PROCEEDINGS OF THE IEEE, VOL. 55, NO.

10, OCTOBER1967 zyxwvutsrz 1701

zyxw
zyxwvut
zyxwvutsr
zyxwvutsr
zyxwvutsrqp
zyxwvutsrqpo
Sampling, Data Transmission, and
the Nyquist Rate
H. J. LANDAU

A bstract-The srmpliag theorem forbandbited signals of W e energy

can be interpreted in two ways, associated with
sbnsaoo.
the names of Nyquist and
error in reading the sample values produces only a
correspondingly small error in the recovered signal.
2) Every square-summable sequence of numbers may be
1) Every signal of W e energy and bandwidth W Hz may be com- transmitted at the rate of 2 W per second over an ideal
pletely recovered, in a simple way, from a knowledge of its samples channel of bandwidth W Hz, by being represented as
taken at the rate of 2W per secwd (Nyquist rate). Moreover, the
reeoverybseMe,inthe~thatasmfllerrorinrefdiagsample
the samples (at the points k/2W) of a bandlimited

Signnl.

zyxwvut
values produces only a correspondingly small error in the recovered

2) Every sqalre-srrmmable sequesce of nmnbers may be tnutsmitted at

the rate of 2Wper second over an ideal channel of bandwidth W H z ,
by being represented as the samples of an easily conshocted band-
limited signrl of W e energy.
signal of finite energy.
Thus the sampling theorem serves as basis for the inter-
changeability of analog signals and digital sequences, so
valuable in modem communication systems. The rate of
2W samples per second for W Hz of bandwidth is often

elaborate compotatioos. zyxwvuts

The practical importaoce of these results, together with the restrictiom

zyxwvutsr
implicit in the sampling theorem, make itnatnral to ask whether the above
rates cannotbe improved, by ppssing to differently chosensampling instam@
or to badpass or mdliband (rather than badimited) signak, or to more
In tbis paper wedraw a distiactioo between re-
combwthgasignalfromitssamples,anddoiisoinastibleway,andwe
argue that only stable sampliq is meaaingfal in practice. We then prove
that :
called the Nyquist rate.
In view of the. practical importance of these results, it is
natural to ask whether the Nyquist rate prescribed above
for stable sampling or for data transmission cannot some-
how be improved upon. Since 1)and 2) refer to a very special
situation-to signals which in frequency occupy only a
single band and to their values at regularly spaced instants-
1) stablesamplingcannotbeperformedataratelowerthanthe it is conceivable that signals might be recoverable from
Nyquist, their values taken at a lower rate, if the sampling instants
2) data cannot be transmitted as samples at a rate higher than the were chosen differently; or if the signals had their fre-
Nyquist,
quencies in a union of several bands; or at thecost of more
regardless of the locatioa of sampling instants, the nature of the set of fre-
computing than is required by the simple formula (1).Were
qwacies whicb the signals occupy,or the method of comhcth. These
this possible, we would have a more efficient reduction of
cwclmions apply not merely to fiuiteswrgy, but Pt90 to bomded, signals.
continuous signals to digital form, with a consequent saving
INTRODUCTION of bandwidth. Similarly, it is conceivable thatarbitrary
HE sampling theorem states that f(t) is a signal of square-summable numbers could be prescribed at a higher
finite energy bandlimited to W Hz if and only if rate as the sample values of a signal, if the sampling instants
were properly chosen, or if the carrying signals were allowed
sin 271W(t- k/2W) frequencies in an appropriate union of bands. Were this
f ( t )= & " 271W(t - k / 2W )
with 1 < cx) ; (1) possible, we would have a more efficientwayof sending
data over given bandwidth.
thereupon The purposeof this paper is to prove that stable sampling
rm of signals cannot be performed at less than the Nyquist
rate, nor can data be transmitted as samples at a greater
rate, regardless of the location of sampling instants, the
nature of the set of frequencies which the signals occupy,
or the method of computation. Although discussed in the
This may be read in two ways, each of which has found context of finite-energy signals, these results also apply to
important applications. bounded signals.
1) Every signal of finite energy and bandwidth W Hz
FORMULATION OF THE PROBLEM AND
may be completely recovered from a knowledge of its
STATEMENT OF RESULTS
samples taken at the rate of 2W per second. More-
over-indispensable for any implementation in prac- We consider a fixed set S of N-frequency bands whose
tice-the recovery is stable, in the sense that a small total finite length (counting negative as well as positive
frequencies) we denote by m(S), and the space g ( S ) of all
Manuscript received March 3,1967; revised June 12, 1967. signals of finite energy with frequencies contained only in
The author is with Bell Telephone Laboratories,Inc., Murray Hill, N. J. S . Specifically, we make the following definition.
 zy
zyxwvutsr
1702

