Rev. Anal. Num´er. Th´eor. Approx., vol. 30 (2001) no. 1, pp. 61–67 ictp.acad.ro/jnaat
EXTENSIONS OF SEMI-LIPSCHITZ FUNCTIONS ON QUASI-METRIC SPACES
COSTIC ˘A MUST ˘AT¸ A
Dedicated to the memory of Acad. Tiberiu Popoviciu
Abstract. The aim of this note is to prove an extension theorem for semi- Lipschitz real functions defined on quasi-metric spaces, similar to McShane ex- tension theorem for real-valued Lipschitz functions defined on a metric space ([2], [4]).
MSC 2000. 46A22, 26A16, 26A48.
1. INTRODUCTION
Let X be a nonvoid set. A quasi-metric on X is a function d:X×X → [0,∞) satisfying the conditions
d(x, y) =d(y, x) = 0⇐⇒x=y; x, y∈X, (i)
d(x, y)≤d(x, z) +d(z, y), x, y, z∈X.
(ii)
If d is a quasi-metric on X, then the pair (X, d) is called a quasi-metric space.
The conjugate of quasi-metricd, denoted by d−1 is defined by d−1(x, y) = d(y, x), x, y∈X.
Obviously the functionds:X×X→[0,∞) defined by ds(x, y) = max
d(x, y), d−1(x, y) ; x, y∈X is a metric on X.
“T. Popoviciu” Institute of Numerical Analysis, P.O. Box 68–1, 3400 Cluj-Napoca, Ro- mania, e-mail: [email protected].
quasi-metric.
Let (X, d) be a quasi-metric space. A function f :X → R is called semi- Lipschitz if there exists a constant K≥0 so that
(1) f(x)−f(y)≤K·d(x, y),
for all x, y∈X. The numberK ≥0 in (1) is called a semi-Lipschitz constant forf.
For a quasi-metric space (X, d) the real-valued functionf :X →R is said to be≤d-increasing if
(2) d(x, y) = 0 implies f(x)−f(y)≤0, x, y∈X or equivalently,
(3) f(x)−f(y)>0 implies d(x, y)>0, x, y∈X.
Note that every semi-Lipschitz function on quasi-metric space (X, d) is≤d- increasing (see (1)).
For a semi-Lipschitz function f : X → R, where (X, d) is a quasi-metric space, denote by kfkd the constant:
(4) kfkd= sup
(f(x)−f(y))∨0
d(x, y) :d(x, y)>0, x, y∈X
.
Theorem 1. Let (X, d) a quasi-metric space and f : X → R a semi- Lipschitz function. Then kfkd defined by (4) is the smallest semi-Lipschitz constant forf.
Proof. If f : X → R is semi-Lipschitz, then f is ≤d-increasing, and then f(x)−f(y)>0 impliesd(x, y)>0.It follows that
(f(x)−f(y))∨0
d(x, y) = f(x)−f(y) d(x, y) >0.
The inequalities f(x)−f(y)≤0 andd(x, y)>0 imply (f(x)−f(y))∨0
d(x, y) = 0.
Consequentlykfkd≥0.
Forf(x)−f(y)<0 it follows (f(x)−f(y))/d(x, y) ≤ kfkd and obviously forf(x)−f(y)≤0 we have f(x)−f(y)≤0≤ kfkd·d(x, y).
Consequently
f(x)−f(y)≤ kfkd·d(x, y) for all x, y∈X.
Now let K ≥0 such that
f(x)−f(y)≤K·d(x, y), for all x, y∈X.
The function f is≤d-increasing, and then (f(x)−f(y))∨0
d(x, y) =
( f(x)−f(y)
d(x,y) ≤K, iff(x)−f(y)>0, 0≤K, iff(x)−f(y)≤0,
Consequently kfkd≤K.
For a quasi-metric (X, d) let us consider the set:
(5)
S Lip X = (
f :X→R|f is ≤d-increasing, sup
d(x,y)6=0
(f(x)−f(y))∨0 d(x,y) <∞
) .
It is straightforward to see that S Lip X is exactly the set of all semi- Lipschitz functions on (X, d) (see [6]).
2. EXTENSIONS OF SEMI-LIPSCHITZ FUNCTIONS
Let Y ⊂ X where (X, d) is a quasi-metric space. Then (Y, d) is a quasi- metric space with the quasi-metric induced by d (denoted by d too). Let us denote byS Lip Y the set of all semi-Lipschitz functions defined onY and let (6) kfkd= sup
(f(x)−f(y))∨0
d(x, y) :x, y∈Y, d(x, y)6= 0
be the smallest semi-Lipschitz constant for f ∈S Lip Y.
If f ∈ S Lip Y, a function F ∈ S Lip X is called an extension (preserving the smallest semi-Lipschitz constant) off if:
(7) F|Y =f and kFkd=kfkd.
Denote by EY (f) the set of all extensions of the functionf ∈S Lip Y, i.e.
(8) EY (f) =
F ∈S Lip X :F|Y =f and kFkd=kfkd
Theorem 2. Let (X, d) be a quasi-metric space andY a nonvoid subset of X. Then for every f ∈S Lip Y the setEY (f) is nonvoid.
Consider the function
(9) F(x) = inf
y∈Y {f(y) +kfkdd(x, y)}, x∈X.
a)First we show that F is well defined.
