Sandwich theorems for radiant functions

Abstract


We show that between two graphs, one of a radiant function and the other of a coradiant, both defined on a real interval containing 0, there exists at least one line which separates the graphs. The conditions for the uniqueness of a separating linear function are also established.

Authors

Costică Mustăța
Tiberiu Popoviciu Institute of Numerical Analysis, Romania

Keywords

Sandwich theorems; radiant functions; coradiant functions.

Paper coordinates

C. Mustăța, Sandwich theorems for radiant functions, J. Numer. Anal. Approx. Theory, 44 (2015) no. 1, 81-90.

PDF

About this paper

Journal

Journal Numer. Anal.Approx. Theory

Publisher Name

Publishing house of the Romanian Academy

Print ISSN

2457-6794

Online ISSN

2501-059X

google scholar link

[1] K. Baron, J. Matkowski and K. Nikodem, A Sandwich with convexity, Mathematica Panonica, 5 (1994) no. 1, pp. 139–144.
[2] W. Forg-Rob, K. Nicodem and Z. Pales, Separation by monotonic functions, Mathematica Panonica, 7 (1996) no. 2, pp. 191–196.
[3] B. Fuchssteiner and W. Lusky, Convex Cones, North Holland Math. Stud., 56 (North Holland, Amsterdam, 1981).
[4] J.A. Johnson, Banach spaces of Lipschitz functions and vector-valued Lipschitz functions, Trans. Amer. Math. Soc., 148(1970), pp. 147–169.
[5] E.J. McShane, Extension of range of functions, Bull. Amer. Math. Soc., 40 (1934), pp. 837–842.
[6] C. Mustata, Norm preserving extension of starshaped Lipschitz functions, Mathematica (Cluj), 19(42)2 (1977), pp. 183–187.
[7] C. Mustata, Extensions of semi-Lipschitz functions on quasi-metric spaces, Rev. Anal. Numer, Theor. Approx., 30 (2001) no. 1, pp. 61–67.
[8] C. Mustata, On the extensions preserving the shape of semi-Holder function, Results. Math., 63 (2013), pp. 425–433.
[9] K. Nikodem and S. Wasowicz, A sandwich theorem and Hyers-Ulam stability of affine functions, Aequationes Math., 49 (1995), pp. 160–164.
[10] A.M. Rubinov, Abstract Convexity and Global Optimization, Kluwer Academic Publisher, Boston-Dordrecht-London, 2000.
[11] A.M. Rubinov and A.P. Shveidel, Radiant and star-shaped functions, Pacific Journal of Optimization, 3 (2007) no. 1, pp. 193–212.
[12] S. Simons, The asymmetric sandwich theorem, Journal of Convex Analysis, 20 (2013) no. 1, pp. 107–124.
[13] S. Suzuki and D. Kuroiwa, Sandwich theorem for quasiconvex functions and its applications, J. Math. Anal. Appl., 379 (2011), pp. 649–655.
[14] A. Szaz, The infimal convolution can be used to derive extensions theorems from sandwich ones, Acta Sci. Math. (Szeged), 76 (2010), pp. 489–499.

Paper (preprint) in HTML form

2015-Mustata-SANDWICH THEOREMS FOR RADIANT FUNCTIONS-JNAAT

SANDWICH THEOREMS FOR RADIANT FUNCTIONS

COSTICĂ MUSTĂŢA*Dedicated to prof. I. Păvăloiu on the occasion of his 75th anniversary

Abstract

We show that between two graphs, one of a radiant function and the other of a coradiant, both defined on a real interval containing 0 , there exists at least one line which separates the graphs. The conditions for the uniqueness of a separating linear function are also established.

MSC 2010. 46A22, 26A16, 47N10.
Keywords. Sandwich theorems, radiant functions, coradiant functions.

1. INTRODUCTION

The paper is concerned with the existence and uniqueness of a linear function whose graph separates the graphs of two real-valued functions defined on an interval in R R R\mathbb{R}R containing zero (i.e. a radiant subset of R R R\mathbb{R}R ), one of which being radiant and the other one coradiant. We show that, under some conditions, this sandwich-type problem has at least one solution. The uniqueness of the solution is also discussed. As application, one gives a sufficient of Hyers-Ulam type stability conditions for positively homogeneous functions.
Sandwich theorems for diverse classes of real-valued functions (monotonic, convex, quasiconvex) were considered in [1], [2], [9], [13] and for more general functions in [3], [12], [14, etc.
Let I I III be an interval in R R R\mathbb{R}R containing 0 . Then I I III is a radiant set, i.e. for every x I x I x in Ix \in IxI and λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] it follows λ x I λ x I lambda x in I\lambda x \in IλxI.
A function f : I R f : I R f:I rarrRf: I \rightarrow \mathbb{R}f:IR is called a radiant function if
(1) f ( λ x ) λ f ( x ) (1) f ( λ x ) λ f ( x ) {:(1)f(lambda x) <= lambda f(x):}\begin{equation*} f(\lambda x) \leq \lambda f(x) \tag{1} \end{equation*}(1)f(λx)λf(x)
for all x I x I x in Ix \in IxI and λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1].
A function g : I R g : I R g:I rarrRg: I \rightarrow \mathbb{R}g:IR is called a coradiant function if
(2) g ( λ x ) λ g ( x ) (2) g ( λ x ) λ g ( x ) {:(2)g(lambda x) >= lambda g(x):}\begin{equation*} g(\lambda x) \geq \lambda g(x) \tag{2} \end{equation*}(2)g(λx)λg(x)
for all x I x I x in Ix \in IxI and λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1].
Obviously, a function f f fff is radiant iff f f -f-ff is coradiant. Every radiant function f f fff verifies the inequality f ( 0 ) 0 f ( 0 ) 0 f(0) <= 0f(0) \leq 0f(0)0 and every coradiant function g g ggg verifies g ( 0 ) 0 g ( 0 ) 0 g(0) >= 0g(0) \geq 0g(0)0.
A function h : I R h : I R h:I rarrRh: I \rightarrow \mathbb{R}h:IR is called a Lipschitz function if there exists a constant K 0 K 0 K >= 0K \geq 0K0 (depending on f f fff and I I III ), such that
(3) | h ( x ) h ( y ) | K | x y | , (3) | h ( x ) h ( y ) | K | x y | , {:(3)|h(x)-h(y)| <= K|x-y|",":}\begin{equation*} |h(x)-h(y)| \leq K|x-y|, \tag{3} \end{equation*}(3)|h(x)h(y)|K|xy|,
for all x , y I x , y I x,y in Ix, y \in Ix,yI.
The constant K K KKK is called a Lipschitz constant (for f f fff ) and the smallest Lipschitz constant is given by following expression:
(4) f I = sup { | f ( x ) f ( y ) | | x y | : x y , x , y I } (4) f I = sup | f ( x ) f ( y ) | | x y | : x y , x , y I {:(4)||f||_(I)=s u p{(|f(x)-f(y)|)/(|x-y|):x!=y,x,y in I}:}\begin{equation*} \|f\|_{I}=\sup \left\{\frac{|f(x)-f(y)|}{|x-y|}: x \neq y, x, y \in I\right\} \tag{4} \end{equation*}(4)fI=sup{|f(x)f(y)||xy|:xy,x,yI}
Denote by L i p 0 I L i p 0 I Lip_(0)IL i p_{0} ILip0I the real linear space of Lipschitz functions on I I III vanishing at 0 , i.e.
(5) Lip 0 I := { f : I R , f ( 0 ) = 0 and f I < } . (5) Lip 0 I := f : I R , f ( 0 ) = 0  and  f I < . {:(5)Lip_(0)I:={f:I rarrR,f(0)=0" and "||f||_(I) < oo}.:}\begin{equation*} \operatorname{Lip}_{0} I:=\left\{f: I \rightarrow \mathbb{R}, f(0)=0 \text { and }\|f\|_{I}<\infty\right\} . \tag{5} \end{equation*}(5)Lip0I:={f:IR,f(0)=0 and fI<}.
The functional I : Lip 0 I [ 0 , ) I : Lip 0 I [ 0 , ) ||||_(I):Lip_(0)I rarr[0,oo):}\left\|\|_{I}: \operatorname{Lip}_{0} I \rightarrow[0, \infty)\right.I:Lip0I[0,) is a norm (called the Lipschitz norm) and Lip 0 I Lip 0 I Lip_(0)I\operatorname{Lip}_{0} ILip0I is a Banach space with respect to this norm [4].
Denote
(6) r L i p 0 I := { f L i p 0 I , f radiant } , (7) c r L i p 0 I := { g L i p 0 I , g coradiant } (6) r L i p 0 I := f L i p 0 I , f  radiant  , (7) c r L i p 0 I := g L i p 0 I ,  g coradiant  {:[(6)r-Lip_(0)I:={f in Lip_(0)I,f" radiant "}","],[(7)cr-Lip_(0)I:={g in Lip_(0)I," g coradiant "}]:}\begin{align*} r-L i p_{0} I & :=\left\{f \in L i p_{0} I, f \text { radiant }\right\}, \tag{6}\\ c r-L i p_{0} I & :=\left\{g \in L i p_{0} I, \text { g coradiant }\right\} \tag{7} \end{align*}(6)rLip0I:={fLip0I,f radiant },(7)crLip0I:={gLip0I, g coradiant }
and
(8) o Lip 0 I := { h Lip 0 I : h ( λ x ) = λ h ( x ) , λ [ 0 , 1 ] , x I } . (8) o Lip 0 I := h Lip 0 I : h ( λ x ) = λ h ( x ) , λ [ 0 , 1 ] , x I . {:(8)o-Lip_(0)I:={h inLip_(0)I:h(lambda x)=lambda h(x),lambda in[0,1],x in I}.:}\begin{equation*} o-\operatorname{Lip}_{0} I:=\left\{h \in \operatorname{Lip}_{0} I: h(\lambda x)=\lambda h(x), \lambda \in[0,1], x \in I\right\} . \tag{8} \end{equation*}(8)oLip0I:={hLip0I:h(λx)=λh(x),λ[0,1],xI}.
Then r r r-r-r Lip 0 I 0 I _(0)I_{0} I0I and c r c r cr-c r-cr Lip 0 I 0 I _(0)I_{0} I0I are convex cones in the linear space Lip 0 I 0 I _(0)I_{0} I0I and o L i p 0 I o L i p 0 I o-Lip_(0)Io-L i p_{0} IoLip0I is a subspace of L i p 0 I L i p 0 I Lip_(0)IL i p_{0} ILip0I. Also
(9) o L i p 0 I = ( r L i p 0 ) ( c r L i p 0 I ) . (9) o L i p 0 I = r L i p 0 c r L i p 0 I . {:(9)o-Lip_(0)I=(r-Lip_(0))nn(cr-Lip_(0)I).:}\begin{equation*} o-L i p_{0} I=\left(r-L i p_{0}\right) \cap\left(c r-L i p_{0} I\right) . \tag{9} \end{equation*}(9)oLip0I=(rLip0)(crLip0I).