zyxwvutsrqp
zyxwvu
zyxwvuts
zyxwvuts
Dejinition: B(S) consists of all square-integrable func-
tions f ( t ) whose Fourier transform F(w)is supported on s,

zyxw
i.e., F(w)= 0 for o 4 S.
PROCEEDINGS OF THE IEEE, OCTOBER 1967

We show in the Appendix that points of a set of inter-

polation for B(S)are necessarily separated by at least some

zyxwv
positive distance d, and that the interpolation can be per-
We have in the past called such signals multiband. formed in a stable way, i.e., the function f ( t ) E B ( S )can be
We start by considering those sets of points t= {tk} at chosen so that
which the values of a signal f ( t ) E B ( S )can be used to re-
construct the signal everywhere. We assume from the outset
that these points separated
are by at least some positive
jrnlf(t)12
-m
dt 1 lak12

distance d, but otherwise do not restrict their location in

any way; this separation hypothesis, while required in the with the constant K independent of {ak}.
proofs, does not seem a significant restriction in cases of There is an extensive literature of particular sampling and
practical interest. The first requirement to impose on the interpolation theorems, especially whenS is a single interval,
(tk} is that the values { f ( t A } determine f ( t ) , i.e., if two but we do not intend to discuss it here. Instead, we will be
functions f ( t ) and g(t)EB(S) agree at the points t = { t k } , concerned with the rate at which stable sampling or inter-
polation can be performed in B(S).This corresponds exactly

zyxwvu
they coincide identically. Writing this in terms of the func-
tion h(t)=f(t) - g(t), we require : to the density, on the t-axis, of sets of sampling or inter-
polation. The facts are as follows.
h E B(S), and h(tk)= 0 for all k, implies h(t) = 0. (2) 1) There exist sets of uniqueness for B(S) having arbi-
trarily low densityf2!This is occasionally reported as a
We will call a set { t k } satisfying (2) a set of uniqueness for sampling result with practical implications, but such
W). an interpretation is somewhat misleading, in view of
If ( t t } is a set of uniqueness for B(S),the samples of the lack of stability. When S is a single interval, so that
f € B ( S )taken at the instants t = { t k } do indeed determine the WS) consists of all signals of prescribed bandwidth, a
signal, but this initself does notconstitute a sampling remarkable result concerning sets of uniqueness is
theorem adequate in practice. The reason is that we know available : given a set of points { t k } , Beurling and
nothing about the effect on the reconstruction process of Malliavin[’l have shown how to determine explicitly

zy
an error in measuring the sample values. It may very well the least upper bound of those bandwidths for which
be thatanarbitrarily small misreading of thenumbers {tk} is a set of uniqueness.
f ( t k )will produce an arbitrarily large error in the recovered 2) If {tk} is a set of stable sampling for then every
function. To guard against this, we must introduce an addi-
tional requirement of stability : small errors in reading the
interval of length r must contain at least (rm(S)(27r)- ’
- A log+ r - B) of the points of {tk}, with A and B
samples shall produce no more than correspondingly small appropriate constants.’ Hence the density of {tk} is at
errors in the recovered function, or, in terms of the differ- least m(S)/2n. Stable sampling cannot be performed at

zyx
ence between the true andreconstructed signal, if the values lower than the Nyquist rate.
of a function ~ E B ( at S )the points { t r } are small, h(t) shall 3) If {tk} is a set of interpolation for B(S),no interval of
itself be small. There are many ways of measuring the size ’
length r can contain more than (rrn(S)(27z)- + C log’ r
of a signal ; we choose energy and put our condition as +D ) of the points of {tk}, with C and D appropriate
follows. constants. Hence the density of {tk} is at most m(S)/27r.
Dejinition : The set of points { tk} is a set of stable sampling Data cannot be sent as samples at higher thanthe
for B(S)if there exists a constant K such that Nyquist rate.
Wewill indicate the proofs of statements 2 ) and 3) and
close with some remarks concerning these problems for
bounded, rather than finite-energy, functions.
for all f ( t ) E B ( S ) .
We can now see that a set of stable sampling is always a DISCUSSION
AND PROOFS

