GEORGIAN MATHEMATICAL JOURNAL : Vol . 6 , No .

1 , 1 9 9 9 , 9 1 - 98

WEAKLY PERIODIC SEQUENCES OF BOUNDED

TRANSFORMATIONS :
A
SPECTRAL
CHARACTERIZATION

LINEAR

A . R . SOLTANI AND Z . SHISHEBOR

Abstract . Let X and Y be two Hilbert spaces , and L(X, Y ) the
space of bounded linear transformations from X into Y. Let {An }
L(X, Y ) be a weakly periodic sequence of period T. Spectral theory
of weakly periodic sequences in a Hilbert space is studied by H . L .
Hurd and V . Mandrekar ( 1 991 ) .
In this work we proceed further
to characterize {An } by a positive measure and a number T of

L(X, X) valued functions a0 , ..., aT 1 ; in the spectral form An =

Z

ein (d)Vn (), whereVn () =

is an

T
1
X

ei

2 k n
T

ak ()and

k=0

L(X, Y ) valued Borel set function on [0, 2) such that

(()x, (0 )x0 )Y = (x, x0 )X( notdef parenrightnotdef notdef

I
ntroducto n
e tX a nd Y b eH l ib ert s aces , a ndl t L(X, Y ) s a ndf r b
ounded near t ansform a t io nsf o m X i tn o Y. upp s o e {A, n
Z braceright comma Z i t he t o fi tn egers , i a s q u e n e c i
n L(X, Y parenright periodST hes q e u n c e{An} i s id t ob e
ea kly p riodic , if fo r e e v ry n, m Z, (Anx , Amx )Y = (An+TAx, m+T x) Y,
, x X, f r s ome T > 0 hes m a l lest T i dn icates t e p r e iod ,
a ndi n is c se w eu s e t hen t oation. T hyphen W P . W ec llt hes q
e un c e {An} L(X, Y ) ron g l y p r e io dc i f An= An+T , n Z.S
u h c a s eq u ec n e i s d en t o e db yT hyphen S P . v ery 1 W
Pi is at ion r a y . S p e c tral t eory o f T W hyphen Ps q u e n e c
s i nH i l bert
aces , a ndi snc rtain B a n a ch s p c a es , w a sfi st s u i d ed b yH
. L . H ud a nd . M andekar [ ] . I X i fin ite d im n e sion a l a nd
Y = Lparenleft two, F, P parenright commarw h r ee
, F, P ) i a p o ba b i lti y s p c a e , t e ne v e ry T W hyphen Ps
eq u ec n e i s a p r e iodically
rrelate ds co ndo der p ocess . u c h a p ocess w s afi st i tn ro du c e d
a nd u d i e db yG a l d y she rv[ ].rF r o m oe o nt ese k nd o fp ocesses
se[],[4].
1 99 1 Mathematics Subject Classification . Primary 47 B 1 5 , 47 B 99 ; Secondary 60
G12.
Key words and phrases .
Hilbert space , bounded linear operators , weakly periodic
sequences , spectral representation .
1 72 - 947 X
ish ing Corporation

91
/99/0100 0091Dollar16.0/0circlecopyrt c1999

P l enum P ubl

92

A . R . SOLTANI AND Z . SHISHEBOR

The significance of a spectral representation for a given sequence is re - alized

in probability theory . It amounts to considering the elements of the sequence
as the Fourier coefficients of a certain ( random ) measure . Let
be a set function on Borel sets in [0, 2) with values in L(X, Y ). Among the
following properties ;
i

x
(periodcenteredparenright x(( ii ), for) forevery
parenrightperiodcenteredx every0)