2. EXTENSIONS PRESERVING RADIANTNESS OF A LIPSCHITZ FUNCTIONS

The following extension result for real valued Lipschitz functions defined on a subset of a metric space was given by McShane [5]:
Theorem 1. Let ( X , d X , d X,dX, dX,d ) be a metric space, Y Y YYY a nonvoid subset of X X XXX and f : Y R f : Y R f:Y rarrRf: Y \rightarrow \mathbb{R}f:YR be a Lipschitz function having the Lipschitz constant K ( f ) K ( f ) K(f)K(f)K(f) (on Y Y YYY ). Then there exists a Lipschitz function F : X R F : X R F:X rarrRF: X \rightarrow \mathbb{R}F:XR such that
(10) F | Y = f and K ( F ) = K ( f ) . (10) F Y = f  and  K ( F ) = K ( f ) . {:(10)F|_(Y)=f quad" and "quad K(F)=K(f).:}\begin{equation*} \left.F\right|_{Y}=f \quad \text { and } \quad K(F)=K(f) . \tag{10} \end{equation*}(10)F|Y=f and K(F)=K(f).
Such a function F F FFF is called a Lipschitz extension of f f fff, preserving the Lipschitz constant. In the proof of this theorem one shows that the following two functions
(11) F ( f ) ( x ) := inf y Y { f ( y ) + K ( f ) d ( x , y ) } , x X (11) F ( f ) ( x ) := inf y Y { f ( y ) + K ( f ) d ( x , y ) } , x X {:(11)F(f)(x):=i n f_(y in Y){f(y)+K(f)d(x","y)}","x in X:}\begin{equation*} F(f)(x):=\inf _{y \in Y}\{f(y)+K(f) d(x, y)\}, x \in X \tag{11} \end{equation*}(11)F(f)(x):=infyY{f(y)+K(f)d(x,y)},xX
and
(12) G ( f ) ( x ) := inf y Y { f ( y ) K ( f ) d ( x , y ) } , x X (12) G ( f ) ( x ) := inf y Y { f ( y ) K ( f ) d ( x , y ) } , x X {:(12)G(f)(x):=i n f_(y in Y){f(y)-K(f)d(x","y)}","x in X:}\begin{equation*} G(f)(x):=\inf _{y \in Y}\{f(y)-K(f) d(x, y)\}, x \in X \tag{12} \end{equation*}(12)G(f)(x):=infyY{f(y)K(f)d(x,y)},xX
are Lipschitz extensions of f f fff, preserving the constant K ( f ) K ( f ) K(f)K(f)K(f), and every other extension H H HHH verifies the inequalities:
(13) F ( x ) H ( x ) G ( x ) , x X (13) F ( x ) H ( x ) G ( x ) , x X {:(13)F(x) >= H(x) >= G(x)","x in X:}\begin{equation*} F(x) \geq H(x) \geq G(x), x \in X \tag{13} \end{equation*}(13)F(x)H(x)G(x),xX
(see also [7], 8] for a more general situation).
In the framework considered above we obtain the following result:
Theorem 2. Let I R I R I subRI \subset \mathbb{R}IR such that 0 I 0 I 0in I0 \in I0I and f : I R f : I R f:I rarrRf: I \rightarrow \mathbb{R}f:IR.
a) If f r f r f in rf \in rfr-Lip 0 I 0 I _(0)I{ }_{0} I0I then the greatest extension of f f fff, namely
(14) F ( f ) ( x ) := inf y I { f ( y ) + f I | x y | } , x R (14) F ( f ) ( x ) := inf y I f ( y ) + f I | x y | , x R {:(14)F(f)(x):=i n f_(y in I){f(y)+||f||_(I)|x-y|}","x inR:}\begin{equation*} F(f)(x):=\inf _{y \in I}\left\{f(y)+\|f\|_{I}|x-y|\right\}, x \in \mathbb{R} \tag{14} \end{equation*}(14)F(f)(x):=infyI{f(y)+fI|xy|},xR
is a radiant function on R R R\mathbb{R}R,
b) If g c r g c r g in crg \in c rgcr-Lip 0 I 0 I _(0)I{ }_{0} I0I, then the smallest extension of g g ggg, namely
(15) G ( g ) ( x ) := sup y I { g ( y ) g I | x y | } , x R (15) G ( g ) ( x ) := sup y I g ( y ) g I | x y | , x R {:(15)G(g)(x):=s u p_(y in I){g(y)-||g||_(I)|x-y|}","x inR:}\begin{equation*} G(g)(x):=\sup _{y \in I}\left\{g(y)-\|g\|_{I}|x-y|\right\}, x \in \mathbb{R} \tag{15} \end{equation*}(15)G(g)(x):=supyI{g(y)gI|xy|},xR
is a coradiant function on R R R\mathbb{R}R.
c) If h o h o h in o-h \in o-ho Lip 0 I 0 I _(0)I_{0} I0I then the extension F ( h ) F ( h ) F(h)F(h)F(h) defined by (14) is in the cone r r rrr-Lip 0 R 0 R _(0)R{ }_{0} \mathbb{R}0R, and G ( h ) G ( h ) G(h)G(h)G(h) defined by (15) is in the cone cr-Lip 0 R 0 R _(0)R{ }_{0} \mathbb{R}0R.
d) If f f f inf \inf Lip 0 R 0 R _(0)R_{0} \mathbb{R}0R and there exists I R I R I subRI \subset \mathbb{R}IR with 0 I 0 I 0in I0 \in I0I such that f | I I = f R f I I = f R ||f|_(I)||_(I)=||f||_(R)\left\|\left.f\right|_{I}\right\|_{I}= \|f\|_{R}f|II=fR and f | I o f I o f|_(I)in o-\left.f\right|_{I} \in o-f|Io Lip 0 I 0 I _(0)I_{0} I0I, then
(16) G ( f | I ) ( x ) f ( x ) F ( f | I ) ( x ) , x R (16) G f I ( x ) f ( x ) F f I ( x ) , x R {:(16)G(f|_(I))(x) <= f(x) <= F(f|_(I))(x)","x inR:}\begin{equation*} G\left(\left.f\right|_{I}\right)(x) \leq f(x) \leq F\left(\left.f\right|_{I}\right)(x), x \in \mathbb{R} \tag{16} \end{equation*}(16)G(f|I)(x)f(x)F(f|I)(x),xR
Proof. The proof is similar to that in [8. For the sake of completeness we sketch the proof.
a) Let f r f r f in rf \in rfr-Lip 0 I 0 I _(0)I{ }_{0} I0I. Then, for all λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] and y I y I y in Iy \in IyI,
F ( f ) ( λ x ) = inf y I { f ( y ) + f I | λ x y | } inf y I { f ( λ y ) + f I | λ x λ y | } inf y I { λ f ( y ) + λ f I | x y | } = λ inf y I { f ( y ) + f I | x y | } = λ F ( f ) ( x ) F ( f ) ( λ x ) = inf y I f ( y ) + f I | λ x y | inf y I f ( λ y ) + f I | λ x λ y | inf y I λ f ( y ) + λ f I | x y | = λ inf y I f ( y ) + f I | x y | = λ F ( f ) ( x ) {:[F(f)(lambda x)=i n f_(y in I){f(y)+||f||_(I)|lambda x-y|}],[ <= i n f_(y in I){f(lambda y)+||f||_(I)|lambda x-lambda y|}],[ <= i n f_(y in I){lambda f(y)+lambda||f||_(I)|x-y|}],[=lambdai n f_(y in I){f(y)+||f||_(I)|x-y|}],[=lambda F(f)(x)]:}\begin{aligned} F(f)(\lambda x) & =\inf _{y \in I}\left\{f(y)+\|f\|_{I}|\lambda x-y|\right\} \\ & \leq \inf _{y \in I}\left\{f(\lambda y)+\|f\|_{I}|\lambda x-\lambda y|\right\} \\ & \leq \inf _{y \in I}\left\{\lambda f(y)+\lambda\|f\|_{I}|x-y|\right\} \\ & =\lambda \inf _{y \in I}\left\{f(y)+\|f\|_{I}|x-y|\right\} \\ & =\lambda F(f)(x) \end{aligned}F(f)(λx)=infyI{f(y)+fI|λxy|}infyI{f(λy)+fI|λxλy|}infyI{λf(y)+λfI|xy|}=λinfyI{f(y)+fI|xy|}=λF(f)(x)
for every x R x R x inRx \in \mathbb{R}xR. It follows
F ( f ) ( λ x ) λ F ( f ) ( x ) , F ( f ) ( λ x ) λ F ( f ) ( x ) , F(f)(lambda x) <= lambda F(f)(x),F(f)(\lambda x) \leq \lambda F(f)(x),F(f)(λx)λF(f)(x),
for all λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] and all x R x R x inRx \in \mathbb{R}xR.
b) Let g c r g c r g in crg \in c rgcr-Lip 0 I 0 I _(0)I{ }_{0} I0I and x R x R x inRx \in \mathbb{R}xR. Then for all λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] and y I y I y in Iy \in IyI,
G ( g ) ( λ x ) = sup y I { g ( y ) g I | λ x y | } sup y I { g ( λ y ) g I | λ x λ y | } λ sup y I { g ( y ) g I | x y | } = λ G ( g ) ( x ) G ( g ) ( λ x ) = sup y I g ( y ) g I | λ x y | sup y I g ( λ y ) g I | λ x λ y | λ sup y I g ( y ) g I | x y | = λ G ( g ) ( x ) {:[G(g)(lambda x)=s u p_(y in I){g(y)-||g||_(I)|lambda x-y|}],[ >= s u p_(y in I){g(lambda y)-||g||_(I)|lambda x-lambda y|}],[ >= lambdas u p_(y in I){g(y)-||g||_(I)|x-y|}],[=lambda G(g)(x)]:}\begin{aligned} G(g)(\lambda x) & =\sup _{y \in I}\left\{g(y)-\|g\|_{I}|\lambda x-y|\right\} \\ & \geq \sup _{y \in I}\left\{g(\lambda y)-\|g\|_{I}|\lambda x-\lambda y|\right\} \\ & \geq \lambda \sup _{y \in I}\left\{g(y)-\|g\|_{I}|x-y|\right\} \\ & =\lambda G(g)(x) \end{aligned}G(g)(λx)=supyI{g(y)gI|λxy|}supyI{g(λy)gI|λxλy|}λsupyI{g(y)gI|xy|}=λG(g)(x)
Consequently G ( g ) ( λ x ) λ G ( g ) ( x ) G ( g ) ( λ x ) λ G ( g ) ( x ) G(g)(lambda x) >= lambda G(g)(x)G(g)(\lambda x) \geq \lambda G(g)(x)G(g)(λx)λG(g)(x) for all λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] and all x R x R x inRx \in \mathbb{R}xR.
c) If h o Lip 0 I = ( r Lip 0 I ) ( c r Lip 0 I ) h o Lip 0 I = r Lip 0 I c r Lip 0 I h in o-Lip_(0)I=(r-Lip_(0)I)nn(cr-Lip_(0)I)h \in o-\operatorname{Lip}_{0} I=\left(r-\operatorname{Lip}_{0} I\right) \cap\left(c r-\operatorname{Lip}_{0} I\right)hoLip0I=(rLip0I)(crLip0I), then the assertions from c) follow from a) and b).
d) Let f Lip 0 R f Lip 0 R f inLip_(0)Rf \in \operatorname{Lip}_{0} \mathbb{R}fLip0R and suppose that there exists I R I R I subRI \subset \mathbb{R}IR with 0 I 0 I 0in I0 \in I0I such that f | I o f I o f|_(I)in o-\left.f\right|_{I} \in o-f|Io Lip 0 I 0 I _(0)I_{0} I0I and f | I = f R f I = f R ||f|_(I)=||f||_(R)\left\|\left.f\right|_{I}=\right\| f \|_{R}f|I=fR. Then f f fff is an extension of f | I f I f|_(I)\left.f\right|_{I}f|I preserving the smallest Lipschitz constant f | I f I ||f|_(I)||\left\|\left.f\right|_{I}\right\|f|I and the assertion follows by (13) and c).