set of uniqueness, whereas the reverse is not true. More- We begin with an intuitive explanation of the proof. In
over, our definition does not commit us to any particular what follows wewilluse the term “sampling” to mean
recovery scheme,and so describes the most general circum- exclusively “stable sampling.”
stances in which sampling can be performed in practice. Suppose { t k } is a set of sampling for B(S) and I is an
Takingup next the problem of transmitting data as interval containing n ofthe points of {tk}.Supposef(t)EW(S)
samples, we consider instants t = {tk} at which square- has little of its energy outside I ; we will referto such a func-
summable values can be prescribed arbitrarily for a signal tion as being well concentrated on I . If we can presume that
f ( t ) E B ( S ) Here
. too we formulate the matterindependently the samples f ( t k ) at points tk outside Z are also small, then,
of any method of computation. by virtue of our discussion of stability, replacing these
Dejinition : The set of points { tk} is a set of interpolation sample values by zero should not too drastically affect our
for W ( S )if, given any set of numbers {ak} with C lak12< co,
there exists f ( t ) E g ( S )with f(tk)=ak. With log’ r = max (log r, 0).
SAMPLING,
LANDAU:

zyxwv
TRANSMISSION,
DATA z
zyxwvutsrq
reconstruction of f ( t ) . We conclude that a function well
concentrated on I can be substantially determined by the
AND NYQUIST RATE 1703

n measurements which give f ( t k ) for tk in I , hence that the

collection of all functions of g(S)well concentrated on I
is, in some sense, no more than n-dimensional. Taken in which represents the dual problem of concentrating on S
conjunction with a similar argument about sets of inter- functions whose Fourier transforms are supported on an
polation, this seems to suggest that determining the number interval of length r. Now by (6)
of independent signals which are well concentrated on I I2
might yield information about our current problem. We
have already had occasion to discuss the latter question in
terms of the eigenvalues of a certain We
will
show that this formulation does indeed have a close con-

cerns us here.

zyxwvu
nection with the sampling and interpolation which con-

zyxwvut
If S consists of intervals of length Zi, we find, after some
manipulation,

zyxwvutsrqpo
With I an arbitrary given set, we begin by determining
that function of g ( S ) which is most concentrated on I , i.e.,
that f ~ g ( Sfor
) which the concentration

fIf (t)I2 dt
N
-
1
n2
log+ zir - 1

Irn
i= 1
(3)
A = I f(t)I2 dt m(S)r
2--
271
A log+ r - B,

I zyxwvuts
-m

achieves its largest value. This is a standard problem of with constants A and B depending only on S . Thus, from
maximizing a quadratic form and leads immediately to the ( 5 ) and (8),
eigenvalue equation 1k j.k(r, s){ 1 - ik(r, s)} A log+ + B. (9)

nk4k(t) =- $k(x)w(t - dx x) , (4) It follows from (9) that the numberof eigenvalues &(r, S )
& I
lying between any two fixed bounds 6 > 0 and y c 1 can
in which the Fourier transform of w is the characteristic grow at most logarithmically with r. Specifically, each such
function of the set S . The kernel being hermitian, positive eigenvalue contributes to the left-hand side of (9) an amount

zyxwvuts
definite, and square-integrable, (4) is of the Hilbert-Schmidt no smaller than a=min {ql- 6), fil- y ) } >O; hence the
type, so that eigenfunctions and eigenvalues do exist (see number of these eigenvalues cannot exceed Ala log' r + B/a.

zyxwvutsr
p. 242, Riesz and NagyI61); we denote the latter, in nonin- Taken together with ( 5 ) and (8), this observation shows that,
creasing order, by &(I, S), k=O, 1, 2, . . . , and by &(r, S ) if asymptotically in r, the number of lk(r, S ) near 1 behaves
I is a single interval of length r. like rn(S)(271)-'r. But we can sharpen this statement by
By definition, i j -l(r, S ) is the largest concentration on an means of the following two lemmas (proved in the Ap-
interval of length r achievable by a function of W ( S )which pendix), whichserve as well to establish the connection
is orthogonal to the j - 1 eigenfunctions 4 0 ( x ) ,. . . , 4 j - 2 ( x ) between the eigenvalue problem and sampling and inter-

zyxwvutsr
previously determined ; intuitively, then, I.j- l(r, S ) repre- polation.
sents the concentration of the jth most concentrated func-
tion. Our inquiry intothenumber of independent well Lemma 1
concentrated functions therefore leads us to consider the Let S be bounded and {tk} be a set of sampling for g ( S ) ,
behavior of the sequence {ik(r, S ) } as a function of r. There whose points are separated by at least d >O. Let I be any
are two simple consequences of (4) whichyield a lot of compact set, I + be the set of points whose distance to I is
information :

