the generalised jump of a function and gibbs phenomenon
TRANSCRIPT
The generalised jump of a function and Gibbs phenomenon.
By K. G. SHmVASTAW, (Un. Saugor, India)
Summary. - By studying certain transforms and applyi~,g his theorem on • the gsneralised jump of a function, author proves certain theorems co~erni~g jump of a fu~t ion and Gibbs phcaome~a.
1. I n t r o d u c t i o n . - L e t f(0) be in tegrab le in the sense of LV, BES(~U~, in ( - - ~ , ~:). W i t h o u t loss of gene ra l i t y we can take f(0) to be an odd funct ion .
The FOURIER Series of f(0) is then
co
(1.1) f(O) c,~ 2 b, sin nO. 1
The Conjuga te ser ies associa ted wi th (1.1) is
oo
{1.2) 2; b,, sin nO = B,d0).
Here
fit) = - - f ( - t) = - fit + 2=).
W e have
(1.3) ,~(t) = f t~ -}- 0 - - f (~ - - t) - - D, D -~ D(~).
I f for some ~ > 0
(i.@ t o (
~ ( 0 = ~ I t - - U)~-lf(u) d u exis ts
o
and tends to l, as n --* oo then 2l is k n o w n as the genera l i sed j u m p o[ f(t) at t "--" 0.
332 K. C. SrIRIW~STAVa: The gene;ralised jump o] a f'tlJtciioJb etc.
DEFINITION. - If lira sup sn(ts} > l. t , - . ~ 0
Then the series (1.1) is said to exhibit Gr]3BS phenomenon at t------0.
2. We shall make use of the following:
(~.1)
Is
s,~(t) = 2] b, sin vt 1
and
s.(t) = ~ b, c o s vt. 1
Also the first ari thmetic means of series (1.1) and (1.2) are given by
(2.2)
n - - 1
:t
and
1 ~-.-:t ~.(t) = - z s~(O.
n 1
We ~lso define the following notations
(2.3) ( a . = a ~ . -1, r z = z r, ,1 r : = ; r,,-~.
We shall make use of the following Lemma.
I ~ t ~ A . - I f
~+~(t) = o(tp-" a s t - - 0
and if
I d t - - O(log~)for h~ l
then for any [~ > ¢~ 2l
the (3ESARO mean of order (1 q-8) of the sequence t nB~ t tends to - - .
This lemma extends our Theorem [5] in the same direction in which ~ r S A K S ~ S U ~ D ~ [4] generalizes a well known t h e o r e m of GHOW [1]. We are omitt ing the proof as it is analogous to that of ] ~ I N A K S H I S U N D A R A M .
K. C. SnRIVASTAVA: The generalised jump o] a ]unction, etc. 333
We prove here the following theorems.
THEOREM 1. - If for some a < 1 (2.4) holds then
1 4 ( - 1)~.-~ (a) - ~ ' f f " ) - - , e " n 2X-- 1
for
and
n t . - - l ( 2 ) , - - 1 ) u = O ( 1 ) a s tw-.*O
=s, , . if,,) 2 l as n - - oc. (b) 10g n ""
TEEOR~.M 2. - If (2.4) holds for some a < 2 then
ta) t s / ( o ) - ~ ; ( t . ) 1 ---2~
and
(b) 21 f l '
{ # . ( 0 ) - ~.(t.) 1 --* ~ ] u ( 1 - - cos uldu
for
(t. -- . o).
THEOREM 3. - I f (~.4} holds for some ~ < 3, then
2X~
~ . t o ) - ~.~t.) - - ~1 t 1-I~ - oo~ . id . - - ~I fJU
f o r
THEOREM 4 . - Under the same assumption s~(~) cover the whole real axis as t~--~ 0 or
ei ther the limit points of
o
or both s tatements hold.