3. SANDWICH THEOREMS

The assertion d) in Theorem 2 suggests the following problem: Let f , g f , g f,gf, gf,g be two functions, on an interval I ( 0 I ) , f I ( 0 I ) , f I(0in I),fI(0 \in I), fI(0I),f radiant (coradiant) and g g ggg coradiant (radiant), and f ( x ) g ( x ) f ( x ) g ( x ) f(x) <= g(x)f(x) \leq g(x)f(x)g(x) for all x I x I x in Ix \in IxI. Is there a function h : I R h : I R h:I rarrRh: I \rightarrow \mathbb{R}h:IR verifying h ( λ x ) = λ h ( x ) h ( λ x ) = λ h ( x ) h(lambda x)=lambda h(x)h(\lambda x)=\lambda h(x)h(λx)=λh(x), for all x I x I x in Ix \in IxI and all λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1], and such that f ( x ) h ( x ) g ( x ) , x I f ( x ) h ( x ) g ( x ) , x I f(x) <= h(x) <= g(x),x in If(x) \leq h(x) \leq g(x), x \in If(x)h(x)g(x),xI ? Problems of such type were considered for example in [1], [2], [9] and, more general in [3], [12], [14], etc. The studies in this direction are motivated by applications to the theory of optimization and numerical analysis.
We show that the answer is affirmative in both cases.
Firstly, we need the following Lemma.
Lemma 3. Let f , g : [ 0 , a ] R , f f , g : [ 0 , a ] R , f f,g:[0,a]rarrR,ff, g:[0, a] \rightarrow \mathbb{R}, ff,g:[0,a]R,f radiant and g g ggg coradiant such that f ( x ) g ( x ) , x [ 0 , a ] f ( x ) g ( x ) , x [ 0 , a ] f(x) <= g(x),x in[0,a]f(x) \leq g(x), x \in[0, a]f(x)g(x),x[0,a]. Then the following assertions hold:
a) The function p f : ( 0 , a ] R p f : ( 0 , a ] R p_(f):(0,a]rarrRp_{f}:(0, a] \rightarrow \mathbb{R}pf:(0,a]R defined by
(17) p f ( x ) = f ( x ) x , (17) p f ( x ) = f ( x ) x , {:(17)p_(f)(x)=(f(x))/(x)",":}\begin{equation*} p_{f}(x)=\frac{f(x)}{x}, \tag{17} \end{equation*}(17)pf(x)=f(x)x,
is nondecreasing on ( 0 , a ] 0 , a ] 0,a]0, a]0,a], and the function p g : ( 0 , a ] R p g : ( 0 , a ] R p_(g):(0,a]rarrRp_{g}:(0, a] \rightarrow \mathbb{R}pg:(0,a]R defined by
(18) p g ( x ) = g ( x ) x (18) p g ( x ) = g ( x ) x {:(18)p_(g)(x)=(g(x))/(x):}\begin{equation*} p_{g}(x)=\frac{g(x)}{x} \tag{18} \end{equation*}(18)pg(x)=g(x)x
is nonincreasing on ( 0 , a 0 , a 0,a0, a0,a ];
b) For every x 0 ( 0 , a ] x 0 ( 0 , a ] x_(0)in(0,a]x_{0} \in(0, a]x0(0,a],
(19) f ( x ) x f ( x 0 ) x 0 g ( x 0 ) x 0 g ( x ) x , x ( 0 , x 0 ] . (19) f ( x ) x f x 0 x 0 g x 0 x 0 g ( x ) x , x 0 , x 0 . {:(19)(f(x))/(x) <= (f(x_(0)))/(x_(0)) <= (g(x_(0)))/(x_(0)) <= (g(x))/(x)","x in(0,x_(0)].:}\begin{equation*} \frac{f(x)}{x} \leq \frac{f\left(x_{0}\right)}{x_{0}} \leq \frac{g\left(x_{0}\right)}{x_{0}} \leq \frac{g(x)}{x}, x \in\left(0, x_{0}\right] . \tag{19} \end{equation*}(19)f(x)xf(x0)x0g(x0)x0g(x)x,x(0,x0].
Proof. a) Let f : [ 0 , a ] R , f f : [ 0 , a ] R , f f:[0,a]rarrR,ff:[0, a] \rightarrow \mathbb{R}, ff:[0,a]R,f radiant. For every x ( 0 , a ] x ( 0 , a ] x in(0,a]x \in(0, a]x(0,a] let λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] be such that x = λ a x = λ a x=lambda ax=\lambda ax=λa. Then
p f ( x ) = f ( x ) x = f ( λ a ) λ a λ f ( a ) λ a = f ( a ) a . p f ( x ) = f ( x ) x = f ( λ a ) λ a λ f ( a ) λ a = f ( a ) a . p_(f)(x)=(f(x))/(x)=(f(lambda a))/(lambda a) <= (lambda f(a))/(lambda a)=(f(a))/(a).p_{f}(x)=\frac{f(x)}{x}=\frac{f(\lambda a)}{\lambda a} \leq \frac{\lambda f(a)}{\lambda a}=\frac{f(a)}{a} .pf(x)=f(x)x=f(λa)λaλf(a)λa=f(a)a.
Let x 1 , x 2 ( 0 , a ] , x 1 x 2 x 1 , x 2 ( 0 , a ] , x 1 x 2 x_(1),x_(2)in(0,a],x_(1) <= x_(2)x_{1}, x_{2} \in(0, a], x_{1} \leq x_{2}x1,x2(0,a],x1x2. Then there exists λ ( 0 , 1 ] λ ( 0 , 1 ] lambda in(0,1]\lambda \in(0,1]λ(0,1] such that x 1 = λ x 2 x 1 = λ x 2 x_(1)=lambdax_(2)x_{1}=\lambda x_{2}x1=λx2, so that
p f ( x 1 ) = f ( x 1 ) x 1 = f ( λ x 2 ) λ x 2 λ f ( x 2 ) λ x 2 = f ( x 2 ) x 2 = p f ( x 2 ) . p f x 1 = f x 1 x 1 = f λ x 2 λ x 2 λ f x 2 λ x 2 = f x 2 x 2 = p f x 2 . p_(f)(x_(1))=(f(x_(1)))/(x_(1))=(f(lambdax_(2)))/(lambdax_(2)) <= (lambda f(x_(2)))/(lambdax_(2))=(f(x_(2)))/(x_(2))=p_(f)(x_(2)).p_{f}\left(x_{1}\right)=\frac{f\left(x_{1}\right)}{x_{1}}=\frac{f\left(\lambda x_{2}\right)}{\lambda x_{2}} \leq \frac{\lambda f\left(x_{2}\right)}{\lambda x_{2}}=\frac{f\left(x_{2}\right)}{x_{2}}=p_{f}\left(x_{2}\right) .pf(x1)=f(x1)x1=f(λx2)λx2λf(x2)λx2=f(x2)x2=pf(x2).
Also
p g ( x 1 ) = g ( x 1 ) x 1 = g ( λ x 2 ) λ x 2 λ g ( x 2 ) λ x 2 = g ( x 2 ) x 2 = p g ( x 2 ) . p g x 1 = g x 1 x 1 = g λ x 2 λ x 2 λ g x 2 λ x 2 = g x 2 x 2 = p g x 2 . p_(g)(x_(1))=(g(x_(1)))/(x_(1))=(g(lambdax_(2)))/(lambdax_(2)) >= (lambda g(x_(2)))/(lambdax_(2))=(g(x_(2)))/(x_(2))=p_(g)(x_(2)).p_{g}\left(x_{1}\right)=\frac{g\left(x_{1}\right)}{x_{1}}=\frac{g\left(\lambda x_{2}\right)}{\lambda x_{2}} \geq \frac{\lambda g\left(x_{2}\right)}{\lambda x_{2}}=\frac{g\left(x_{2}\right)}{x_{2}}=p_{g}\left(x_{2}\right) .pg(x1)=g(x1)x1=g(λx2)λx2λg(x2)λx2=g(x2)x2=pg(x2).
Consequently p f p f p_(f)p_{f}pf is nondecreasing and p g p g p_(g)p_{g}pg is nonincreasing on ( 0 , a 0 , a 0,a0, a0,a ].
b) Taking into account the hypothesis and the conclusions in a), the inequalities (19) follow.
Remark 4. The inequalities (17), (18) and (19) imply that every line passing through the points ( 0,0 ) and ( x , f ( x ) x , f ( x ) x,f(x)x, f(x)x,f(x) ) is above the graph of f f fff on the interval [ 0 , x 0 , x 0,x0, x0,x ], and every line passing through the points ( 0,0 ) and ( x , g ( x ) x , g ( x ) x,g(x)x, g(x)x,g(x) ) is bellow the graph of g g ggg on [ 0 , x ] [ 0 , x ] [0,x][0, x][0,x]. Also, the line passing through ( 0,0 ) and ( x , g ( x ) ) ( x , g ( x ) ) (x,g(x))(x, g(x))(x,g(x)) is above the line passing through ( 0 , 0 ) ( 0 , 0 ) (0,0)(0,0)(0,0) and ( x , f ( x ) ) ( x , f ( x ) ) (x,f(x))(x, f(x))(x,f(x)).
We obtain the main result of the paper.
Theorem 5. Let f , g : [ 0 , a ] R , f f , g : [ 0 , a ] R , f f,g:[0,a]rarrR,ff, g:[0, a] \rightarrow \mathbb{R}, ff,g:[0,a]R,f radiant and g g ggg coradiant, and f ( x ) g ( x ) f ( x ) g ( x ) f(x) <= g(x)f(x) \leq g(x)f(x)g(x), for all x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a]. Then there exists at least a function h : [ 0 , a ] R h : [ 0 , a ] R h:[0,a]rarrRh:[0, a] \rightarrow \mathbb{R}h:[0,a]R having the form h ( x ) = α x , α R h ( x ) = α x , α R h(x)=alpha x,alpha inRh(x)=\alpha x, \alpha \in \mathbb{R}h(x)=αx,αR and such that
(20) f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . (20) f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . {:(20)f(x) <= h(x) <= g(x)","x in[0","a].:}\begin{equation*} f(x) \leq h(x) \leq g(x), x \in[0, a] . \tag{20} \end{equation*}(20)f(x)h(x)g(x),x[0,a].
If there exists x 0 ( 0 , a ] x 0 ( 0 , a ] x_(0)in(0,a]x_{0} \in(0, a]x0(0,a] such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0) then the unique function h : [ 0 , a ] R h : [ 0 , a ] R h:[0,a]rarrRh:[0, a] \rightarrow \mathbb{R}h:[0,a]R verifying 20 is h ( x ) = f ( a ) a x h ( x ) = f ( a ) a x h(x)=(f(a))/(a)xh(x)=\frac{f(a)}{a} xh(x)=f(a)ax.
Proof. By Lemma 3, taking into account (19), one obtains
f ( x ) x f ( a ) a g ( a ) a g ( x ) x , x ( 0 , a ] . f ( x ) x f ( a ) a g ( a ) a g ( x ) x , x ( 0 , a ] . (f(x))/(x) <= (f(a))/(a) <= (g(a))/(a) <= (g(x))/(x),x in(0,a].\frac{f(x)}{x} \leq \frac{f(a)}{a} \leq \frac{g(a)}{a} \leq \frac{g(x)}{x}, x \in(0, a] .f(x)xf(a)ag(a)ag(x)x,x(0,a].
