
x
1
h(1)
h
_
x x
0
x
2
x
0
_
+x
1
(x
0
) = x
1
; (x
2
) = x
3
.
Since x
3
x
1
< 0 and h is increasing, it follows that .
We can take x
1
=
x
0
(r
2
h(1)(1 +b) +ab)
1 a
2
+b ar
2
h(1)
(namely, we shall prove that
for this x
1
one has x
1
[r
2
x
0
, r
1
x
0
]).
According to the denition of the set B, it holds h(1)
_
r
1
r
2
, 1
_
.
We now consider g :
_
r
1
r
2
, 1
_
R, given by g(x) =
x
0
(r
2
(1 +b)x +ab)
ar
2
x + 1 a
2
+b
.
Taking into account that
1 a
2
+b
ar
2
> 1, it will follow that g is well dened.
It is well known that the function x
mx +n
px +q
is strictly decreasing if and
only if

m n
p q

< 0. We have

(1 +b)r
2
ab
ar
2
1 a
2
+b

r
2
+r
1
r
2
2
r
2
1
r
2
r
2
2
r
1
r
2
2
+r
1
r
2
1 r
2
1
r
2
2
r
1
r
2

=
=

r
2
+r
1
r
2
2
r
2
r
2
1
r
2
r
2
2
+r
1
r
2
1 r
2
1

= r
2
_
1 r
2
1
_

1 +r
1
r
2
r
2
r
1
+r
2
1

=
= r
2
_
1 r
2
1
_ _
1 r
2
2
_
< 0.
Therefore it follows that g is decreasing. Then, for x
_
r
1
r
2
, 1
_
, it will follow
that g(x)
_
g
_
r
1
r
2
_
, g(1)
_
. But g
_
r
1
r
2
_
= x
0
r
1
and g(1) = x
0
r
2
, so
x
1
[x
0
r
2
, x
0
r
1
].
Let us dene T(h) = ; T : B V .
a) We shall prove that T is well dened. More precisely, starting with
h B and dening as previously, let us show that V (x
1
). Obviously
(x
0
) = x
1
and (x
2
) = x
3
. Since h is increasing it follows that is de-
creasing. Therefore is surjective and decreasing, i.e. is bijective.
254 Traian Cristian Gdea
Let us prove that
(y) (x)
x y
[r
1
, r
2
], for x > y. By the equalities
_
x = x
0
+t
1
(x
2
x
0
)
y = x
0
+t
2
(x
2
x
0
) ,
we get x y = (t
1
t
2
) (x
2
x
0
) . Hence x > y if and only if t
1
> t
2
. We
have
(y) (x)
x y
=
x
3
x
1
h(1)

_
h(t
2
) h(t
1
)
t
1
t
2
_

_
t
1
t
2
x y
_
=
=
x
1
x
3
h(1)

_
h(t
1
) h(t
2
)
t
1
t
2
_

_
1
x
2
x
0
_
.
According to the denition of x
1
, it follows that
x
0
r
2
h(1)(1 +b) +abx
0
= x
1
_
1 a
2
+b
_
ax
1
r
2
h(1).
We deduce
h(1) =
x
1
_
1 a
2
+b
_
abx
0
r
2
(x
0
(1 +b) +ax
1
)
=
1
r
2
_
x
1
x
3
x
0
x
2
_
. (2.6)
Consequently
(y) (x)
x y
=
_
h(t
1
) h(t
2
)
t
1
t
2
_
r
2
. (2.7)
Since
h(t
1
) h(t
2
)
t
1
t
2

_
r
1
r
2
, 1
_
, it follows that
(y) (x)
x y
[r
1
, r
2
].
Hence V (x
1
) and then it follows that T is well dened.
We shall prove that if
C =
_
x
1
[x
0
r
2
,x
0
r
1
]
_
V (x
1
)
_
(y) (x)
y x

y > x; x, y [x
0
, x
2
]
_
,
then C = [r
2
, r
1
]. It obviously holds C [r
2
, r
1
]. On the other hand, one
has C
1
C, where
C
1
=
_
x
1
[x
0
r
2
, x
0
r
1
]
_
W (x
1
)
D(x
1
, ) with
D(x
1
, ) = {
(y) (x)
y x
| y > x, x, y [x
0
, x
2
]} and
W (x
1
) = V (x
1
)

{ : [0, 1] R | (x) = px +q}.