1
m
&(r, S ) = trace =
less than d/2, and n(Z+) be the number of points of (tk}
contained in I +. Then &+)(I, S)I y < 1, where y depends
on S and (tk} but not on I .
k= 0 2n Lemma 2
30

1 I.z(r, S ) = double integral of (kernel12. Let S be bounded and {tk} be a set of interpolation for
(6)
k=O g ( S ) , whose points are separated by at least d >O. Let I be
any compact set, I - be the set of points whose distance to
To evaluate the latter most easily, we observe that (4) may the complement of I exceeds d/2, and n ( I - ) be the number
be converted to other equationswithout changmg the eigen- of points of { tk} contained in I - . Then An(l-) - l(r, S )2 6 >0,
values. One such transformation, obtained by introducing where 6 depends on S and {tk} but not on I .
into (4) the definition of w, multiplying both sides by e-iz", We have already remarked that the eigenvalues A k ( z , S )
and integrating over I , gives remain unchanged when we reverse the roles of the time and
 zy
zyxwvutsrqponm
zyxwvu
1704

zyxwvutsrq
zyxwvutsrq
zyxwvutsr
frequency domains. It is also easy to see from (4) that they
are unaffected by a rescaling of time and frequency which
replaces Z by aZ and S by a- ' S , with a any constant.2 By
applying both of these permissible transformations, we con-
clude that i k - l(r, S ) is also the kth eigenvalue p f the prob-
lem of concentrating functions whose frequencies lie in a
single interval of length 2n (i.e., bandlimited functions) on
PROCEEDINGS OF THE IEEE,OCTOBER

Proof: We follow the proof of Theorem 1, interchanging

the roles of Lemmas 1 and 2. Thus with Z an interval of
length r, n(Z-)2 n(Z)-2, and, by Lemma 2, there exists a
constant 6 >0 independent of r such that
1967

the set (r/2n)S. Butnow theordinary sampling theorem

shows the integer points to be a set both of sampling and
interpolation. When S is the union of N intervals, the num-

zyxwvutsrqp
ber of these points contained in S + is at most (2n)-'rrn(S) so that
+ 2N, and their number in S - is at least (2n)-'rm(S)- 2 N .

zyxwvut
Thus Lemmas 1 and 2 establish the existence of absolute
constants yo and 6, such that
4(2x)-'rm(s)1+2N(r,S ) Yo < 1,
i[(2x)-'rm(S)]-2N-1(r, 2 > O;
here we use the notation [x] to denote the integer part of
x. It follows from their construction that y o and do may be
taken to be 3, but we will not prove this here.
(10)
(11)
(n(Z) - 3) - ([rm(S)(2n)-'] + 2 N ) < C'log' r

whence
n(Z) 5 rm(S)(2n)-'

Theorem 2 is established.
+ Clog' r + D.
+ D',

It follows immediately that the density of a set of inter-

polation cannot be higher than rn(S)/2n.
In the preceding argument we used Lemmas 1 and 2, to- CONCLUDING
REMARKS
gether with known sets of sampling and interpolation, to
determine the behavior of the { i k ( r ,S ) } . Now the lemmas 1 ) The aforementioned density results apply equally well
allow us in turn to apply this information to arbitrary sets to functions of several variables and to frequency sets S
of sampling or interpolation. more complicated than finite unions of intervals. A complete
account of this extension is to be found elsewhere[31.
Theorem 1 2) It may well be argued that functions of finite energy
If { t k } is a set of sampling for @S), every interval of are too special a class, and that bounded functions present
'
length r must contain at least (rm(S)(2n)- -A log' r - B) greater practical interest. The problem of describing fre-
of the points of { t k } ,with A and B appropriate constants. quency content is rather severe in this case, the Fourier
Proof: Let Z be an interval of length r, and n(Z) the num- transform not being available, but meaningful definitions
ber of points of { t k } contained in I ; since I is a single in- can be given (using distributions, for example) for the class
+
terval, n(Z + ) < n(Z) 2. By Lemma 1, there exists a constant g b ( S )of bounded signals with frequencies in S ; we will not
y < 1 independent of r such that discuss the details here. We can then formulate the notions
of stable sampling and interpolation as follows.
& ( I ) + 2(r, S ) Y < 1. (12) Definition : The points { t k } form a set of stable sampling
But from ( 1 1) for g b ( S ) if there exists a constant K such that