for n t , - = = O ( ~ )
334 K. C. SHRIVASTAVA: The genera~ised jump of a function, etc.
3. Necessary and sufficient conditions that for any convergent sequence Ts----T i. e. the t ransform
n
T , , - - Y~ a, , , ~ , T~ 1
has a l imit hare
(3.1) lira a~,~--a~ exists for = 1, 2, 3... fcl.**4- ~
n
(a.~) t
and
s
(3.3) lim E a,, ~ -- p exists. f$ --.-~ o o 1
These are Ko#u~A SOHUR [2] conditions of regulari ty. We then have
13.4) T. --.. (0 - - Z a~)T + ~ a~T~ . 1 1
4. We study the t ransform
A . - - 1 ~T~cosv t . ' log n 1 v
Applying ABED' s t ransformation twice we get :
15 ~$--2
(4.1) ~ r ~ = • r~a% + ' T~_~(o._~ -- 20.) + r~c., 1 l
Taking T~ = 1, we get
T~ ---- v, T~ =
and therefore (4.1) becomes
I --~" C#s *
Regarding the t ransformation (4.1) we have the following theorem.
K. C, SHRIVASTAVA: The generalised jump of a /unction, etc. 335
THEOREM B. - Ir (C, 2)T,~ - - T and if
then
nt. --~(2X _ = l o o n )
since T. has a (C, 2) l imit say T then
Taking
therefore
2TJ T as n(n + 1)
cos yr. (~v - - - -
V
no. = cos nt,----- O( 1
Fur ther
so also
and
• log n)
t
~%-~ cos v t - - A%--x __ A2 / ' s in v~dx 0
t
-_ 52v-~ _ _ I /'h2e,~xdx 0
t
o
< t
0
t
< / I 1 - - e '~ l~dx 0
< P
'C 33 ) K. ('. SnrtlVASa'av.l: TI~c gcl~erali.~cd jump of a ] l t n c t i o . . etc. . . . . . . . . . . . . . . . . . . . . .
t h u s we h a v e
a l so the c o n d i t i o n
I A=v-~ cos vt I < 2v-~ + t ~
~, cos vt ~' - v .... ---- 0{11
i s s a t i s f i e d . T h u s tlle t r a n s f o r m a t i o n i s r e g u l a r .
T h e r e f o r e
s Vl~n v2h 21e°s l 1
--= Ollog ~}.
A g a i n
n n
(4.4} v v-~ .~ cos vt = ~ v-~ - - ,, v-~t l - - cos vtt i 1 1
7/ n , ,, 1
---- Z v-* - - 2 v v-* s in- ~ yr. 1 a
: ~ v - ' i _ _ n2t, , z 1
----- (log n + Oqll).
n COS Vt 1 £ - - 1 . .'.._~.lim ~ g n ~ v
T h u s b y 13.4) w e see t h a t A, ~ T.
COROLLARY. - I f nt , = 0 t l ) t h e n
T,, cos vt, c,2 T log n 1 Y
i m p l i e s
W e se t
Tv cos vt, c,z T log n. 1 Y
1~, T,, Ao'= O, to=O, A°'--F, (
v
/ = ~ 1 - , n ~ 1.
t V
K. C. SnRIVASTAVA: The geY~er(dised jump o] a funct ion, etc. 337
By Summat ion by parts we get :
A . log n - - A.'ln cos nt . q- Y~ A./l~ 1 cos yr. - - cos (v q- 1)t. }. I
In order that logari thmic summabil i ty imply A . summabil i ty we must have
(4.5) lim l~ i cos y r . - - cos(v -b 1)t. t - - v exists for v -- 1, 2, 3...
(4.6)
(4,7)
l~' I cos vt, ~ cos (v ~ 1)t. I - - 0(log n) 1
l im [1. cos nt~ q- ~, { cos yr. - - cos (v Jr 1)t. }] : p n ----~ CD 1
also (3.1) is satisfied for a ~ - - 1 . Now (3.6) is equivalent to
which is t rue if n t . - 0(l). Substi tut ing T~----1, in (3.7), we get :
cos vt~ ., c,~ p log n 1 ?
which is t rue for p - - 1 by (4.4). Thus we see that logari thmic Summabil i ty imply A , summabili ty. Fu r the r for (4.8} to hold good it is necessary that n t . - - 0 ( 1 ) and corol-
lary 1 i s proved. From Theorem C and Theorem B and the above Corollary we get
s.(t.) --. ~ (,log n)
f o r
looo) which proves Theorem 1 (b).