We shall prove that C
1
= [r
2
, r
1
]. Let [r
2
x
0
, r
1
x
0
]. We look for
linear ((x) = px + q) such that p = . So (x
0
) = x
1
, (x
2
) = x
3
, that is
Unicity properties and algebraic properties for the solutions. . . 255
px
0
+q = x
1
, px
2
+q = x
3
. These equalities imply that p (x
2
x
0
) = x
3
x
1
.
The latter relation is equivalent to
p =
x
3
x
1
x
2
x
0
=
x
1
x
3
x
0
x
2
=
x
1
x
1
(a
2
b) abx
0
x
0
+ax
1
+bx
0
=
x
1
_
1 a
2
+b
_
abx
0
ax
1
+ (b + 1)x
0
,
x
1
[r
2
x
0
, r
1
x
0
] .
We consider g
1
: [x
0
r
2
, x
0
r
1
] R, given by g
1
(x) =
_
1 a
2
+b
_
x abx
0
ax + (b + 1)x
0
.
Obviously
_
b + 1
a
_
=
r
1
r
2
+ 1
r
1
+r
2
> r
1
. Hence g
1
is well dened. We now
establish the monotony of g
1
. One has

1 a
2
+b ab
a b + 1

= (b + 1)
2
a
2
=
= (a +b + 1)(b + 1 a) = (r
2
1
1)(r
2
2
1) > 0,
and it follows g
1
increasing. It is clear that g
1
(x
0
r
1
) =
x
0
_
r
3
1
r
1
_
x
0
_
r
2
1
1
_ = r
1
and g
1
(x
0
r
2
) =
x
0
_
r
3
2
r
2
_
x
0
_
r
2
2
1
_ = r
2
. Since g
1
is continuous and increasing, it
follows that g
1
([x
0
r
2
, x
0
r
1
]) = [r
2
, r
1
] . Consequently, for all [x
0
r
2
, x
0
r
1
]
there exists linear ((x) = px + q) such that p = . Hence C
1
= [r
2
, r
1
]
and it follows that C = [r
2
, r
1
].
b) We shall prove that T is injective.
Let h
1
, h
2
B with h
1
= h
2
. Let us prove that T(h
1
) = T(h
2
). Next,
we will denote by
1
the function T(h
1
) and by
2
the function T(h
2
), and

1
,
2
V . There exist x
1
, x

1
such that
1
V (x
1
) and
2
V (x

1
).
1) If x
1
= x

1
, then
1
(x
0
) = x
1
and
2
(x
0
) = x

1
, with x
1
= x

1
, so

1
=
2
.
2) If x
1
= x

1
hence
1
,
2
V (x
1
), we have x
2
= x

2
with obvious
notations. But h
1
(1) =
1
r
2
_
x
1
x
3
x
0
x
2
_
=
1
r
2
_
x

1
x

3
x
0
x

2
_
= h
2
(1). Since
h
1
= h
2
, there exists (0, 1) such that h
1
() = h
2
(). We choose
x

= x
0
+ (x
2
x
0
); x

(x
0
, x
2
).
We shall prove that
1
() =
2
(). One has

1
(x

) =
x
3
x
1
h
1
(1)
h
1
_
x

x
0
x
2
x
0
_
+x
1
=
x
3
x
1
h
1
(1)
h
1
() +x
1
.
In the same way

2
(x

) =
x

3
x

1
h
2
(1)
h
2
() +x

1
.
256 Traian Cristian Gdea
Since x
1
= x

1
, x
3
= x

3
and h
1
() = h
2
(), it follows that
1
(x

) =
2
(x

).
Consequently
1
=
2
and it follows that T is injective.
c) We shall show that T is surjective. Let V ; there exists x
1
such
that V (x
1
). Let us prove that there exists h B with T(h) = . Since
V (x
1
), it follows that
x x
0
x
2
x
0
[0, 1] for x [x
0
, x
1
]. We can write
t =
x x
0
x
2
x
0
[0, 1].
We dene h(t) =
(x) x
1
x
3
x
1