Then, for α [ f ( a ) a , g ( a ) a ] α f ( a ) a , g ( a ) a alpha in[(f(a))/(a),(g(a))/(a)]\alpha \in\left[\frac{f(a)}{a}, \frac{g(a)}{a}\right]α[f(a)a,g(a)a] the function h ( x ) = α x , x [ 0 , a ] h ( x ) = α x , x [ 0 , a ] h(x)=alpha x,x in[0,a]h(x)=\alpha x, x \in[0, a]h(x)=αx,x[0,a] verifies the properties from the conclusions of first part of theorem.
Now, if f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0) at x 0 ( 0 , a ] x 0 ( 0 , a ] x_(0)in(0,a]x_{0} \in(0, a]x0(0,a] one obtains
f ( x 0 ) x 0 f ( a ) a g ( a ) a g ( x 0 ) x 0 f x 0 x 0 f ( a ) a g ( a ) a g x 0 x 0 (f(x_(0)))/(x_(0)) <= (f(a))/(a) <= (g(a))/(a) <= (g(x_(0)))/(x_(0))\frac{f\left(x_{0}\right)}{x_{0}} \leq \frac{f(a)}{a} \leq \frac{g(a)}{a} \leq \frac{g\left(x_{0}\right)}{x_{0}}f(x0)x0f(a)ag(a)ag(x0)x0
and then f ( a ) = g ( a ) f ( a ) = g ( a ) f(a)=g(a)f(a)=g(a)f(a)=g(a). For x [ x 0 , a ] x x 0 , a x in[x_(0),a]x \in\left[x_{0}, a\right]x[x0,a] one obtains
f ( x ) x = g ( x ) x f ( x ) x = g ( x ) x (f(x))/(x)=(g(x))/(x)\frac{f(x)}{x}=\frac{g(x)}{x}f(x)x=g(x)x
i.e f ( x ) = g ( x ) = f ( a ) a x f ( x ) = g ( x ) = f ( a ) a x f(x)=g(x)=(f(a))/(a)xf(x)=g(x)=\frac{f(a)}{a} xf(x)=g(x)=f(a)ax. Then h ( x ) = f ( a ) a x h ( x ) = f ( a ) a x h(x)=(f(a))/(a)xh(x)=\frac{f(a)}{a} xh(x)=f(a)ax verifies f ( x ) h ( x ) g ( x ) , x [ 0 , a ] f ( x ) h ( x ) g ( x ) , x [ 0 , a ] f(x) <= h(x) <= g(x),x in[0,a]f(x) \leq h(x) \leq g(x), x \in [0, a]f(x)h(x)g(x),x[0,a], and the second part of the theorem follows.
In the case I = [ 0 , ) I = [ 0 , ) I=[0,oo)I=[0, \infty)I=[0,) one obtains.
Theorem 6. Let f , g : [ 0 , ) R , f f , g : [ 0 , ) R , f f,g:[0,oo)rarrR,ff, g:[0, \infty) \rightarrow \mathbb{R}, ff,g:[0,)R,f radiant and g g ggg coradiant, and f ( x ) g ( x ) f ( x ) g ( x ) f(x) <= g(x)f(x) \leq g(x)f(x)g(x) for all x [ 0 , ) x [ 0 , ) x in[0,oo)x \in[0, \infty)x[0,). Then there exists at least a function h : [ 0 , ) R h : [ 0 , ) R h:[0,oo)rarrRh:[0, \infty) \rightarrow \mathbb{R}h:[0,)R, of the form h ( x ) = α x , α R h ( x ) = α x , α R h(x)=alpha x,alpha inRh(x)=\alpha x, \alpha \in \mathbb{R}h(x)=αx,αR fixed such that f ( x ) h ( x ) g ( x ) , x [ 0 , ) f ( x ) h ( x ) g ( x ) , x [ 0 , ) f(x) <= h(x) <= g(x),x in[0,oo)f(x) \leq h(x) \leq g(x), x \in[0, \infty)f(x)h(x)g(x),x[0,).
If there exists x 0 > 0 x 0 > 0 x_(0) > 0x_{0}>0x0>0 such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0) then the unique function h : [ 0 , ) R h : [ 0 , ) R h:[0,oo)rarrRh:[0, \infty) \rightarrow \mathbb{R}h:[0,)R satisfying f ( x ) h ( x ) g ( x ) f ( x ) h ( x ) g ( x ) f(x) <= h(x) <= g(x)f(x) \leq h(x) \leq g(x)f(x)h(x)g(x) is h ( x ) = f ( x 0 ) x 0 x h ( x ) = f x 0 x 0 x h(x)=(f(x_(0)))/(x_(0))xh(x)=\frac{f\left(x_{0}\right)}{x_{0}} xh(x)=f(x0)x0x.
Proof. By Lemma 3
f ( x ) x f ( a ) a g ( a ) a g ( x ) x , x ( 0 , a ] f ( x ) x f ( a ) a g ( a ) a g ( x ) x , x ( 0 , a ] (f(x))/(x) <= (f(a))/(a) <= (g(a))/(a) <= (g(x))/(x),x in(0,a]\frac{f(x)}{x} \leq \frac{f(a)}{a} \leq \frac{g(a)}{a} \leq \frac{g(x)}{x}, x \in(0, a]f(x)xf(a)ag(a)ag(x)x,x(0,a]
for every a ( 0 , ) a ( 0 , ) a in(0,oo)a \in(0, \infty)a(0,),
Then
sup 0 < x a f ( x ) x inf 0 < x a g ( x ) x sup 0 < x a f ( x ) x inf 0 < x a g ( x ) x s u p_(0 < x <= a)(f(x))/(x) <= i n f_(0 < x <= a)(g(x))/(x)\sup _{0<x \leq a} \frac{f(x)}{x} \leq \inf _{0<x \leq a} \frac{g(x)}{x}sup0<xaf(x)xinf0<xag(x)x
and
lim a sup 0 < x a f ( x ) x lim a inf 0 < x a g ( x ) x lim a sup 0 < x a f ( x ) x lim a inf 0 < x a g ( x ) x lim_(a rarr oo)s u p_(0 < x <= a)(f(x))/(x) <= lim_(a rarr oo)i n f_(0 < x <= a)(g(x))/(x)\lim _{a \rightarrow \infty} \sup _{0<x \leq a} \frac{f(x)}{x} \leq \lim _{a \rightarrow \infty} \inf _{0<x \leq a} \frac{g(x)}{x}limasup0<xaf(x)xlimainf0<xag(x)x
By considering α [ lim a sup 0 < x a f ( x ) x , lim a inf 0 < x a g ( x ) x ] α lim a sup 0 < x a f ( x ) x , lim a inf 0 < x a g ( x ) x alpha in[lim_(a rarr oo)s u p_(0 < x <= a)(f(x))/(x),lim_(a rarr oo)i n f_(0 < x <= a)(g(x))/(x)]\alpha \in\left[\lim _{a \rightarrow \infty} \sup _{0<x \leq a} \frac{f(x)}{x}, \lim _{a \rightarrow \infty} \inf _{0<x \leq a} \frac{g(x)}{x}\right]α[limasup0<xaf(x)x,limainf0<xag(x)x] one obtains
f ( x ) x α g ( x ) x , x ( 0 , ) f ( x ) x α g ( x ) x , x ( 0 , ) (f(x))/(x) <= alpha <= (g(x))/(x),x in(0,oo)\frac{f(x)}{x} \leq \alpha \leq \frac{g(x)}{x}, x \in(0, \infty)f(x)xαg(x)x,x(0,)
and consequently
f ( x ) α x g ( x ) , x [ 0 , ) f ( x ) α x g ( x ) , x [ 0 , ) f(x) <= alpha x <= g(x),x in[0,oo)f(x) \leq \alpha x \leq g(x), x \in[0, \infty)f(x)αxg(x),x[0,)
Then h ( x ) = α x , x [ 0 , ) h ( x ) = α x , x [ 0 , ) h(x)=alpha x,x in[0,oo)h(x)=\alpha x, x \in[0, \infty)h(x)=αx,x[0,), satisfies the first conclusion of the theorem.
Let x 0 > 0 x 0 > 0 x_(0) > 0x_{0}>0x0>0 such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0). For every x > x 0 x > x 0 x > x_(0)x>x_{0}x>x0 we have
f ( x 0 ) x 0 f ( x ) x g ( x ) x g ( x 0 ) x 0 f x 0 x 0 f ( x ) x g ( x ) x g x 0 x 0 (f(x_(0)))/(x_(0)) <= (f(x))/(x) <= (g(x))/(x) <= (g(x_(0)))/(x_(0))\frac{f\left(x_{0}\right)}{x_{0}} \leq \frac{f(x)}{x} \leq \frac{g(x)}{x} \leq \frac{g\left(x_{0}\right)}{x_{0}}f(x0)x0f(x)xg(x)xg(x0)x0
It follows f ( x ) = g ( x ) = f ( x 0 ) x 0 x f ( x ) = g ( x ) = f x 0 x 0 x f(x)=g(x)=(f(x_(0)))/(x_(0))xf(x)=g(x)=\frac{f\left(x_{0}\right)}{x_{0}} xf(x)=g(x)=f(x0)x0x, for all x > x 0 x > x 0 x > x_(0)x>x_{0}x>x0 and consequently, for x [ 0 , ) x [ 0 , ) x in[0,oo)x \in[0, \infty)x[0,) the function h ( x ) = f ( x 0 ) x 0 x h ( x ) = f x 0 x 0 x h(x)=(f(x_(0)))/(x_(0))xh(x)=\frac{f\left(x_{0}\right)}{x_{0}} xh(x)=f(x0)x0x verifies
f ( x ) h ( x ) g ( x ) . f ( x ) h ( x ) g ( x ) . f(x) <= h(x) <= g(x).f(x) \leq h(x) \leq g(x) .f(x)h(x)g(x).
Now we consider I = [ a , 0 ] I = [ a , 0 ] I=[a,0]I=[a, 0]I=[a,0].
Theorem 7. Let f , g : [ a , 0 ] R ( a < 0 ) , f f , g : [ a , 0 ] R ( a < 0 ) , f f,g:[a,0]rarrR(a < 0),ff, g:[a, 0] \rightarrow \mathbb{R}(a<0), ff,g:[a,0]R(a<0),f radiant and g g ggg coradiant, and f ( x ) g ( x ) f ( x ) g ( x ) f(x) <= g(x)f(x) \leq g(x)f(x)g(x), for all x [ a , 0 ] x [ a , 0 ] x in[a,0]x \in[a, 0]x[a,0]. Then there exists at least a function h : [ a , 0 ] R h : [ a , 0 ] R h:[a,0]rarrRh:[a, 0] \rightarrow \mathbb{R}h:[a,0]R having the form h ( x ) = α x , α R h ( x ) = α x , α R h(x)=alpha x,alpha inRh(x)=\alpha x, \alpha \in \mathbb{R}h(x)=αx,αR fixed, and verifying the inequalities:
(21) f ( x ) h ( x ) g ( x ) , x [ a , 0 ] . (21) f ( x ) h ( x ) g ( x ) , x [ a , 0 ] . {:(21)f(x) <= h(x) <= g(x)","x in[a","0].:}\begin{equation*} f(x) \leq h(x) \leq g(x), x \in[a, 0] . \tag{21} \end{equation*}(21)f(x)h(x)g(x),x[a,0].
If there exists x 0 [ a , 0 ) x 0 [ a , 0 ) x_(0)in[a,0)x_{0} \in[a, 0)x0[a,0) such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0) then the only function h : [ a , 0 ] R h : [ a , 0 ] R h:[a,0]rarrRh:[a, 0] \rightarrow \mathbb{R}h:[a,0]R such that f ( x ) h ( x ) g ( x ) f ( x ) h ( x ) g ( x ) f(x) <= h(x) <= g(x)f(x) \leq h(x) \leq g(x)f(x)h(x)g(x) is h ( x ) = f ( a ) a x h ( x ) = f ( a ) a x h(x)=(f(a))/(a)xh(x)=\frac{f(a)}{a} xh(x)=f(a)ax.