X, finesax , x0

()x

is
, pl ( ex an(m
Xcm

( i i s sa t i s fied , t h en w e a cl l a a r ndo ms ectrum , R

S semicolon i f ( i ) a n d i parenleft) b th esa t i s fied , t h en w ca
l l a a ndo m s p ectru mw th o thogonalin crem n e t s comma
SO .
i s a R comma S t h en
n=
w l l d e fined n i

two pi in(d),

Z .1)

s e nse h t at

te
Z

nx, Bmx )

two pi pi two i(n m)((d)x, d0parenrightx 0) =

Z
two pi pi two i(n m) xx(d, d0parenright comma

brackleft zero 2)..E Ayn

ery sh quencermonizable {B}s
periodcenteredparenlef t

s a
m ea sure
h ereX, Y xx)gven
)iby( b.1) is c ll eonbrackleft zerodhrm2noizable[4 )
,

encein LX, Y, f o rw h icha it s a o sa t s i fies ( i ) , i s va t ationaryse

quence . B yu (ing a c e rtain T h o r otof a l u itary o erator H u rd
a d M n a drekar
p ovided t h e G adyshev s t y pe e r presentation f o ra T W
Ps quencein a l b ertsp ace . I c n b ed duced r f om t e i Rr w
rk t at {A} LX, Y ) s T W Ps quenceif an d o ly i fA n= 0 e
inoZ (d), n Z w e h re Zi s a
So n brackleft zero 2), t h atit sre st i r ctio n t o e ch two brackleftpT
1)), p = 0, ..., T 1, is a RSOI ; more precisely ,
notdef

p
Z() = p=0
TP 1 Wp (2notdef T

, 2(T p+

, notdef two parenleft notdef notdefp notdef notdef plusT no

(1.2)

for every , 0 [0, 2T ) with notdef notdef notdef notdef a d e notdefe notdef notdefnotdef
r y notdef notdef notdef notdef pbracelef tqnotdef notdef equal notdef0, notdef
notdef notdef period. period notdef notdef commaT notdef minus
inf inity.notdef notdef
h i sre presentation w s a al so i ndependently, d rived b yt e a thorsin
six brackleft] b notdef notdef y ing a d fferent a proach .

t is w o r k w ep oceed f rther t d i ent i f y e ch T W Ps quence {A}

LX, Y ) w th a p sit iv e m a e sure a a n m b e r T o LX, Xhyphen parenright
v alued nctions , a, ..., aT minus one. T e h i d entification s i
g ven i n T e h orem 3 4 , w h i ch

A SPECTRAL CHARACTERIZATION

93

is a refinement of Gladyshev s result on periodically correlated sequences .

The relationship between the functions a0 , ..., aT 1 and the matrix f, given
in ( 2 . 3 ) , introduced by Gladyshev is also provided in Theorem 3 . 4 . An
application of our representation to the regularity of periodically correlated
sequences is as follows .
T
A periodically correlated sequence Xn , n Z, is regular , i . e ., span
{Xt ,
n
R 2/T

t n} = {0}, if and only if 0

log det f ()d > , with f as in ( 2 . 3 ) (
QT 1
see [ 2 ] ) . It follows from ( 3 . 3 ) that det f () = j=0 a0 ( + 2jT )2,
R 2
[0, 2/T ). Therefore the sequence is regular if and only if 0 log a0 ()d >
. Theorem 3 . 1 in [ 3 ] easily follows from this observation , as f 0 in [ 3 ]

dominatesa0 .
We summarize our characterization by the spectral representation An =
in
e
(d)Vn (), where , is a certain RSOI , as in the abstract , and
0
for each [0, 2), {Vn ()} L(X, X) is a T SP in n, as is defined in
Theorem 3 . 4 .
2 . Notation and Preliminaries
Suppose Z is given by ( 1 . 4 ) ; let F be the L(X, X) valued Borel set
func tion on [0, 2) [0, 2) for which

R 2

(x, F (, 0 )x0 )X = (Z()x, Z(0 )x0 )Y,

x, x0 X.

(2.1)

We call F the spectral measure of the sequence {An }. Note that the support of
F lies on the lines dk : s t = 2kT, k = T + 1, ..., T 1. Furthermore

(An x, Am x0 )Y =

Z
0

eint+ims (x, F (dt, ds)x0 )X, x, x0 X.LetFk be

the restriction of F to the line dk , k = T + 1, ..., T 1. Let be a positive

finite measure for which each Fk is continuous and
2T k(dx) = (dx) on [0, 2/T ), k = 1 T, ..., T 1,
(2.2)
R
and (x, Fk ()x0 ) = (x, f k(s)x0 )(ds), where f k are L(X, X) valued func tionson[0, 2), k = 1 T, ..., T 1, [7].
To see the existence of such a measure , note that corresponding to each Fk
there is a positive finite measure k such that Fk is k continuous [ 8 ] .