Letz∈Yand x∈X.For any y∈Y we have
f(y) +kfkdd(x, y) =f(z) +kfkdd(x, y)−(f(z)−f(y))
≥f(z) +kfkdd(x, y)− kfkdd(z, y)
=f(z)− kfkd(d(z, y)−d(x, y)). The inequality d(z, y)−d(x, y)≤d(z, x) =d−1(x, z) implies (10) f(y) +kfkdd(x, y)≥f(z)− kfkd·d−1(x, z)
showing that for everyx∈Xthe set{f(y) +kfkdd(x, y) :y∈Y}is bounded from above by f(z)− kfkdd−1(x, z),and the infimum (9) is finite.
b) We show now that F(y) =f(y) for all y∈Y.
Lety∈Y.Then
F(y)≤f(y) +kfkdd(y, y) =f(y). For anyv∈Y we have
f(y)−f(v)≤ kfkd·d(y, v) so that
f(v) +kfkd·d(y, v)≥f(y) and
F(y) = inf{f(v) +kfkdd(y, v) :v∈Y} ≥f(y). It follows F(y) =f(y).
c)We prove that kFkd=kfkd.
Since F|Y =f, the definitions ofkFkd andkfkdyield kFkd≥ kfkd. Letx1, x2 ∈X and ε >0.Choosing y∈Y such that
F(x1)≥f(y) +kfkdd(x1, y)−ε we obtain
F(x2)−F(x1)≤f(y) +kfkdd(x2, y)−(f(y) +kfkd·d(x1, y)−ε)
=kfkd[d(x2, y)−d(x1, y)] +ε
≤ kfkd·d(x2, x1) +ε.
Since ε >0 is arbitrary, it follows
F(x2)−F(x1)≤ kfkd·d(x2, x1) for any x1, x2 ∈X andkFkd≤ kfkd.
d) The function F is ≤d-increasing.
Indeed, let beu, v∈Xandd(u, v) = 0.We haved(u, y)≤d(u, v)+d(v, y). Consequently
d(u, y)≤d(v, y). Then
f(y) +kfkdd(u, y)≤f(y) +kfkdd(v, y). It follows that
F(u)≤F(v), and consequently d(u, v) = 0 impliesF(u)≤F(v).
It follows that F ∈EYd(f) so thatEYd (t)6=∅.
Remarks 1. 10 Similarly, the function
(11) G(x) = sup
y∈Y
f(y)− kfkdd−1(x, y)
is ≤d-increasing, andGbelongs to EYd (f) too.
20 The inequality
(12) G(x)≤F(x),
holds for everyx∈X.
Indeed, taking the infimum with respect to z∈Y and then the supremum with respect toy ∈Y in (10) we find
G(x) = sup
y∈Y
f(y)− kfkdd−1(x, y) ≤ inf
z∈Y {f(z) +kfkdd(x, z)}=F(x). In fact, the following theorem holds:
Theorem 3. Let (X, d) be a quasi-metric space, Y a nonvoid subset of X and f ∈S Lip Y.
Then for any H ∈EYd (f) we have
(13) G(x)≤H(x)≤F(x), x∈X.
H(x)−H(y)≤ kfkdd(x, y) implying
H(x)≤H(y) +kfkdd(x, y) =f(y) +kfkd(x, y). Taking the imfimum with respect to y∈Y we get
H(x)≤ inf
y∈Y
f(y) +kfkdd(x, y) =F(x).
The inequality H(x)≥G(x), x∈X can be proved similarly.
Corollary 4. A function f ∈S Lip Y has a unique extension in S Lip X if and only if the following relation
(14) inf
y∈Y {f(y) +kfkdd(x, y)}= sup
y∈Y
{f(y)− kfkd(y, x)}, holds for every x∈X.
Example.
LetR be the real axis andd:R×R→[0,∞) the quasi-metric defined by d(x, y) =
x−y, if x≥y 1, if x < y.
Let Y be given by Y = [0,1] ⊂ R and f :Y → R, f(y) = 2y. Then f is semi-Lipschitz onY andkfkd= 2.The extensionF defined by (9) is
F(x) =
2, if x <0 2x, if x≥0 and the extensionG defined by (11) is
G(x) =
2x, x≤1 0, x >1 Obviously,G(x)≤F(x), x∈R.
REFERENCES
[1] S. CobzasandC. Must˘at¸a,Norm preserving extension of convex Lipschitz functions, J. Approx. Theory,29(1978), 555–569.
[2] J. CzipserandL. Geh´er,Extension of functions satisfying a Lipschitz condition, Acta Math. Sci. Hungar.,6(1955), 213–220.
[3] P. FletcherandW. F. Lindgren,Quasi-Uniform Spaces, Dekker, New York, 1982.
[4] J. A. McShane,Extension of range of functions, Bull. Amer. Math. Soc.,40(1939), 837–842.
[5] C. Must˘at¸a,Best approximation and unique extension of Lipschitz functions, J. Ap- prox. Theory, 19(1977), 222–230.
[6] S. Romagueraand M. Sanchis,Semi-Lipschitz functions and best approximation in quasi-metric spaces, J. Approx. Theory,103(2000), 292–301.
[7] J. H. WellsandL. R. Williams,Embeddings and Extensions in Analysis, Springer- Verlag, Berlin, 1975.
Received: August 8, 2000.