Proof. Observe that for all x [ a , 0 ] x [ a , 0 ] x in[a,0]x \in[a, 0]x[a,0] one obtains f ( x ) f ( a ) a x f ( x ) f ( a ) a x f(x) <= (f(a))/(a)xf(x) \leq \frac{f(a)}{a} xf(x)f(a)ax. Indeed, if x [ a , 0 ) x [ a , 0 ) x in[a,0)x \in[a, 0)x[a,0), there exists λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] such that x = λ a x = λ a x=lambda ax=\lambda ax=λa. Then
f ( x ) = f ( λ a ) λ f ( a ) = f ( a ) a ( λ a ) = f ( a ) a x f ( x ) = f ( λ a ) λ f ( a ) = f ( a ) a ( λ a ) = f ( a ) a x f(x)=f(lambda a) <= lambda f(a)=(f(a))/(a)(lambda a)=(f(a))/(a)xf(x)=f(\lambda a) \leq \lambda f(a)=\frac{f(a)}{a}(\lambda a)=\frac{f(a)}{a} xf(x)=f(λa)λf(a)=f(a)a(λa)=f(a)ax
Analogously,
g ( x ) = g ( λ a ) λ g ( a ) = g ( a ) a ( λ a ) = g ( a ) a x g ( x ) = g ( λ a ) λ g ( a ) = g ( a ) a ( λ a ) = g ( a ) a x g(x)=g(lambda a) >= lambda g(a)=(g(a))/(a)(lambda a)=(g(a))/(a)xg(x)=g(\lambda a) \geq \lambda g(a)=\frac{g(a)}{a}(\lambda a)=\frac{g(a)}{a} xg(x)=g(λa)λg(a)=g(a)a(λa)=g(a)ax
But g ( a ) f ( a ) g ( a ) f ( a ) g(a) >= f(a)g(a) \geq f(a)g(a)f(a) implies g ( a ) a f ( a ) a g ( a ) a f ( a ) a (g(a))/(a) <= (f(a))/(a)\frac{g(a)}{a} \leq \frac{f(a)}{a}g(a)af(a)a, and for α [ g ( a ) a , f ( a ) a ] α g ( a ) a , f ( a ) a alpha in[(g(a))/(a),(f(a))/(a)]\alpha \in\left[\frac{g(a)}{a}, \frac{f(a)}{a}\right]α[g(a)a,f(a)a] it follows
g ( a ) a x α x f ( a ) a x , x [ a , 0 ] g ( a ) a x α x f ( a ) a x , x [ a , 0 ] (g(a))/(a)x >= alpha x >= (f(a))/(a)x,x in[a,0]\frac{g(a)}{a} x \geq \alpha x \geq \frac{f(a)}{a} x, x \in[a, 0]g(a)axαxf(a)ax,x[a,0]
and, consequently f ( x ) α x g ( x ) , x [ 0 , a ] f ( x ) α x g ( x ) , x [ 0 , a ] f(x) <= alpha x <= g(x),x in[0,a]f(x) \leq \alpha x \leq g(x), x \in[0, a]f(x)αxg(x),x[0,a].
Now, let x 0 [ a , 0 ) x 0 [ a , 0 ) x_(0)in[a,0)x_{0} \in[a, 0)x0[a,0) such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0).
Then
f ( a ) a f ( x 0 ) x 0 = g ( x 0 ) x 0 g ( a ) a f ( a ) a f x 0 x 0 = g x 0 x 0 g ( a ) a (f(a))/(a) <= (f(x_(0)))/(x_(0))=(g(x_(0)))/(x_(0)) <= (g(a))/(a)\frac{f(a)}{a} \leq \frac{f\left(x_{0}\right)}{x_{0}}=\frac{g\left(x_{0}\right)}{x_{0}} \leq \frac{g(a)}{a}f(a)af(x0)x0=g(x0)x0g(a)a
We are lead to f ( a ) g ( a ) f ( a ) g ( a ) f(a) >= g(a)f(a) \geq g(a)f(a)g(a), and because of hypothesis f ( a ) g ( a ) f ( a ) g ( a ) f(a) <= g(a)f(a) \leq g(a)f(a)g(a) it follows f ( a ) a = g ( a ) a f ( a ) a = g ( a ) a (f(a))/(a)=(g(a))/(a)\frac{f(a)}{a}=\frac{g(a)}{a}f(a)a=g(a)a.
Then h ( x ) = f ( a ) a x h ( x ) = f ( a ) a x h(x)=(f(a))/(a)xh(x)=\frac{f(a)}{a} xh(x)=f(a)ax is the only function having the graph between the graphs of f f fff and g g ggg on [ a , 0 ] [ a , 0 ] [a,0][a, 0][a,0].
Suppose now that f f fff is radiant, g g ggg is coradiant and f ( x ) g ( x ) f ( x ) g ( x ) f(x) >= g(x)f(x) \geq g(x)f(x)g(x) for all x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a]. Because f ( 0 ) 0 f ( 0 ) 0 f(0) <= 0f(0) \leq 0f(0)0 and g ( 0 ) 0 g ( 0 ) 0 g(0) >= 0g(0) \geq 0g(0)0 it follows f ( 0 ) = g ( 0 ) = 0 f ( 0 ) = g ( 0 ) = 0 f(0)=g(0)=0f(0)=g(0)=0f(0)=g(0)=0.
The following theorem holds.
Theorem 8. Let f , g : [ 0 , a ] R , f f , g : [ 0 , a ] R , f f,g:[0,a]rarrR,ff, g:[0, a] \rightarrow \mathbb{R}, ff,g:[0,a]R,f radiant, g g ggg coradiant and f ( x ) g ( x ) f ( x ) g ( x ) f(x) >= g(x)f(x) \geq g(x)f(x)g(x), x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a]. Then there exists at least a function h : [ 0 , a ] R h : [ 0 , a ] R h:[0,a]rarrRh:[0, a] \rightarrow \mathbb{R}h:[0,a]R having the form h ( x ) = α x , α R h ( x ) = α x , α R h(x)=alpha x,alpha inRh(x)=\alpha x, \alpha \in \mathbb{R}h(x)=αx,αR fixed, and such that
(22) f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . (22) f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . {:(22)f(x) >= h(x) >= g(x)","x in[0","a].:}\begin{equation*} f(x) \geq h(x) \geq g(x), x \in[0, a] . \tag{22} \end{equation*}(22)f(x)h(x)g(x),x[0,a].
If there exists x 0 ( 0 , a ] x 0 ( 0 , a ] x_(0)in(0,a]x_{0} \in(0, a]x0(0,a] such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0) then f ( x ) = g ( x ) f ( x ) = g ( x ) f(x)=g(x)f(x)=g(x)f(x)=g(x) for every x [ 0 , x 0 ] x 0 , x 0 x in[0,x_(0)]x \in\left[0, x_{0}\right]x[0,x0] and the only function h : [ 0 , a } R h : [ 0 , a } R h:[0,a}rarrRh:[0, a\} \rightarrow \mathbb{R}h:[0,a}R verifying 22 is h ( x ) = f ( x 0 ) x 0 x h ( x ) = f x 0 x 0 x h(x)=(f(x_(0)))/(x_(0))xh(x)=\frac{f\left(x_{0}\right)}{x_{0}} xh(x)=f(x0)x0x.
Proof. Let x 0 ( 0 , a ] x 0 ( 0 , a ] x_(0)in(0,a]x_{0} \in(0, a]x0(0,a] be such that f ( x 0 ) = g ( x 0 ) f x 0 = g x 0 f(x_(0))=g(x_(0))f\left(x_{0}\right)=g\left(x_{0}\right)f(x0)=g(x0). If x ( 0 , x 0 ] x 0 , x 0 x in(0,x_(0)]x \in\left(0, x_{0}\right]x(0,x0], by Lemma 3 one obtains
g ( x ) x g ( x 0 ) x 0 = f ( x 0 ) x 0 f ( x ) x . g ( x ) x g x 0 x 0 = f x 0 x 0 f ( x ) x . (g(x))/(x) >= (g(x_(0)))/(x_(0))=(f(x_(0)))/(x_(0)) >= (f(x))/(x).\frac{g(x)}{x} \geq \frac{g\left(x_{0}\right)}{x_{0}}=\frac{f\left(x_{0}\right)}{x_{0}} \geq \frac{f(x)}{x} .g(x)xg(x0)x0=f(x0)x0f(x)x.
It follows g ( x ) f ( x ) , x [ 0 , x 0 ] g ( x ) f ( x ) , x 0 , x 0 g(x) >= f(x),x in[0,x_(0)]g(x) \geq f(x), x \in\left[0, x_{0}\right]g(x)f(x),x[0,x0] and because f ( x ) g ( x ) f ( x ) g ( x ) f(x) >= g(x)f(x) \geq g(x)f(x)g(x) (by hypothesis) one obtains f ( x ) = g ( x ) f ( x ) = g ( x ) f(x)=g(x)f(x)=g(x)f(x)=g(x) for every x [ 0 , x 0 ] x 0 , x 0 x in[0,x_(0)]x \in\left[0, x_{0}\right]x[0,x0].
The line h ( x ) = f ( x 0 ) x 0 x ( = g ( x 0 ) x 0 x ) h ( x ) = f x 0 x 0 x = g x 0 x 0 x h(x)=(f(x_(0)))/(x_(0))x(=(g(x_(0)))/(x_(0))x)h(x)=\frac{f\left(x_{0}\right)}{x_{0}} x\left(=\frac{g\left(x_{0}\right)}{x_{0}} x\right)h(x)=f(x0)x0x(=g(x0)x0x) is between the graphs of f f fff and g g ggg over the interval [ 0 , x 0 ] 0 , x 0 [0,x_(0)]\left[0, x_{0}\right][0,x0].
For x > x 0 , x a x > x 0 , x a x > x_(0),x <= ax>x_{0}, x \leq ax>x0,xa
f ( x 0 ) x 0 f ( x ) x and f ( x 0 ) x 0 g ( x ) x f x 0 x 0 f ( x ) x  and  f x 0 x 0 g ( x ) x (f(x_(0)))/(x_(0)) <= (f(x))/(x)" and "(f(x_(0)))/(x_(0)) <= (g(x))/(x)\frac{f\left(x_{0}\right)}{x_{0}} \leq \frac{f(x)}{x} \text { and } \frac{f\left(x_{0}\right)}{x_{0}} \leq \frac{g(x)}{x}f(x0)x0f(x)x and f(x0)x0g(x)x
and consequently the function h ( x ) = f ( x 0 ) x 0 x , x [ 0 , a ] h ( x ) = f x 0 x 0 x , x [ 0 , a ] h(x)=(f(x_(0)))/(x_(0))x,x in[0,a]h(x)=\frac{f\left(x_{0}\right)}{x_{0}} x, x \in[0, a]h(x)=f(x0)x0x,x[0,a] verifies
f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . f(x) >= h(x) >= g(x),x in[0,a].f(x) \geq h(x) \geq g(x), x \in[0, a] .f(x)h(x)g(x),x[0,a].
Consider now the remaining case f ( x ) > g ( x ) , x ( 0 , a ] f ( x ) > g ( x ) , x ( 0 , a ] f(x) > g(x),x in(0,a]f(x)>g(x), x \in(0, a]f(x)>g(x),x(0,a]. Because
g ( a ) a g ( x ) x < f ( x ) x f ( a ) a , x ( 0 , a ] g ( a ) a g ( x ) x < f ( x ) x f ( a ) a , x ( 0 , a ] (g(a))/(a) <= (g(x))/(x) < (f(x))/(x) <= (f(a))/(a),x in(0,a]\frac{g(a)}{a} \leq \frac{g(x)}{x}<\frac{f(x)}{x} \leq \frac{f(a)}{a}, x \in(0, a]g(a)ag(x)x<f(x)xf(a)a,x(0,a]