5. Next we study the transform,
B. =-i ~. Tv c o s yr.. n l
Annali di Matematica 4 3
338 K. C. SI-IRIVASWAVA: The generalized lump of a function,, etc.
t h e n
THEOREM C. - If (C, 2) T. - - T and if
As b e f o r e
n t . - - ~ ( 2 1 ( - 1 ) ~ = 0 ( 1 )
B. 2{--i)~-1
n m S
n B . - - ~, T~A2o~ + T],~1(o._1 - - 2~,,) + T~o,, 1
N o w
S i m i l a r l y
a n d
on = cos nt.= 0(1).
11\ On-1--COS ( n t . - - t n ) = 0 ~ ) ,
h 2 cos vt = R#~t(1 - - e i t ) 2,
] AScosvtn I < ts,
¢ I AS cos Yr. I < # t . ' = O(n). 1
F r o m (4.2) a n d t3.3j we g e t :
- 1 - n -~ 2; co s vt~ --* - - • t~ ~ c o s vtn --~ p.
n t ~ 1
{ (2X--1)u
tn ~ cos vtn --~ COS z d x 1
0
= ( - - 1) x-1.
al~ Vg --~ o o
w h e r e n o w c~ --- cos v t . . A g a i n A s cos vt n --~ 0 as n ~ oo
K. C~ SI/aIViSTAVA: The generalised lump of a function, etc. 339
Hence
~1 ~ cos yr. 2 (-- 1) x-1 n l = ~ 2 ~ - - i = ~
which proves Theorem C. Tak ing T~ = vb~ and applying Theorem i we get
1 ~ vB~ cos yr, -* 2 1 . 2 (1 --)x-~
i . e.~
a n d
41 (-- 1) x-~ s.'(t.) -- ~: 2), -- 1 "
Thus Theorem 1 (a) is proved.
6. The t ransforms
v. __ ~ T~ (1 - - cos vt.) ~ T
D ° = 1- ~ T~(1 - c o s ~ t , ) .
THEOREM D. - If ~ , '~ 3) T, .... ~'I' _~--d ..if
nt.--2),7: = 0 ( 1 ) ,
then
and
(a) c~ - - T ~ (1 -- u cos u) du
o
(b) D.- T.
By repea ted ABEL'S t ransformat ion we get n--8
1
T._~(~._~ -- 3o.) + " + '
340 K. C. SHRIVASTAVA: The geawralised .lump o] a ]unction, etc.
W e t ake
~ = ~(1 - - cos vtn)
C l e a r l y
h~ (1 - - cos vt,) - - 0 (n - - o% v - - 1, 2, ...).
c,---O(1) fo r ( v - - n , n - - l , n - - 2 , . . . ) .
F u r t h e r
h e n c e
and
$
h~ (1 - - v cos vt) _ _ i fd , t ( 1 _ _ d~)adx " o
t
I h~v-~( 1 - - cos vt) I < j l 1 - - e ~" IAd~ < 0 +~
0
n
E v a I A%- ' (1 - - cos vt,) I < n ' t , 4 - - 0(1). 1
F i n a l l y as in o t h e r t r a n s f o r m s
v-~(i - - cos vt.) - - t. :~ (vt.)-~(l - - cos v~.) 1 1
_ f (1 - uc°s u)du 0
T h u s T h e o r e m D (a) is c o m p l e t e .
A g a i n
(6.1) ha(1 - - cos vt) - - - - R% ~a
= - - ~ t ( 1 - - ea) a
i t f o l l o w s t h a t
(6.2) [ h*( 1 - - cos vt) I < t*
K. C. SHRIVASTAVA: The generalised jump of a function, etc. 341
and
n 1 . t. E (I - - cos yr.). ~; v ~ I ~ ( t - - cos v t j I =
1
Therefore
2).r: 1/ lira ~ v ~t A S " ( 1 - c ° s v t . ) { ~ 2 X u ( 1 - - cos u)du - - 1 .
0
Thus Theorem D tb) is complete. Setting T~--vbv and applying theorem B, we get
i , e ,
~ bv(1-- c°s vt") "-" 2~l~ f ( 1 - cos u) ¢¢
0
l i m { S.(0) - - s.(t.)} 21 f l - - cos u du U
o
and also
lim 1 ~ vb~(1 - - cos yr.) -.- 2_/
i, e,
l im -1 [ s . ' ( 0 ) - s.'(t.}] .-. _21
(6.3) and (6.4) together complete the proof of the theorem.