x
1
x
3
r
2
(x
0
x
2
)
. Hence h(t) =
(x) x
1
(x
2
x
0
) r
2
.
1) Let us prove that h B. Obviously h is continuous and increasing.
One has h(0) = 0 and h(1) =
1
r
2
_
x
1
x
3
x
0
x
2
_
(see (2.6)). According to (2.7),
if x
1
[x
0
r
1
, x
0
r
2
], it follows that h(1)
_
r
1
r
2
, 1
_
. Consequently h is well
dened in 1.
Let 0 t
1
< t
2
1, and write
x = x
0
+t
1
(x
2
x
0
), y = x
0
+t
2
(x
2
x
0
).
One has the equality
(y) (x)
y x
= r
2
_
h(t
2
) h(t
1
)
t
2
t
1
_
(see (2.7)). Hence
(y) (x)
y x
[r
2
, r
1
] if and only if
h(t
2
) h(t
1
)
t
2
t
1

_
r
1
r
2
, 1
_
and h B.
2) Let us prove that T(h) = . One has
T(h)(x) =
_
x
3
x
1
h(1)
_

_
(x) x
1
(x
2
x
0
) r
2
_
+x
1
=
=
x
3
x
1
_
x
1
x
3
x
0
x
2
_

1
r
2

(x) x
1
(x
2
x
0
) r
2
+x
1
= (x).
Consequently T(h) = and T is bijective.
We dene F = H T, H : V S and T : B V , so H T : B S.
Since H and T are bijective, it follows that F is also bijective. 2
C. Case 0 < r
1
< 1 < r
2
.
The set of solutions is given by [4, Theorem 1.4]. Let us observe that the
set of solutions on R, for which f(x) > x, depends upon two parameters:
x
1
> 0 and the function f
0
: [0, x
1
] [x
1
, x
2
], where (x
n
)
nZ
is that one
of [4, Theorem 1.4] and f
0
is continuous bijective and veries the relation
(2.1). Under these conditions it follows from Theorem 1.1 that, starting with
(x
1
, f
0
) under previous conditions, these parameters determine uniquely the
solution on R, which additionally fullls the condition f(0) = x
1
(because
Unicity properties and algebraic properties for the solutions. . . 257
two solutions f and g with f(0) = g(0) = x
1
which coincide on [0, x
1
],
coincide on R according to Theorem 1.1).
Notation 2.3. In the sequel we will use the following notation.
(i) V (x
1
) = {f
0
: [0, x
1
] [x
1
, ax
1
] f
0
is continuous, bijective, increasing,
f
0
satises (2.1)}.
(ii) V =

x
1
(0,)
V (x
1
).
(iii) S
+
= {f : R R | f is (strictly) increasing and is solution
of equation (1.1), f(x) > x}.
(iv) B = {h : [0, 1] [0, 1], h(0) = 0, h is continuous, increasing,
h(1) =
r
1
+r
2
1
r
2
;
h(x) h(y)
x y