it follows that inf { f ( x ) x : x ( 0 , a ] } inf f ( x ) x : x ( 0 , a ] i n f{(f(x))/(x):x in(0,a]}\inf \left\{\frac{f(x)}{x}: x \in(0, a]\right\}inf{f(x)x:x(0,a]} and sup { g ( x ) x : x ( 0 , a ] } sup g ( x ) x : x ( 0 , a ] s u p{(g(x))/(x):x in(0,a]}\sup \left\{\frac{g(x)}{x}: x \in(0, a]\right\}sup{g(x)x:x(0,a]} are finite, and
sup 0 < x a g ( x ) x inf 0 < x a f ( x ) x . sup 0 < x a g ( x ) x inf 0 < x a f ( x ) x . s u p_(0 < x <= a)(g(x))/(x) <= i n f_(0 < x <= a)(f(x))/(x).\sup _{0<x \leq a} \frac{g(x)}{x} \leq \inf _{0<x \leq a} \frac{f(x)}{x} .sup0<xag(x)xinf0<xaf(x)x.
By considering α [ sup 0 < x a g ( x ) x , inf 0 < x a f ( x ) x ] α sup 0 < x a g ( x ) x , inf 0 < x a f ( x ) x alpha in[s u p_(0 < x <= a)(g(x))/(x),i n f_(0 < x <= a)(f(x))/(x)]\alpha \in\left[\sup _{0<x \leq a} \frac{g(x)}{x}, \inf _{0<x \leq a} \frac{f(x)}{x}\right]α[sup0<xag(x)x,inf0<xaf(x)x], the line h ( x ) = α x , x [ 0 , a ] h ( x ) = α x , x [ 0 , a ] h(x)=alpha x,x in[0,a]h(x)=\alpha x, x \in[0, a]h(x)=αx,x[0,a] lies between the graphs of f f fff and g g ggg, i.e.
f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . f(x) >= h(x) >= g(x),x in[0,a].f(x) \geq h(x) \geq g(x), x \in[0, a] .f(x)h(x)g(x),x[0,a].
Remark 9. The result in Theorem 8 is valid also if f , g : ( , 0 ] R f , g : ( , 0 ] R f,g:(-oo,0]rarrRf, g:(-\infty, 0] \rightarrow \mathbb{R}f,g:(,0]R. In this case there exists α [ lim a sup a < x 0 g ( x ) x , lim a inf a < x 0 f ( x ) x ] α lim a sup a < x 0 g ( x ) x , lim a inf a < x 0 f ( x ) x alpha in[lim_(a rarr-oo)s u p_(a < x <= 0)(g(x))/(x),lim_(a rarr-oo)i n f_(a < x <= 0)(f(x))/(x)]\alpha \in\left[\lim _{a \rightarrow-\infty} \sup _{a<x \leq 0} \frac{g(x)}{x}, \lim _{a \rightarrow-\infty} \inf _{a<x \leq 0} \frac{f(x)}{x}\right]α[limasupa<x0g(x)x,limainfa<x0f(x)x] such that the function h ( x ) = α x h ( x ) = α x h(x)=alpha xh(x)=\alpha xh(x)=αx satisfies the inequalities
f ( x ) h ( x ) g ( x ) , x ( , 0 ] . f ( x ) h ( x ) g ( x ) , x ( , 0 ] . f(x) <= h(x) <= g(x),x in(-oo,0].f(x) \leq h(x) \leq g(x), x \in(-\infty, 0] .f(x)h(x)g(x),x(,0].
Now consider I = [ a , b ] I = [ a , b ] I=[a,b]I=[a, b]I=[a,b] where a < 0 < b a < 0 < b a < 0 < ba<0<ba<0<b, or I = R I = R I=RI=\mathbb{R}I=R.
By the above results it follows:
Corollary 10. a) Let f , g : [ a , b ] R , a < 0 < b f , g : [ a , b ] R , a < 0 < b f,g:[a,b]rarrR,a < 0 < bf, g:[a, b] \rightarrow \mathbb{R}, a<0<bf,g:[a,b]R,a<0<b, f f fff radiant, g g ggg coradiant and such that f ( x ) g ( x ) , x [ a , b ] f ( x ) g ( x ) , x [ a , b ] f(x) <= g(x),x in[a,b]f(x) \leq g(x), x \in[a, b]f(x)g(x),x[a,b]. Then there exists h : [ a , b ] R h : [ a , b ] R h:[a,b]rarrRh:[a, b] \rightarrow \mathbb{R}h:[a,b]R,
h ( x ) = { α x , x [ a , 0 ] , α [ g ( a ) a , f ( a ) a ] , β x , x ( 0 , b ] , β [ f ( b ) b , g ( b ) b ] . h ( x ) = α x , x [ a , 0 ] , α g ( a ) a , f ( a ) a , β x , x ( 0 , b ] , β f ( b ) b , g ( b ) b . h(x)={[alpha x","x in[a","0]","alpha in[(g(a))/(a),(f(a))/(a)]","],[beta x","x in(0","b]","beta in[(f(b))/(b),(g(b))/(b)].]:}h(x)=\left\{\begin{array}{l} \alpha x, x \in[a, 0], \alpha \in\left[\frac{g(a)}{a}, \frac{f(a)}{a}\right], \\ \beta x, x \in(0, b], \beta \in\left[\frac{f(b)}{b}, \frac{g(b)}{b}\right] . \end{array}\right.h(x)={αx,x[a,0],α[g(a)a,f(a)a],βx,x(0,b],β[f(b)b,g(b)b].
such that f ( x ) h ( x ) g ( x ) , x [ a , b ] f ( x ) h ( x ) g ( x ) , x [ a , b ] f(x) <= h(x) <= g(x),x in[a,b]f(x) \leq h(x) \leq g(x), x \in[a, b]f(x)h(x)g(x),x[a,b].
b) Let f , g : R R f , g : R R f,g:RrarrRf, g: \mathbb{R} \rightarrow \mathbb{R}f,g:RR, f f fff radiant, g g ggg coradiant, f ( x ) g ( x ) f ( x ) g ( x ) f(x) <= g(x)f(x) \leq g(x)f(x)g(x), for all x R x R x inRx \in \mathbb{R}xR.
Then there exists h : R R h : R R h:RrarrRh: \mathbb{R} \rightarrow \mathbb{R}h:RR of the form
h ( x ) = { α x , x ( , 0 ] , β x , x ( 0 , ) , h ( x ) = α x , x ( , 0 ] , β x , x ( 0 , ) , h(x)={[alpha x","x in(-oo","0]","],[beta x","x in(0","oo)","]:}h(x)=\left\{\begin{array}{c} \alpha x, x \in(-\infty, 0], \\ \beta x, x \in(0, \infty), \end{array}\right.h(x)={αx,x(,0],βx,x(0,),
where
α [ lim a sup x a g ( x ) x , lim a inf x a f ( x ) x ] α lim a sup x a g ( x ) x , lim a inf x a f ( x ) x alpha in[lim_(a rarr-oo)s u p_(x >= a)(g(x))/(x),lim_(a rarr-oo)i n f_(x >= a)(f(x))/(x)]\alpha \in\left[\lim _{a \rightarrow-\infty} \sup _{x \geq a} \frac{g(x)}{x}, \lim _{a \rightarrow-\infty} \inf _{x \geq a} \frac{f(x)}{x}\right]α[limasupxag(x)x,limainfxaf(x)x]
and
β [ lim a sup x a f ( x ) x , lim a inf x a g ( x ) x ] , β lim a sup x a f ( x ) x , lim a inf x a g ( x ) x , beta in[lim_(a rarr oo)s u p_(x <= a)(f(x))/(x),lim_(a rarr oo)i n f_(x <= a)(g(x))/(x)],\beta \in\left[\lim _{a \rightarrow \infty} \sup _{x \leq a} \frac{f(x)}{x}, \lim _{a \rightarrow \infty} \inf _{x \leq a} \frac{g(x)}{x}\right],β[limasupxaf(x)x,limainfxag(x)x],
verifying the inequalities:
f ( x ) h ( x ) g ( x ) , x R f ( x ) h ( x ) g ( x ) , x R f(x) <= h(x) <= g(x),x inRf(x) \leq h(x) \leq g(x), x \in \mathbb{R}f(x)h(x)g(x),xR
A similar result is valid if f ( x ) g ( x ) , f f ( x ) g ( x ) , f f(x) >= g(x),ff(x) \geq g(x), ff(x)g(x),f radiant and g g ggg coradiant on [ a , b ] [ a , b ] [a,b][a, b][a,b], a < 0 < b a < 0 < b a < 0 < ba<0<ba<0<b, respectively on R R R\mathbb{R}R.
In Corollary 11 a), if [ g ( a ) a , f ( a ) a ] [ f ( b ) b , g ( b ) b ] g ( a ) a , f ( a ) a f ( b ) b , g ( b ) b [(g(a))/(a),(f(a))/(a)]nn[(f(b))/(b),(g(b))/(b)]!=O/\left[\frac{g(a)}{a}, \frac{f(a)}{a}\right] \cap\left[\frac{f(b)}{b}, \frac{g(b)}{b}\right] \neq \emptyset[g(a)a,f(a)a][f(b)b,g(b)b] and μ μ mu\muμ is a number from this set, then h ( x ) = μ x , x [ a , b ] h ( x ) = μ x , x [ a , b ] h(x)=mu x,x in[a,b]h(x)=\mu x, x \in[a, b]h(x)=μx,x[a,b] verify
f ( x ) μ x g ( x ) , x [ a , b ] . f ( x ) μ x g ( x ) , x [ a , b ] . f(x) <= mu x <= g(x),x in[a,b].f(x) \leq \mu x \leq g(x), x \in[a, b] .f(x)μxg(x),x[a,b].
A similar result follows in the case b).
If f : I R ( 0 I ) f : I R ( 0 I ) f:I rarrRquad(0in I)f: I \rightarrow \mathbb{R} \quad(0 \in I)f:IR(0I), and f f fff is a convex (concave) function on I , f ( 0 ) = 0 I , f ( 0 ) = 0 I,f(0)=0I, f(0)=0I,f(0)=0, then f f fff is radiant (coradiant). The above results may be enounced for convex and concave function defined on I I III, vanishing at zero.
Examples. 1 0 1 0 1^(0)1^{0}10 Let f m , g : [ 0 , 2 ] R f m , g : [ 0 , 2 ] R f_(m),g:[0,2]rarrRf_{m}, g:[0,2] \rightarrow \mathbb{R}fm,g:[0,2]R be the functions defined by
f m ( x ) = { m x 3 , x [ 0 , 1 ] m x , x ( 1 , 2 ] , m R , g ( x ) = { x 2 + 2 , x [ 0 , 1 ] x , x ( 1 , 2 ] f m ( x ) = m x 3 , x [ 0 , 1 ] m x , x ( 1 , 2 ] , m R , g ( x ) = x 2 + 2 , x [ 0 , 1 ] x , x ( 1 , 2 ] {:[f_(m)(x)={[mx^(3)",",x in[0","1]],[mx",",x in(1","2]],m inR,:}],[g(x)={[-x^(2)+2",",x in[0","1]],[x",",x in(1","2]]:}]:}\begin{aligned} & f_{m}(x)=\left\{\begin{array}{ll} m x^{3}, & x \in[0,1] \\ m x, & x \in(1,2] \end{array}, m \in \mathbb{R},\right. \\ & g(x)= \begin{cases}-x^{2}+2, & x \in[0,1] \\ x, & x \in(1,2]\end{cases} \end{aligned}fm(x)={mx3,x[0,1]mx,x(1,2],mR,g(x)={x2+2,x[0,1]x,x(1,2]