COROLLARY. - I f (C, 3) T . - - T a n d i f
95t.
then
(6.5) 0 . - - D . = l _ ~ v_~( n _ v)T~(1 - - cos yr.)
T]
342 K. C. SHRIVASTAVA: The generalised lump of a function, etc.
7. Now we shall show that the transform (C, 3) can be replaced by (C, 4). Consider the transform
| - :Z T~(v -'n - - 1)(1 - - cos yr . ) . ~ x
THEOREM E. - If (C, 4) T. -- T and if
then the result (6.5) holds. Set
We have therefore
n t . - 2 ~ = 0 ( 1 ) .
c~ -- (v-~n -- 1)(1 -- cos yr.).
By repeated ABI~,L~ S t ransformation we have
Now
hence
r~ B - - 4
1 1
T • - I " Ca-I" T ; . _ 2 ( c . _ , - - 4~._,) -t- ' + '
1 - - cos yr. = 0 [ 1 )
o
Fur the r from (6,1) and (6.2) we have
I h ' ~ I < n [ h ' v - ' ( l - cos vt.)l + I a ' ( l - cos vt.}l
< nt. 8 -1- t2
hence
r}-
v" 1 a% I < n't°'( n2t. + n) 1
= o ( n ) .
K. C. Snmv.~sT.~v.~: The ge~erati.~ed )ump of a function, etc. 343
H e n c e the condi t ions of r egu l a r i t y are Sat isf ied. Now se t t ing T ~ - vb~, and app ly ing (6.5) we get
1 ~ v-~(n - v)vb~(1 - - cos yr.)
_ [ ~ (~ _ ~)~ + 1 ~ ta - - ~)b~ cos ~tn
= -~.(o) - ~ . ( t~) .
lira [ ~ . ( 0 ) - ~.(t.)] - - ~ t cos .----® u(1 - - u ) - 1
o
T h e r e f o r e
wh ich comple te s the proof of T h e o r e m 3.
1 T~(k sin t . - - sin kvt.).
8. The t r ans fo rm
Let k be a pos i t ive integer , we have iden t ica l ly
k
(8.1} k sin ~¢ - - sin kx -~ sin ~ E { l - - cos (k - - 2v + 1)x }. 1
T a k i n g
k~c,~ = k sin vtn - - s in kvt.
w h e r e n t . = k l : + ¢,, and ), an integer , then u s ing (8.l) we have
k n . c, ----- 2(-- 11~ sin ~. E sin 2 (k - - 2v + llE. + ~ (k "4- 1)),~ . IL
A s s u m e that (K-{-1) ~, is an even number , ----21, > O, then
k . no. - - 2 ( - - 1) ~- s in ~. Y. s in 2 ( k - - 2v q- 1)%
hence a s s u m i n g that
344 K. C. SIIRIVASTh.VA: The generaHsed jump o/ a funvlion, ctc.
w e have :
F u r t h e r
and
kn ] v. [ < kSe.~--O (1-s).
( n - - ~ ) t . = ) , ~ : @ e , - - ~ t , , ~ = 1 , 2,...
k(n - - @)c,-~ - - ( - - 1) ~ { k sin (¢. - - ~t.} - - sin k(t, - - bt. }
/¢
- - 2 ( - - I) )" s in (t. - - ~t.) ~ s in ' • 1
!},21 (k - 2v -[- 1). (~. - - bt.)}t
o(1) o(1) for any f ixed k and ~. Also
W h e r e
H e n c e
h%~ = h ~ sin yr. t A" sin kvt. v k v
and
t
1 A' sin kvt_____. _ _ A ~ / c o s kv~d~ k v
o
t
o
k-~ I ~%-~ s in kvt I < kU ~
1 3.
< 2 k ' t . ' n °
= o(1).
K. C. SHaIVASTAVA: The generolised jump of a ]unvtion,, etc. 345 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L - - . -
Finally
1 ~ sin kvt. - - 1 t. Z sin kvt.
Thus
~ W
I fsin k u -~ ~ J u du.