_
r
1
r
2
, 1
_
; x > y; x, y [0, 1]
_
.
Lemma 2.3. There exists a bijection H : V S
+
.
Proof. Let f
0
V . There exists x
1
> 0 such that f
0
V (x
1
). We dene
H(f
0
) = f, where f is the unique function which fullls the conditions
f(x
0
) = x
1
, f S
+
and f(x) = f
0
(x) for all x [0, x
1
]. Then it follows that
f satises (2.1), for all x > y 0. Hence H is well dened.
a) We shall prove that H is injective. Let f
1
, f
2
S
+
with H
_
f
1
0
_
= f
1
;
H
_
f
2
0
_
= f
2
and f
1
= f
2
. Let x
1
> 0, such that f
1
(0) = f
2
(0) = x
1
,
f
1
0
= f
1
|
[0,x
1
]
; f
2
0
= f
2
|
[0,x
1
]
. Since f
1
|
[0,x
1
]
= f
2
|
[0,x
1
]
it follows that f
1
0
= f
2
0
,
so H is injective.
b) We shall prove that H is surjective. Let f S
+
. We look for f
0
V
with H(f
0
) = f. Take x
1
= f(0); f
0
= f|
[0,x
1
]
; x
2
= f f(0); then one has
f
0
: [0, x
1
] [x
1
, x
2
]. We shall prove that f
0
V (x
1
). Since f S
+
, it
follows that x
1
> 0. The equation f(f(0)) = a f(0) b 0 implies x
2
= ax
1
.
Since f veries (2.1), f
0
veries (2.1), because f
0
= f|
[0,x
1
]
. Hence H is
surjective and then it follows that H is bijective. 2
Theorem 2.3. There exists a bijection F : B S
+
.
Proof. Firstly we shall prove that there exists T : B V , T bijective.
Consider x
1
> 0 and put x
2
= ax
1
.
Let h B. If x [0, x
1
], then there exists t [0, 1] such that t =
x
x
1
.
Consequently t [0, 1] if and only if x [0, x
1
]. We now dene by
: [0, x
1
] [x
1
, ax
1
] , (x) = x
1
r
2
h
_
x
x
1
_
+x
1
.
For x = x
1
it holds
(x
1
) = x
1
r
2
h(1) +x
1
= x
1
r
2

(a 1)
r
2
+x
1
= ax
1
= x
2
.
258 Traian Cristian Gdea
Hence (x) [x
1
, ax
1
], because is increasing.
We dene T : B V by T(h) = .
a) We shall prove that T is well dened. More precisely, starting with
h B and dening as above, let us prove that V (x
1
).
Obviously (0) = x
1
and (x
1
) = ax
1
. Since is increasing, continuous,
(0) = x
1
and (x
1
) = ax
1
, it follows that is bijective.
Let us prove that
(y) (x)
y x
[r
1
, r
2
]. For x
1
> y > x > 0 put
x = t
1
x
1
, y = t
2
x
1
, where t
1
, t
2
are in [0, 1]. One has y x = (t
2
t
1
) x
1
and y > x if and only if t
2
> t
1
. Then we have:
_
(y) = x
1
r
2
h(t
2
) +x
1
(x) = x
1
r
2
h(t
1
) +x
1
and it follows (y) (x) = x
1
r
2
(h(t
2
) h(t
1
)). We deduce that
(y) (x)
y x
= x
1
r
2

h(t
2
) h(t
1
)
t
2
t
1

_
t
2
t
1
y x
_
=
=
h(t
2
) h(t
1
)
t
2
t
1
x
1
r
2

1
x
1
.
In conclusion, it holds
(y) (x)
y x
= r
2

_
h(t
2
) h(t
1
)
t
2
t
1
_
. (2.8)
Since
h(t
2
) h(t
1
)
t
2
t
1

_
r
1
r
2
, 1
_
, we have
(y) (x)
y x
[r
1
, r
2
] . Hence
V (x
1
). Then it follows that T is well dened.
We shall prove that if
C =
_
x
1
[0,]
_
V (x
1
)
_
(y) (x)
y x

y > x; x, y [0, x
1
]
_
,
then C = [r
1
, r
2
]. We have C [r
1
, r
2
] and it remains to prove the inclusion
[r
1
, r
2
] C. We look for functions : [0, x
1
] [x
1
, ax
1
] (a = r
1
+r
2
) of the
form
(x) =
_
rx +c x [0, x

1
)
r

x +c

x [x

1
, x
1
] ,
where r = r

and we study what values can take the ratio

(x) (y)
x y
,
where 0 y < x x

, in order to have V (x
1
). Consequently the
following necessary conditions are compulsory: (0) = x
1
, (x
1
) = ax
1
,
rx

1
+c = r

1
+c

and r
1
r, r

r
2
. Consequently
_
_
_
c = x
1
r

x
1
+c

= ax
1
rx

1
+c = r

1
+c,

which implies
_
_
_
c = x
1
c

= (a r

)x
1
rx

1
+x
1
= r

1
+ (a r

)x
1
.
Unicity properties and algebraic properties for the solutions. . . 259
It follows that
x