Then, for m > 0 m > 0 m > 0m>0m>0 the function f m f m f_(m)f_{m}fm is radiant, g g ggg is coradiant and f m ( x ) g ( x ) f m ( x ) g ( x ) f_(m)(x) <= g(x)f_{m}(x) \leq g(x)fm(x)g(x), x [ 0 , 2 ] x [ 0 , 2 ] x in[0,2]x \in[0,2]x[0,2]. Every function h : [ 0 , 2 ] R , h ( x ) = α x h : [ 0 , 2 ] R , h ( x ) = α x h:[0,2]rarrR,h(x)=alpha xh:[0,2] \rightarrow \mathbb{R}, h(x)=\alpha xh:[0,2]R,h(x)=αx, where α [ m , 1 ] α [ m , 1 ] alpha in[m,1]\alpha \in[m, 1]α[m,1] verifies
f m ( x ) h ( x ) g ( x ) , x [ 0 , 2 ] f m ( x ) h ( x ) g ( x ) , x [ 0 , 2 ] f_(m)(x) <= h(x) <= g(x),x in[0,2]f_{m}(x) \leq h(x) \leq g(x), x \in[0,2]fm(x)h(x)g(x),x[0,2]
Also, for m = 1 , f 1 ( x ) g ( x ) , x [ 0 , 2 ] m = 1 , f 1 ( x ) g ( x ) , x [ 0 , 2 ] m=1,f_(1)(x) <= g(x),x in[0,2]m=1, f_{1}(x) \leq g(x), x \in[0,2]m=1,f1(x)g(x),x[0,2] and because f 1 ( 1 ) = g ( 1 ) = 1 f 1 ( 1 ) = g ( 1 ) = 1 f_(1)(1)=g(1)=1f_{1}(1)=g(1)=1f1(1)=g(1)=1 one obtains that h ( x ) = x h ( x ) = x h(x)=xh(x)=xh(x)=x is the unique function verifying f 1 ( x ) h ( x ) g ( x ) f 1 ( x ) h ( x ) g ( x ) f_(1)(x) <= h(x) <= g(x)f_{1}(x) \leq h(x) \leq g(x)f1(x)h(x)g(x), x [ 0 , 2 ] x [ 0 , 2 ] x in[0,2]x \in[0,2]x[0,2]. Consequently Theorem 6 is fulfilled.
2 0 2 0 2^(0)2^{0}20 Let a > 1 , m ( 0 , 1 ] a > 1 , m ( 0 , 1 ] a > 1,m in(0,1]a>1, m \in(0,1]a>1,m(0,1] and let f m , g a : [ 0 , ) R f m , g a : [ 0 , ) R f_(m),g_(a):[0,oo)rarrRf_{m}, g_{a}:[0, \infty) \rightarrow \mathbb{R}fm,ga:[0,)R, be the functions:
c f m ( x ) = { m x 3 a 2 , x [ 0 , a ] , m x , x ( a + ) , g a ( x ) = { x 2 + ( 1 + a ) x , x [ 0 , a ] x , x ( a , + ) c f m ( x ) = m x 3 a 2 ,      x [ 0 , a ] , m x ,      x ( a + ) , g a ( x ) = x 2 + ( 1 + a ) x ,      x [ 0 , a ] x ,      x ( a , + ) cf_(m)(x)={[(mx^(3))/(a^(2))",",x in[0","a]","],[mx",",x in(a+oo)","]g_(a)(x)={[-x^(2)+(1+a)x",",x in[0","a]],[x",",x in(a","+oo)]:}c f_{m}(x)=\left\{\begin{array}{ll} \frac{m x^{3}}{a^{2}}, & x \in[0, a], \\ m x, & x \in(a+\infty), \end{array} g_{a}(x)= \begin{cases}-x^{2}+(1+a) x, & x \in[0, a] \\ x, & x \in(a,+\infty)\end{cases}\right.cfm(x)={mx3a2,x[0,a],mx,x(a+),ga(x)={x2+(1+a)x,x[0,a]x,x(a,+)
Then f m ( x ) g a ( x ) f m ( x ) g a ( x ) f_(m)(x) <= g_(a)(x)f_{m}(x) \leq g_{a}(x)fm(x)ga(x), for x [ 0 , ) , f m x [ 0 , ) , f m x in[0,oo),f_(m)x \in[0, \infty), f_{m}x[0,),fm is radiant and g a g a g_(a)g_{a}ga is coradiant.
For every α [ m , 1 ] α [ m , 1 ] alpha in[m,1]\alpha \in[m, 1]α[m,1] one obtains
f m ( x ) h ( x ) = α x g ( x ) , x [ 0 , ) f m ( x ) h ( x ) = α x g ( x ) , x [ 0 , ) f_(m)(x) <= h(x)=alpha x <= g(x),x in[0,oo)f_{m}(x) \leq h(x)=\alpha x \leq g(x), x \in[0, \infty)fm(x)h(x)=αxg(x),x[0,)
For m = 1 , f 1 ( a ) = g a ( a ) = a m = 1 , f 1 ( a ) = g a ( a ) = a m=1,f_(1)(a)=g_(a)(a)=am=1, f_{1}(a)=g_{a}(a)=am=1,f1(a)=ga(a)=a and consequently h ( x ) = x h ( x ) = x h(x)=xh(x)=xh(x)=x is the only function verifying f 1 ( x ) h ( x ) = x g ( x ) , x [ 0 , ) f 1 ( x ) h ( x ) = x g ( x ) , x [ 0 , ) f_(1)(x) <= h(x)=x <= g(x),x in[0,oo)f_{1}(x) \leq h(x)=x \leq g(x), x \in[0, \infty)f1(x)h(x)=xg(x),x[0,). Theorem 7 is fulfilled.
3 0 3 0 3^(0)3^{0}30 Let f , g : [ 2 , 0 ] R , f ( x ) = x 2 + x f , g : [ 2 , 0 ] R , f ( x ) = x 2 + x f,g:[-2,0]rarrR,f(x)=x^(2)+xf, g:[-2,0] \rightarrow \mathbb{R}, f(x)=x^{2}+xf,g:[2,0]R,f(x)=x2+x and g ( x ) = x 2 4 x g ( x ) = x 2 4 x g(x)=-x^(2)-4xg(x)=-x^{2}-4 xg(x)=x24x. Then f f fff is radiant, g g ggg is coradiant and f ( x ) g ( x ) , x [ 2 , 0 ] f ( x ) g ( x ) , x [ 2 , 0 ] f(x) <= g(x),x in[-2,0]f(x) \leq g(x), x \in[-2,0]f(x)g(x),x[2,0]. Every line h ( x ) = α x h ( x ) = α x h(x)=alpha xh(x)=\alpha xh(x)=αx, where α [ 2 , 1 ] α [ 2 , 1 ] alpha in[-2,-1]\alpha \in[-2,-1]α[2,1] has the graph between the graphs of f f fff and g g ggg.
Now let f , g : [ 3 , 0 ] R f , g : [ 3 , 0 ] R f,g:[-3,0]rarrRf, g:[-3,0] \rightarrow \mathbb{R}f,g:[3,0]R be the functions
f ( x ) = { x 2 , x [ 3 , 1 ] x 2 + x 2 , x ( 1 , 0 ] g ( x ) = { x 2 , x [ 3 , 1 ] 4 x 2 9 2 x , x ( 1 , 0 ] f ( x ) = x 2 , x [ 3 , 1 ] x 2 + x 2 , x ( 1 , 0 ] g ( x ) = x 2 , x [ 3 , 1 ] 4 x 2 9 2 x , x ( 1 , 0 ] {:[f(x)={[(-x)/(2)",",x in[-3","-1]],[x^(2)+(x)/(2)",",x in(-1","0]]:}],[g(x)={[(-x)/(2)",",x in[-3","-1]],[-4x^(2)(9)/(2)x",",x in(-1","0]]:}]:}\begin{aligned} & f(x)= \begin{cases}\frac{-x}{2}, & x \in[-3,-1] \\ x^{2}+\frac{x}{2}, & x \in(-1,0]\end{cases} \\ & g(x)= \begin{cases}\frac{-x}{2}, & x \in[-3,-1] \\ -4 x^{2} \frac{9}{2} x, & x \in(-1,0]\end{cases} \end{aligned}f(x)={x2,x[3,1]x2+x2,x(1,0]g(x)={x2,x[3,1]4x292x,x(1,0]
Then f f fff is radiant, g g ggg is coradiant and f ( x ) g ( x ) , x [ 3 , 0 ] f ( x ) g ( x ) , x [ 3 , 0 ] f(x) <= g(x),x in[-3,0]f(x) \leq g(x), x \in[-3,0]f(x)g(x),x[3,0].
The only function h h hhh such that f ( x ) h ( x ) g ( x ) , x [ 3 , 0 ] f ( x ) h ( x ) g ( x ) , x [ 3 , 0 ] f(x) <= h(x) <= g(x),x in[-3,0]f(x) \leq h(x) \leq g(x), x \in[-3,0]f(x)h(x)g(x),x[3,0] is h ( x ) = 1 2 x h ( x ) = 1 2 x h(x)=-(1)/(2)xh(x)= -\frac{1}{2} xh(x)=12x. Thus Theorem 8 is fulfilled.
4 0 4 0 4^(0)4^{0}40 Let f m , g : [ 0 , a ] R , ( a > 0 , m 0 ) f m , g : [ 0 , a ] R , ( a > 0 , m 0 ) f_(m),g:[0,a]rarrR,(a > 0,m >= 0)f_{m}, g:[0, a] \rightarrow \mathbb{R},(a>0, m \geq 0)fm,g:[0,a]R,(a>0,m0) be the functions
f m ( x ) = x 2 + m x g ( x ) = x 3 f m ( x ) = x 2 + m x g ( x ) = x 3 {:[f_(m)(x)=x^(2)+mx],[g(x)=-x^(3)]:}\begin{aligned} f_{m}(x) & =x^{2}+m x \\ g(x) & =-x^{3} \end{aligned}fm(x)=x2+mxg(x)=x3
Then f m f m f_(m)f_{m}fm is radiant, g g ggg is coradiant and f ( x ) g ( x ) , x [ 0 , a ] f ( x ) g ( x ) , x [ 0 , a ] f(x) >= g(x),x in[0,a]f(x) \geq g(x), x \in[0, a]f(x)g(x),x[0,a]. The function h ( x ) = α x h ( x ) = α x h(x)=alpha xh(x)=\alpha xh(x)=αx, where α [ 0 , m ] α [ 0 , m ] alpha in[0,m]\alpha \in[0, m]α[0,m] is such that
f ( x ) h ( x ) g ( x ) , x [ 0 , a ] f ( x ) h ( x ) g ( x ) , x [ 0 , a ] f(x) >= h(x) >= g(x),x in[0,a]f(x) \geq h(x) \geq g(x), x \in[0, a]f(x)h(x)g(x),x[0,a]
For m = 0 m = 0 m=0m=0m=0, the function h ( x ) = 0 h ( x ) = 0 h(x)=0h(x)=0h(x)=0 verify
f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . f ( x ) h ( x ) g ( x ) , x [ 0 , a ] . f(x) >= h(x) >= g(x),x in[0,a].f(x) \geq h(x) \geq g(x), x \in[0, a] .f(x)h(x)g(x),x[0,a].
5 0 5 0 5^(0)5^{0}50 Let f , g : [ 0 , 2 ] R f , g : [ 0 , 2 ] R f,g:[0,2]rarrRf, g:[0,2] \rightarrow \mathbb{R}f,g:[0,2]R be given by
f ( x ) = { x , x [ 0 , 1 ] 2 ( x 1 ) , x ( 1 , 2 ] g ( x ) = { x , x [ 0 , 1 ] 3 ( x 1 ) , x ( 1 , 2 ] f ( x ) = x , x [ 0 , 1 ] 2 ( x 1 ) , x ( 1 , 2 ] g ( x ) = x , x [ 0 , 1 ] 3 ( x 1 ) , x ( 1 , 2 ] {:[f(x)={[x",",x in[0","1]],[2(x-1)",",x in(1","2]]:}],[g(x)={[x",",x in[0","1]],[-3(x-1)",",x in(1","2]]:}]:}\begin{aligned} & f(x)= \begin{cases}x, & x \in[0,1] \\ 2(x-1), & x \in(1,2]\end{cases} \\ & g(x)= \begin{cases}x, & x \in[0,1] \\ -3(x-1), & x \in(1,2]\end{cases} \end{aligned}f(x)={x,x[0,1]2(x1),x(1,2]g(x)={x,x[0,1]3(x1),x(1,2]
Then f f fff is radiant, g g ggg is coradiant, f ( 1 ) = g ( 1 ) f ( 1 ) = g ( 1 ) f(1)=g(1)f(1)=g(1)f(1)=g(1) and f ( x ) g ( x ) , x [ 0 , 2 ] f ( x ) g ( x ) , x [ 0 , 2 ] f(x) >= g(x),x in[0,2]f(x) \geq g(x), x \in[0,2]f(x)g(x),x[0,2]. The function h ( x ) = x , x [ 0 , 2 ] h ( x ) = x , x [ 0 , 2 ] h(x)=x,x in[0,2]h(x)=x, x \in[0,2]h(x)=x,x[0,2] is the only function having the graph between graphs of f f fff and g g ggg. Thus Theorem 8 is fulfilled.