PT 1
Now let = k=1T k, and 2T k (E) = ((E+2kT )notdef comma2notdef notdef
parenright) notdef comma t notdef e F n d notdeffi notdef notdef notdef n
e notdef notdef notdef notdef
E)Now =

T j equal1

minusone
equal k T1T

kT

T notdef
(Etwo intersectionnotdefnotdef
pi(j1) com

defines an operator - matrix ; we refer to a matrix whose entries are operators

as an operator - matrix . We call f the spectral density of {An } with respect

94

A . R . SOLTANI AND Z . SHISHEBOR

to .
Recall that an operator - matrix T = [Ti,j ]i, j = 0, ..., T 1 is called
positive definite if and only if the matrix [(xkj , Tj,l xkl )]j, l = 0, ..., T 1, is posi
tive definite for each xk0 , ..., xkT 1 .
We recall that L2 (, X) stands for
R 2
Bochner integrable functions g : (0, 2 ] X such that 0 k g() k2X (d) <
the proof of which is omitted [ 9 ] .
3 . Spectral Characterization
R 2
Hn integral ( sto chastic integral ) of the form 0 (d)g(), where g
L2 (, X) and is the RSOI given below in ( 3 . 1 ) , can be defined by using
the well known result on a existence of the weak sto chastic integral .
The
detail is given in the following theorem .
For more on a stochastic integral
we refer the readers to [ 1 0 ] and [ 1 ] .
Theorem 3 . 1 .
Let be a positive finite Borel measure on [0, 2)
and be a RSOI for which
(()x, (0 )x0 )Y = (x, x0 )X( notdef parenrightnotdef notdef
.onenotdef ){notdef notdef
R
en f o r e ery g L(, Xcomma parenright
d)g() i
w ll d
fined a
a ne e m e n t hY a d
v
Z
0two pi d)g()vextenddoublevextenddouble

zero integraldisplaytwo pi

Moreover , the integral satisfies the fo l lowing properties :

2

Z
(i)

Z
(d)(g1() + g2()) =

Z
(d)g1() +

(d)g2(),

g1, g2

L2 (, X);
( ii ) if gn g in L2 (, X), th en

R 2
0

(d)gn()

R 2
0

(d)g() in

Y
asn .
Corollary 3 . 2 .

Let and be as in Theorem 3 . 1 .

Z

k a()x k2X (d) < , x X}.

A = {a; a : [0, 2) L(X, X),

Then for e very a A, x X,
in Y, and
Z
k
0

R 2
0

Let

(d)a()x is well defined as an e lement

Z
(d)a()xkY = {
0

k a()x k2X (d)}21 .

() k 2Xd)

Moreover , for every x, x0 X and every a, b A,

Z

(i)
Z
(ii)

(d)a()(x + x0 ) =

0
2

given

0
2

(d)a()x0 ,

X,

Z
(d)an ()x

(d)b()x ,

(d)a()x +
0

for
th en

(d)a()x +

( ii i )
L2 (, X),

Z
(d)(a() + b())x =

if

an ()x

(d)a()xinY.
0

a()x

in

A SPECTRAL CHARACTERIZATION
95 Proof .
For each fix x
X, a()x is an X valued function . If a A, then
a()x L2 (, X).
Now apply Theorem 3 . 1 and use the linear property of

a().
The following lemma concerns the Cholesky decomposition for a positive
definite operator - matrix ( Lemma 3 . 5 in [ 1 1 ] ) and it plays a crucial role
in
our characterization .
Lemma 3 . 3 .
Let Mn be an n n positive operator - matrix , th
en Mn =
Un Un , where Un is an n n upper triangular matrix .
The following theorem is the essential result of this paper .
Theorem 3 . 4 .
Let {An } L(X, Y ). Then {An } is a T WP
if and only if
2

eins (ds)Vn (s),

An =

(3.2)

where
( i ) is a RSOI that satisfies ( 3 . 1 )

and is a finite measure

satisfying
(2.2);
(ii)Vn (s) =

T
1
X

ei

2 k n
T

ak (s + 2kT ),

s [0, 2), n Z,

k=0

where each ak
for

is an L(X, X) valued function on [0, 2) with ak (s) = 0

s [0, 2kT ).
Furthermore , th e triangular operator - matrix
A(x) = [a j k (x + 2jT )]k j,
k, j = 0, ..., T 1, satisf ies

f (x) = A (x)A(x),

x [0, 2T ),

(3.3)

i . e . , it is the Cholesky decomposition of the density f (x) given by ( 2 .

