• Nie Znaleziono Wyników

Sehauder bases in the space of continuous functions

N/A
N/A
Protected

Academic year: 2021

Share "Sehauder bases in the space of continuous functions"

Copied!
4
0
0

Pełen tekst

(1)

ROCZNIKI POLSKIEGO TOW ARZYSTW A MATEMATYCZNEGO Seria I : PRACE MATEMATYCZNE X I I I (1970) ANNALES SOCIETATIS MATHEMATICAE POLONAE Series I : COMMENTATIONES MATHEMATICAE X III (1970)

J. B

adecki

(Poznań)

Sehauder bases in the space of continuous functions

Denote by {hn{t)} (n = 1 , 2 , . . . ) the orthonormal Haar system.

t

Ciesielski [1] proved that functions 1 and f hn(s)ds (0 < t < 1 and 0

n = 1 , 2 , ...) form a Schauder basis in (7<0,1> and that

oo 1

t

F(t) = F { 0) + I f hn(s)dF(s) J hn(u)du

71 = 1 о о

for any function F e C < 0 ,1>.

The following theorem generalizes this result:

T

heorem

. I f g e C ( 0 , 1>, g(t) Ф 0 for 7e<0,1), and g is of bounded variation on < 0 ,1 ), then the functions

t

(p 0(t) = 1, cpn(t) = j g ( u ) h n(u)du ( n ■= 1 , 2 , ...)

о

form a basis in the space of continuous functions on <0, 1>, and for every -Pe(7<0,l) there holds

(1)

CO 1

F(t) = # (0 ) + ^ f n(t) f

71=1 0

hn(s)

g{s) dF(s),

where the series is uniformly convergent over the interval <0,1).

P r o o f . It is well known that every positive integer n > 1 can be written as n = 2m+ 7c, where m ^ 0, 0 < h < 2m, and m,Tc are uniquely determined by n. Let us set t10 — 0, h,i = 1 and for n > 1

for i = 0 , 1 , ..., 2 h,

for i = 2A:+1? • ••, n.

Roczniki PTM — Prace Matematyczne X III 13

(2)

194 J. Radecki

Evidently, {/п^} (г = О , 1, . .., п) is a sequence of normal partitions of the interval <0, 1>. Let us write P n>i = {t: te (tn>i_x, tn>i)} for i — 1, 2, n —1 and P n>n = {t : t€<dn,n-u tn>n)j. If f e L ( 0 , 1>, then f [geL<0, 1>

and for tePni the following equality

n

1

У f hj(s)ds = f

J 9(S) i

Pn,i\

m

\Pn,i\ 9 ( s ) n,i

ds

j=i о

holds [2], where \Pn,i\ = 1 - Therefore, for F(t) = F(0) + f f(s)ds о we have

n 1

V i Г h ( s) 1 Г dF(s)

< 2>

j =

1 0 WnApJ, g{s) J

n,%

where tePni . Since absolutely continuous functions are dense in G<0,1), so the equality (2) holds for any F e C <0, 1> and n = 1 , 2 , . . . Multiplying both sides of this equality by g and integrating we obtain

(3)

lb i ь

an( F: t ) = У f — —- dF(s) f g(u)hj(u)du

J g(s) J '

/-1 о

= Y — f g ( » ) d u + - i - Г , ( « ) * , ,

\Pn,j\ pJ g(s) pd \Pn,i\ pJ g(s) t y

where t€Pn i .

Let us write Sn(t) = —

[ J - n j l P n J

gn(t) = Sn(t)lg[t). Then (3) can be statet in the form

j g(u)du for tePnj , j = 1 , 2 , and

(4) F(0) + an( F - , t ) - F ( t ) = E(tM_ 1) [ ^ ( W - i ) - 1] +

t t

+ F ( t K_i_l ) - F ( t ) - F ( m g n { 0 ) - 1 1 - § F { s ) d g n(s)

+ J

F ( s ) d g n (s) +

0 ^n,i— 1

f ł W 4 ^ _ z ! V f c . ) _ r J W 4 4 - ) i . J y Lg{tn,i) g(tn,i-

1

) pJ. \g(s)lJ l-Pndl, -

{гс,г— 1

Let V { Щ denote the total variation of В over <a, &> and let

a

31(H) = sup \H(t)\. Note that for 0 < a < b < 1 there holds

<£<0,1>

(5) V Ы <

M ( 8 n)