4. APPLICATIONS

Let I = [ 0 , a ] I = [ 0 , a ] I=[0,a]I=[0, a]I=[0,a] and let f : I R f : I R f:I rarrRf: I \rightarrow \mathbb{R}f:IR. The function f f fff is called ε ε epsi\varepsilonε-positively homogeneous if | f ( λ x ) λ f ( x ) | < ε | f ( λ x ) λ f ( x ) | < ε |f(lambda x)-lambda f(x)| < epsi|f(\lambda x)-\lambda f(x)|<\varepsilon|f(λx)λf(x)|<ε, for all x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a] and λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1].
The function f : [ 0 , a ] R f : [ 0 , a ] R f:[0,a]rarrRf:[0, a] \rightarrow \mathbb{R}f:[0,a]R is both radiant and coradiant iff f f fff is positively homogeneous, i.e., f ( λ x ) = λ f ( x ) f ( λ x ) = λ f ( x ) f(lambda x)=lambda f(x)f(\lambda x)=\lambda f(x)f(λx)=λf(x), for all x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a] and λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1].
The above results gives sufficient stability conditions of Hyers-Ulam type for positively homogeneous functions.
Corollary 11. Let f : [ 0 , a ] R f : [ 0 , a ] R f:[0,a]rarrRf:[0, a] \rightarrow \mathbb{R}f:[0,a]R such that f f fff is radiant or coradiant. Let ε > 0 ε > 0 epsi > 0\varepsilon>0ε>0 be a real number. In order to obtain
| f ( λ x ) λ f ( x ) | < ε | f ( λ x ) λ f ( x ) | < ε |f(lambda x)-lambda f(x)| < epsi|f(\lambda x)-\lambda f(x)|<\varepsilon|f(λx)λf(x)|<ε
for all x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a] and λ [ 0 , 1 ] λ [ 0 , 1 ] lambda in[0,1]\lambda \in[0,1]λ[0,1] it is sufficient that
| f ( x ) f ( a ) a x | < ε , x [ 0 , a ] . f ( x ) f ( a ) a x < ε , x [ 0 , a ] . |f(x)-(f(a))/(a)x| < epsi,x in[0,a].\left|f(x)-\frac{f(a)}{a} x\right|<\varepsilon, x \in[0, a] .|f(x)f(a)ax|<ε,x[0,a].
Proof. Let f : [ 0 , a ] R f : [ 0 , a ] R f:[0,a]rarrRf:[0, a] \rightarrow \mathbb{R}f:[0,a]R be radiant. By Lemma 3 and Theorem 6 it follows f ( x ) f ( a ) a x f ( x ) f ( a ) a x f(x) <= (f(a))/(a)xf(x) \leq \frac{f(a)}{a} xf(x)f(a)ax, for all x [ 0 , a ] x [ 0 , a ] x in[0,a]x \in[0, a]x[0,a].
Then
0 λ f ( x ) f ( λ x ) λ f ( a ) a x f ( λ x ) = = f ( a ) a ( λ x ) f ( λ x ) = = | f ( a ) a ( λ x ) f ( λ x ) | < ε . 0 λ f ( x ) f ( λ x ) λ f ( a ) a x f ( λ x ) = = f ( a ) a ( λ x ) f ( λ x ) = = f ( a ) a ( λ x ) f ( λ x ) < ε . {:[0 <= lambda f(x)-f(lambda x) <= lambda(f(a))/(a)x-f(lambda x)=],[=(f(a))/(a)(lambda x)-f(lambda x)=],[=|(f(a))/(a)(lambda x)-f(lambda x)| < epsi.]:}\begin{aligned} 0 & \leq \lambda f(x)-f(\lambda x) \leq \lambda \frac{f(a)}{a} x-f(\lambda x)= \\ & =\frac{f(a)}{a}(\lambda x)-f(\lambda x)= \\ & =\left|\frac{f(a)}{a}(\lambda x)-f(\lambda x)\right|<\varepsilon . \end{aligned}0λf(x)f(λx)λf(a)axf(λx)==f(a)a(λx)f(λx)==|f(a)a(λx)f(λx)|<ε.
Consequently
λ f ( x ) f ( λ x ) = | f ( λ x ) λ f ( x ) | < ε λ f ( x ) f ( λ x ) = | f ( λ x ) λ f ( x ) | < ε lambda f(x)-f(lambda x)=|f(lambda x)-lambda f(x)| < epsi\lambda f(x)-f(\lambda x)=|f(\lambda x)-\lambda f(x)|<\varepsilonλf(x)f(λx)=|f(λx)λf(x)|<ε
Now, if f f fff is coradiant, then by Theorem 6 it follows f ( x ) f ( a ) a x f ( x ) f ( a ) a x f(x) >= (f(a))/(a)xf(x) \geq \frac{f(a)}{a} xf(x)f(a)ax and
0 f ( λ x ) λ f ( x ) f ( λ x ) λ f ( a ) a x = f ( λ x ) f ( a ) a λ x = | f ( λ x ) f ( a ) a λ x | < ε 0 f ( λ x ) λ f ( x ) f ( λ x ) λ f ( a ) a x = f ( λ x ) f ( a ) a λ x = f ( λ x ) f ( a ) a λ x < ε {:[0 <= f(lambda x)-lambda f(x) <= f(lambda x)-lambda(f(a))/(a)x],[=f(lambda x)-(f(a))/(a)lambda x=|f(lambda x)-(f(a))/(a)lambda x| < epsi]:}\begin{aligned} 0 & \leq f(\lambda x)-\lambda f(x) \leq f(\lambda x)-\lambda \frac{f(a)}{a} x \\ & =f(\lambda x)-\frac{f(a)}{a} \lambda x=\left|f(\lambda x)-\frac{f(a)}{a} \lambda x\right|<\varepsilon \end{aligned}0f(λx)λf(x)f(λx)λf(a)ax=f(λx)f(a)aλx=|f(λx)f(a)aλx|<ε
Consequently
f ( λ x ) λ f ( x ) = | f ( λ x ) λ f ( x ) | < ε f ( λ x ) λ f ( x ) = | f ( λ x ) λ f ( x ) | < ε f(lambda x)-lambda f(x)=|f(lambda x)-lambda f(x)| < epsif(\lambda x)-\lambda f(x)=|f(\lambda x)-\lambda f(x)|<\varepsilonf(λx)λf(x)=|f(λx)λf(x)|<ε
if
| f ( x ) f ( a ) a x | < ε . f ( x ) f ( a ) a x < ε . |f(x)-(f(a))/(a)x| < epsi.\left|f(x)-\frac{f(a)}{a} x\right|<\varepsilon .|f(x)f(a)ax|<ε.
By Theorem 6, it follows that if f : [ 0 , a ] R f : [ 0 , a ] R f:[0,a]rarrRf:[0, a] \rightarrow \mathbb{R}f:[0,a]R is positively homogeneous, x 0 ( 0 , a ] x 0 ( 0 , a ] x_(0)in(0,a]x_{0} \in(0, a]x0(0,a] and f ( x 0 ) f x 0 f(x_(0))f\left(x_{0}\right)f(x0) is given, then the function is exactly f ( x ) = f ( x 0 ) x 0 x , x [ 0 , a ] f ( x ) = f x 0 x 0 x , x [ 0 , a ] f(x)=(f(x_(0)))/(x_(0))x,x in[0,a]f(x)=\frac{f\left(x_{0}\right)}{x_{0}} x, x \in [0, a]f(x)=f(x0)x0x,x[0,a].

REFERENCES

[1] K. Baron, J. Matkowski and K. Nikodem, A Sandwich with convexity, Mathematica Panonica, 5 (1994) no. 1, pp. 139-144.
[2] W. Förg-Rob, K. Nicodem and Z. Páles, Separation by monotonic functions, Mathematica Panonica, 7 (1996) no. 2, pp. 191-196.
[3] B. Fuchssteiner and W. Lusky, Convex Cones, North Holland Math. Stud., 56 (North Holland, Amsterdam, 1981).
[4] J.A. JoHNSON, Banach spaces of Lipschitz functions and vector-valued Lipschitz functions, Trans. Amer. Math. Soc., 148(1970), pp. 147-169. 준
[5] E.J. McShane, Extension of range of functions, Bull. Amer. Math. Soc., 40 (1934), pp. 837-842. 중
[6] C. Mustăţa, Norm preserving extension of starshaped Lipschitz functions, Mathematica (Cluj), 19(42)2 (1977), pp. 183-187.
[7] C. Mustăţa, Extensions of semi-Lipschitz functions on quasi-metric spaces, Rev. Anal. Numér, Théor. Approx., 30 (2001) no. 1, pp. 61-67. ©
[8] C. Mustăta, On the extensions preserving the shape of semi-Hölder function, Results. Math., 63 (2013), pp. 425-433. [ x x x\boldsymbol{x}x
[9] K. Nikodem and S. Wasowicz, A sandwich theorem and Hyers-Ulam stability of affine functions, Aequationes Math., 49 (1995), pp. 160-164. 짖
[10] A.M. Rubinov, Abstract Convexity and Global Optimization, Kluwer Academic Publisher, Boston-Dordrecht-London, 2000.
[11] A.M. Rubinov and A.P. Shveidel, Radiant and star-shaped functions, Pacific Journal of Optimization, 3 (2007) no. 1, pp. 193-212.
[12] S. Simons, The asymmetric sandwich theorem, Journal of Convex Analysis, 20 (2013) no. 1, pp. 107-124.
[13] S. Suzuki and D. Kuroiwa, Sandwich theorem for quasiconvex functions and its applications, J. Math. Anal. Appl., 379 (2011), pp. 649-655. 짖
[14] Á. SzÁz, The infimal convolution can be used to derive extensions theorems from sandwich ones, Acta Sci. Math. (Szeged), 76 (2010), pp. 489-499.
Received by the editors: December 9, 2014.
2015

Related Posts