3),
a (x) s tands for the adjo int of a(x). Proof .
Let {An } be a sequence for
which ( 3 . 2 ) holds , i . e . ,

An =

k=0
X
T 1

Z
0

eins 2T k(ds)ak (s).

(3.4)

Thus
(An x, Am x0 )Y =
k=0
l=0
XX
T 1 T 1

Z
0

Z
0

eins+imt ( 2T k(ds)ak (s)x, 2lT (dt)al (t)x0 )Y.(3.5)

96

It also follows from ( 3 . 1 ) that

A . R . SOLTANI AND Z . SHISHEBOR

( 2T k(ds)ak (s)x, 2lT (dt)al (t)x0 )Y =

={

a (s)x,

(0 k

, ap+k (s

+ p2T )x0 )X 2T p(ds),

s 2kT = otherwiset 2lT

where p = l k. Since 2pT (ds) = (ds), p = 1 T, ..., T 1. We obtain

(An x, Am x0 )Y =
(T

= p = P 1 1)

2p
T )

eins+im(splus

0
k=p
TP 1 (ak (s)x, ak+p (s + 2p/T )x )X (ds)+
2

eins+ims

+
0
PR
1
2 ins+im(s+2p
T )
+Tp =1
0 e

k=0
X

(ak (s)x, ak (s)x0 )X(ds)+

T 1

(s
(ak=0
TP 1p (s)x, ak+p
k

+ 2p/T )x0 )X (ds).(3.6)

It is clear from ( 3 . 6 ) that {An } is a T WP sequence .

Conversely , let
{An } be a T WP sequence ; then it follows from ( 2 . 1 ) that

(An x, Am x0 )Y =

Z
0

eins+imt (Z(ds)x, Z(dt)x0 )Y =

0
2 Z 2

=
0

p=0 X
l=0 Z
X
T 1 T 1

l R
T2

Tp

(p+

(l+1)

eins+imt (x, F (ds, dt)x0 )X =

2T

2(p l)m

1) i(nm)siT

(x, Fpl (ds)y)X =

2T

T 1 T 1
p=0
X

2(p+T 1)

2pT

p=0 X
l=0
X

eins+imt (x, F (ds, dt)x0 )X =

1R 2(T p+1
)ei (n
2T p

m)s i2 km
T (x, f k(ds)y)X(ds)+

Pk=T + T +k
1

p=0 Z
X

T 1

=k
X

Tp

Pk=1

T 1 p T l

R
2T

2(p+T 1)

ei(nm)s (x, f 0(ds)y)X(ds)+

2T p

(p+

km
1) i(mn)si2 T

(x, f k(ds)y)X(ds) =

A SPECTRAL CHARACTERIZATION

k=T
X+

2 mk

1eiT

2(TT +k)

97

ei(nm)s (x, f k(ds)y)X(ds)+

ei(nm)s (x, f 0(ds)y)X(ds)+

+
0

k=1
X

2 mk

eiT

2kRT 2 ei(nm)s (x, f k(ds)y)X(ds).

T 1

By using ( 2 . 3 ) and comparing the last equality with ( 3 . 6 ) we obtain that

if
ak ,
k=p
X

k = 1 T, ..., T 1, satisfy

(ak (s)x, ap+k (s + 2T p)x0 )X = (x, fp (s)x0 )X, p = T + 1, ..., 1,

T 1

T 1
X

(ak (s)x, ak (s)x0 )X = (x, f 0(s)x0)X ,

(3.7)
k=0

k=0
TP 1p (ak (s)x, ak+p (s

+ 2T p)x )X = (x, fp (s)x )X, p = 1, ..., T 1,

then {An } will satisfy ( 3 . 2 ) , as ( 3 . 2 ) , ( 3 . 4 ) , ( 3 . 5 ) and ( 3 . 6 )