1
=
x
1
(1 a +r

)
r

r
.
Hence one has
0 < x

1
< x
1
if and only if
1 a +r

r
> 0
and
x
1
(1 a +r

)
r

r
< x
1
if and only if
1 a +r

r
< 0.
If r

> r, then
_
1 a +r < 0
1 a +r

> 0
, so
_
r < a 1
r

> a 1.
Then r [r
1
, a 1)
and r

(a 1, r
2
]. Hence if r < r

, then r can take any value from the

interval [r
1
, a 1).
If r

< r, then
_
1 a +r > 0
1 a +r

< 0
, so
_
r > a 1
r

< a 1.
Then r (a 1, r
2
]
and r

[r
1
, a 1). Therefore if r > r

, then r can take any value from the

interval (a 1, r
2
].
Our next objective is to prove that V (x
1
). We observe that it is
enough to prove that
0 < x
1
implies that r
1
( ) () () r
2
( ).
If , [0, x

1
] or , [x

1
, x
1
] it is obvious that
r
1
( ) () () r
2
( ),
according to the previous considerations.
Let us suppose that < x

1
< . We shall prove that the inequality
() () r
2
( ) holds. Indeed, one has
() = r +c = r +x
1
, () = r

+c

= r

+ (a r

)x
1
,
deducing that () () r
2
( ) if and only if
r

+ax
1
r

x
1
r x
1
r
2
r
2
.
The latter inequality is equivalent to
>
r
2
r
r
2
r

+
x
1
(a r

1)
r
2
r

.
Since > x

1
, it is sucient to prove that
x

1
>
_
r
2
r
r
2
r

_
+x
1
_
a r

1
r
2
r

_
.
260 Traian Cristian Gdea
By replacing x

1
, we get the inequality
x
1
_
1 a +r

r
_
>
_
r
2
r
r
2
r

_
+x
1
_
a r

1
r
2
r

_
.
By direct computation one shows that this is equivalent to
x
1
_
1 a +r

_
r
2
r
(r

r)(r
2
r

)
>
r
2
r
r
2
r

.
The latter inequality is equivalent to
< x
1
_
1 a +r

r
_
,
that is < x

1
. We already know that this is true, so ()() r
2
().
We prove now that () () r
1
( ), that is
r

+ax
1
r

x
1
r x
1
r
1
r
1
.
This relation is equivalent to
>
_
r r
1
r

r
1
_
+x
1
_
r

a + 1
r

r
1
_
.
Since > x
1
, it is enough to prove that
x

1
>
_
r r
1
r

r
1
_
+x
1
_
r

a + 1
r

r
1
_
.
By replacing x

1
, we get the equivalent inequality
x
1
_
1 a +r

r
_
>
_
r r
1
r

r
1
_
+x
1
_
r

a + 1
r

r
1
_
.
It is easy to check that this relation is equivalent to
x
1
_
1 a +r

_
r r
1
(r

r)(r

r
1
)
>
_
r r
1
r

r
1
_
,
that is < x

1
. From our hypothesis we know that this inequality is true.
Hence
() () r
1
( ).
Thus we have V (x
1
).
We have proved also that: [r
1
, r
2
] {a 1} C.
Obviously, choosing (x) = (a 1)x +x
1
, we have
(0) = x
1
, (x
1
) = ax
1
and
(y) (x)
y x
= a 1.
Unicity properties and algebraic properties for the solutions. . . 261
Since a 1 (r
1
, r
2
), it follows that V (x
1
). Hence [r
1
, r
2
] C. Then it
follows that C = [r
1
, r
2
].
b) Let us prove that T is injective. Let h
1
, h
2
B with h
1
= h
2
. We
shall prove that T (h
1
) = T (h
2
). Put T (h
1
) =
1
and T (h
2
) =
2
, where