0
. j ' s i _ p t f s i n k u ~ - - du - - ~ j u du .
1 0 0
(7.1)
for
Applying Theorem h we get the following theorem.
THEORE~ 5, - If (C, 4 t T. ~ T and (k -t- 1)X is an even number then
lira E v - l T ~ { s i n v t . - - s i n k v t . } - - T -~ ! sin u -- s in -~ du 0
nt - - X~: = 0 (~)
as n - - - ~ choosing ), = 1, k odd, and T ~ - vb~ we get
l s.{kt.)] 2l f l ( s i n u s i~ku) {7.2) li.m [s.(t .)--~ -'* r c j u -- . du. 0
Since (C, 4) {nb.)--* 2/ aud k is an odd integer write
lira sup s.(t.) ~-. ~ . t ~ 0
If Y - - - { - - ~ then the limit points of s.(t,} cover the whole real axis, if
Y i s finite then f rom(7.2)as n t , - - 7 : = O ( 1 )
7~
21 [" sin u du ] lira sup l s . ( t . ) - - ~ j T I
0
W
1, 0
Anna l i di Ma tcma t i ca 44
346 K. C. SHRIVASTAVA: The geJ~eralised j,u~p of ~r fU~('t~O~, ~tC"
But k is arbiSrarily large; hence
T~
0
Hence in ei ther case the limit points of s,(t,) contain the intervi~l
"ff 7T
2~lsinud~, ÷21ls inud~ ) ( - - ~ ] u -~ j u 0 0
This and the Theorem A completes the proof of Theorem 4.
9. The results of Theorems 1, 2, 3, 4 also hold under the following ~onditions of generalised jump which have a l ready been proved.
THEOREm. S (a) [6]. - If for a > 0, - - 1 < p < 1, a ~- p > 0
J ~(u) l du = 1 log~
0
as t - -+0
then we have
2l - - - - ----- 0 ( ~ o ) - P
where ~[~+e_l_:(~) stands for i -{- ~ -~ pth (~ESARO mean of the sequence { nB.(x) }.
T m ~ o ~ M S (b) [6]. - I f for o :>0 , - - l < p < l , a - ] - p > 0 , p > - - I and
7T
f I . ' • u ~ t / ! 0
t h e n
21 g ~ + ~ - l ( ~ ) - - ~- "- 0 t ~ o - P ( l o g ¢o) J ' + l ! as ~o - - c ¢ .
Tm~oa~M M [3]. - If
y - - ~ + l O < ~ < y 0 * < a < 2
K. C. Sn~xv.~s'rAvA: The ge~era[ i sed jum~p o] a f u u e t i o n , etc. 347
and
t
:
0
then the sequence nB.(x) is summable to 2/ by C~-i~ mean.
ACKNOWLEDGEMENT
I a m i n d e b t e d to P r o f . M. L. MISRA fo r g u i d a n c e a n d e n c o u r a g e m e n t in the p r e p a r a t i o n of th i s p a p e r .
REFERENCES
[1] CHOW, H. C., On a theorem of O. Szas#, • Journal Load. Math. See. ~, 16 (1941), pp. 29,27. [2] C0OK% R. @., Infinite Matrices and Sequence Spaces, Macmillan Ltd. [3] KINUKAMA MASAKUTI, On the Integro Jump of a Function determined by its Fourier
Coefficients, • Prec. Jap. Academy ,, 3t (1955), 4548. [4] .'~II~AKSHI SUNVERAM S., A note on the Theory of Fourier Series, • Prec. ~at. Inst. Sci.
India ,, 10 (1944), pp. 205.215. [5] SHRIVASTAVA, K. C., On the Determination of the Jump of a Function by iJs Fourier
Coefficients, , ~ohoku Mathematical Journal ,, 12 (1960), pp. 120.129. [6] SULAYANA KUMARI, Determination of the Jump of a Function by its Fourier Series,
•Proc. Nat. Inst. Sci. India ,, 24~ (1958). 204.216 [7] SZASZ OTTO, On the General~sed Jump of a Function and ~ibbs phenomenon, • Duke
Math. J'ourn. ~, 11 (1944), pp. 323.3~3. [8] ZrG~UND, A., Trigonometric Series, Warsaw (1950).