are equivalent . But it easily follows that ( 3 . 7 ) and ( 3 . 3 ) are identical .
On the other hand , ( 3 . 3 ) follows by applying Lemma 3 . 3 to f.
Remark 3 . 5 .
We were led by the referee to the representation that
{An } L(X, Y ) is T WP if and only if
Z
An =

ein (d)Dn ,

(3.8)

where is an L(X, Y ) valued measure and {Dn } L(X, X) is a T SP

sequence . In (3.8) lacks property ( 3 . 1 ) , but instead each {Dn } is inde
- pendent of .
Let us provide the referee s proof which is a nice application of the sym bolic calculus to unitary operators .
Proof . Since {An } is T weakly periodic , An+T = U An for some unitary U
in Y and each n and T. We define Bn = U n/T An . It is easy to make sure
B }

n/T B .
we n Writing

n
thethat {spectral
L(X, Y )isT stronglyrepresentationfortheunitaryperiodic andAn operator = U U1/T

R 2

ein E(d)Bn , where E(d) is a solution of the identity of Y. Now if X

is infinite - dimensional , one can establish an isomorphism ( without loss
of generality one can assume that X and Y are separable )j L(Y, X) and we
come to ( 3 . 8 ) with (d) = E(d)j 1 and Dn = jBn .
0

98

A . R . SOLTANI AND Z . SHISHEBOR

Acknowledgement
This research was supported by the AEOI , Center for Theoretical Physics
and Mathematics .
The authors would like to thank the referee for reading carefuuly the paper
and providing valuable comments such as the representation cited in Remark 3
. 5.
References
1 .
H . L . Hurd and V . Mandrekar , Spectral theory of periodically
and quasi - periodically stationary S S sequences .
Tech .
Report No
.
349 , Cen - t er for Stochastic Processes , Department of Statistics ,
University of North
Carolina ,
Chapel Hill , NC 27599 - 3260 , USA , 1 991 .
2 . E . G . Gladyshev , Periodically correlated random sequences .
Soviet
Math .
Dokl .
2 ( 1 96 1 ) , 385 388 .
3 . A . G . Miamee , Periodically correlated processes and their
stationary
dilations . SIAM J . Appl .
Math .
50 ( 1 990 ) , 1 1 94 1 1 99 .
4 . M . Pourahmadi and H . Salehi , On subordination and linear transfor
- mation of harmonizable and periodically correlated processes . Probability
Theory on Vector Spaces , III ( Lublin ) , Lecture Notes in Math . , 1 80 , 1
95 2 1 3 , Springer , Berlin New York , 1 983 .
5 . Yu . A . Rozanov , Some approximation problems in the
theory of
stationary processes . J . Multivariate Anal .
2 ( 1 972 ) , 1 35 1 44
6 .
Z . Shishebor and A . R . Soltani , A spectral representation weakly periodic sequences of bounded linear transformations . Technical
report , Department of Statistics , Shiraz University , Shiraz , Iran , 1 995 .
7 . V . Mandrekar and H . Salehi , On singularity and Lebesgue type de composition for operator - valued measures . J . Multivariate Anal . 2 ( 1
971 ) ,
1 67 1 85 .
8 . N . Dunford and J . T . Schwartz , Linear operators , I . Interscience ,
New York , 1 958 .
9.
J . Diestel and J . J . Uhl , Vector measures .
Amer .
Math .
Soc . , Provi - dence , R . I . , 1 977 .
10.
Yu .
A . Rozanov , Stationary random processes .
Holden Day ,
San Francisco , 1 967 .
1 1 . G . T . Adams , P . J . McGuire , and V . I . Paulsen , Analytic
reproducing
kernels and multiplication operators . Il linois J . Math .
36 ( 1 992 ) ,
404 4 1 9 .
( Received 1 8 . 1 2 . 1 996 ; revised 5 . 9 . 1 997 )
Authors addresses :
Center for Theoretical Physics Department of Statistics
and Mathematics AEOI Shiraz University
P . O . Box 1 1 365 - 8486 , Tehran Shiraz 7 1 454
Iran Iran