1
,
2
V . There exist x
1
and x

1
such that
1
V (x
1
) and
2
V (x

1
). If
x
1
= x

1
we have
1
(0) = x
1
and
2
(0) = x

1
, so
1
=
2
. If x
1
= x

1
, hence

1
,
2
V (x
1
) we have x
2
= x

2
= ax
1
. Since h
1
= h
2
, there exists (0, 1)
such that h
1
() = h
2
(). We choose x

= x
1
, hence x

(0, x
1
). We
have

1
(x

) = x
1
r
2
h
1
() +x
1
,

2
(x

) = x

1
r
2
h
2
() +x

1
.
Since x
1
= x

1
and h
1
() = h
2
(), we deduce that
1
(x

) =
2
(x

). Conse-
quently
1
=
2
and we deduce that T is injective.
c) Let us prove that T is surjective. Let V ; there exists x
1
such that
V (x
1
). Let us prove that there exists h B such that T(h) = . For
any x [0, x
1
], write t =
x
x
1
[0, 1]. We dene h : [0, 1] [0, 1] via
h(t) =
(tx
1
) x
1
x
1
r
2
.
1) Let us prove that h B. Obviously h is continuous and increasing.
One has h(0) = 0, h(1) =
(x
1
) x
1
x
1
r
2
=
a 1
r
2
. Hence h is well dened at
t = 1. Let 0 t
1
< t
2
1, and write x = t
1
x
1
, y = t
2
x
1
. According to
relation (2.8) we have:
(y) (x)
y x
= r
2
_
h(t
2
) h(t
1
)
t
2
t
1
_
.
Hence
(y) (x)
y x
[r
1
, r
2
] if and only if
h(t
2
) h(t
1
)
t
2
t
1

_
r
1
r
2
, 1
_
and this
implies that h B.
2) Let us prove that T(h) = . One has
T(h)(x) = x
1
r
2
_
(x) x
1
r
2
x
1
_
+x
1
= (x).
Consequently it follows that T(h) = . Thus it follows that T is surjective
and nally that T is bijective.
We dene F = H T, H : V S
+
and T : B V , so H T : B S
+
.
Since H and T are bijective, it follows that F is also bijective. 2
262 Traian Cristian Gdea
References
[1] K. Baron and W. Jarczyk, Recent results on functional equations in a single
variable perspectives and open problems, Aequationes Math., 61 (2001), 1-48.
[2] I. Chitescu and T. Gdea, The functional equation f f +af +b1
R
= 0, An. Stiint .
Univ. Al. I. Cuza Iasi. Mat. (N.S.), LVI (2010), 279-310.
[3] B. Choczewski and M. Kuczma, On iterative roots of polynomials, European Con-
ference on Iteration Theory (Lisbon, 1991), pp. 59-67, World Sci. Publishing, Singa-
pore, 1992.
[4] T. Gdea, Unicity properties and algebraic properties for the solutions of the func-
tional equation f f +af +b1
R
= 0 (I), Ann. Univ. Bucharest (Math. Ser.), 1 (LIX)
(2010), 99-110.
[5] J. Matkowski and W. Zhang, Method of characteristic for functional equations in
polynomial form, Acta Math. Sin., 13 (1997), 421-432.
[6] J. Matkowski and W. Zhang, On linear dependence of iterates, J. Appl. Anal., 6
(2000), 149-157.
[7] W. Zhang, Stability of the solution of the iterated equation
n
P
i=1
if
i
(x) = F(x), Acta
Math. Sci., 8 (1988), 421-424.
[8] W. Zhang, Iteration. Iterative roots and iterative equations, Trends in Mathematics,
4 (2001), 151-160.
[9] W. Zhang and W. Zhang, Computing iterative roots of polygonal functions, J.
Comput. Appl. Math., 205 (2007), 497-508.
[10] L. R. Zhao, Existence and uniqueness theorem for the solution of the functional
equation 1f(x) + 2f
2
(x) = F(x), J. China Univ. Sci. Tech., 13 (1983), 21-27.
Traian Cristian Gdea
University Politehnica of Bucharest, Faculty of Applied Sciences
313 Splaiul Independentei Str., Bucharest, Romania
E-mail: gtcmnop@yahoo.com