V (--)

+ m ( - \

V («„) < Jf(9) V

{ - )

( 1 ) V (ff).

a a \gl \gI « a \ 0 / \ 0 / a

(3)

Schauder bases 195

Since the sequence {8n(t)} (n = 1 , 2 , . . . ) converges uniformly to g(t) on < 0 ,1>, so lim gn(t) = G(t) = 1 uniformly for /e<0,1>. By little

П—9-00

modifications in the proof of Helly Theorem we get: If aeC(a, by, functions

0n($) (n 1 , 2 , . . . ) are of uniformly bounded variation on <a, by and

t

if /?„($) converges uniformly to 0(<) on <a, 6> as w -► oo, then f a(s)dpn(s)

t a

converges uniformly to f a(s)d(3{s) on {a, by. So we obtain

о

t t

lim JF($)dgn(s) = JF(s) dG( s) = 0

n~+oo о

о

t t

uniformly over < 0 ,1>. Functions F, F/g, \J(g), V (1/^7) are uniformly

о о

continuous on < 0 , 1>, so for given e > 0 there exists a 6 = 3(e) > 0 such that for 0 < x < у < 1 and y — x < <5 there holds \F(x)—F(y)\ < e,

F( x) F(y)

g(x) g(y)

У

V(g)<e, < e.

Moreover, for sufficiently large n we have sup \Pn>i\ < 6, \gn(t)~

t

1

—1| < e, I / F(s)dgn(s)\ < s for all <e<0, 1>. So, from (4) and (5) we 0

obtain

\F(0) + <rn(F-,t)-F(t)\

< M { F ) E + e + M{ F) e + e + M{ F) [ M{ g) e + M ^ e ] + M{g)[e + M{ F) e]

= e[2M{ F) M{ g) + M { F ) M ^ + 2M( F) + M(g) + 2].

Let now F(t) = £ On<Pn(t), where the series is uniformly convergent n=o

on < 0 , 1>. Obviously, a 0 = F (0). By the continuity of the functionals г Ля(в)

I ——- dF( s) , we obtain о 9(s)

f %т(

g ( 8 )

8) dF(s) = ^ ~ d ( p n(s)t = ^ anj Jim(s)hn{s)ds

n=o о ' n=i о

— Q>m

for m = 1 , 2 , that proves the uniqueness of (1).

(4)

196 J. Rad ecki

R e m a r k . If the continuous derivative g'(t) exists for g(t) Ф 0 in

<0,1) and if the improper integral

/ dF{s) g{*)

= lim / dF(s) g{$)

exists for continuous F(t) in <0, 1), then the equality (1) holds in <0, 1) and the convergence is almost uniform.

References

[1] Z. C ie s ie ls k i, On Haar functions and on the Schauder basis of the space G <0, 1>, Bull. Acad. Polon. Sci. 7 (1959), -pp. 227-232.

[2] R. S ik o r s k i, Funkcje rzeczywiste, tom II, str. 135, Warszawa 1959.

UNIW ERSYTET im. ADAMA MICKIEWICZA W POZNANIU K A TE D R A M ATEMATYKI II

UNIW ERSITY OF ADAM MICKIEWICZ IN POZNAŃ

Cytaty

Powiązane dokumenty

Taberski for his kind criticism and valuable

W ell, Bounded continuous vector-valued functions on a locally compact space, Michigan Math. W en j en, Generalization of Stone-Weierstrass approximation theorem,

This answers questions raised in [5] where it is shown that in various models of set theory there are universally bad Darboux functions, Darboux functions whose sum with any

Although our generic construction produces skew products with a good cyclic approximation, hence of simple spectrum, there also exist ergodic real-analytic cocycles (of topological

Let K be the compact space generated by this family of functions (see the beginning of this section for the definition of K).. Then the space K has the

At this point, the problem of supports of homomorphisms of G(X, 01) have been completely solved. It turns out that theorems of the above type hold true for

For a real Banach space with a smooth dual we then introduce a simple family of sublinear functions which contains an exhaustive family of upper convex approximations for

The Hausdorff measure is often more convenient that Kuratowski measure since in many spaces there are formulae allowing to calculate or evaluate its values ([1], [2]) while