i[rm(S)(2n)-']-ZN-1(T,S ) 2 > O, (13)

and, as we have argued in connection with (9), the number
of eigenvalues lying in the range 6, <&(r, S ) I increases at for all C#J E g b ( S ) .
most logarithmically with r. Thus Definition : The points { t k } form a set of interpolation
for if,given any bounded sequence of numbers {a,},
([rm(S)(2n)-'] - 2N - 1 ) - (n(Z) + 2) IA'log' r + B', there exists a function $(t)eab(S)with & t k ) = a k .
with constants A' and H,whence Such sets were fust defined and studied by Beurling. One
can show the following.
'
n(Z) 2 r m ( S ) ( 2 ~ ) -- A log' r - B.
a) The density of a set of sampling for g b ( S ) is always
Theorem 1 is established. strictly larger than m(S)/2n;thus, for bounded signals,
It follows immediately that thedensity of a set of sampling stable sampling requires a greater than Nyquist rate.
cannot be lower than m(S)/2n. b) The density of a set of interpolation for g b ( S ) is at
most m(S)/2n,and strictly smaller if S is a single in-
Theorem 2 terval. We suspect it to be always strictly smaller, but
If { t k } is a set of interpolation for g ( S ) , no interval of have not succeeded in proving this. In any case, trans-
length r can containmore than (rm(S)(2n)-' + C log'r+ D) mission of data as samples of bounded signals can
of the points of I t k } , with C and D appropriate constants. certainly not be performed at a greater than Nyquist
rate.
We denote by aI the set of points of the form af for t E I. Details may again be found elsewherer3].
 zyxwvut
zyxwvutsrqp
zyxwvutsrqp
SAMPLING,
LANDAU: TRANSMISSION,
DATA AND NYQUIST RATE 1705

APPENDIX K such that the interpolating functions f ( t ) in 6(S) satisfy

Proposition I
1
Let S be bounded, and let ( t k } be a set of interpolation
for W(S). Then the points of { t k } are separated by at least
oc

-m
If(')12
dt I 1 lakl2 = 1lf(tk)12.

formed in a stable way.

zyxwvu
some positive distance d, and the interpolation can be per-

1zyxwvut
Proof: By writing f € W ( S ) as the inverse transform of its
This shows that when interpolation is performed in d?(S),it
is necessarily stable. Proposition 1 is established.

zyxw
Fourier transform, and applying Schwarz's inequality and Lemma 1
Parseval's theorem to this representation, we find

zyxwvutsrqp
Let S be bounded and { t k } be a set of sampling for B(S),
whose points are separated by at least d >O. Let I be any
IK, lf(t)I2 dt, (15) compact set, I + be the set of points whose distance to I
-m is less than 4 2 , and n(I+ ) be the number of points of { t k }

If'(t)12 IK, lm-m

lf(t)I2 dl, (16)
+
contained in I . Then
on S and ( t k ) but not on I .
)(I, S)I y < 1 , where y depends

Proof: By the Weyl-Courant lemma,

with constants K , and K , depending only on S . Thus for
signals in W ( S )a bound on energy implies a corresponding
bound on both the signal and its derivative.
If the points ( t k } are not separated by at least some fixed
positive distance, we can find two disjoint subsequences, J-m
which we denote by { p k } and { q k ) , with the property that

zyxwvutsrqp
Ipk-qkI I l / k 2 , k = 1 , 2 , . . . . Now let us prescribe the values with ck any subspace of dimension k. Thus to prove Lemma
l / k at t = p k , - l / k at t = 4 k , and 0 at every remaining point 1 , it is sufficient to show that we can keep the concentration
of { t k } ; since these values are square-summable, and { t k } is of a function of B(S)over I bounded away from 1 by im-
a set of interpolation, they will be assumed by some posing n(I +) orthogonality conditions. The following
simple device accomplishes this. Let h b ) be a square-
f(t)Eg(S).But then 2 / k = f b k ) - f ( q k ) 5 l p k - qkl lf'(sk)l with
sk apoint lyingbetween p k and q k . Hence I f ' ( s k ) l > 2 k , integrable function which vanishes for ly[ > d / 2 and whose
k = 1 , 2, . . . , contradicting ( 1 6 ) . This establishes the first Fourier transform H ( o ) satisfies IH(o)l> 1 for OES.Such a
part of the proposition. function certainly exists, since the transforms of functions

zyxwvutsrqpon
To prove the second part, we appeal toan abstract vanishing on lyl> d / 2 can be made to approximate uniformly
principle. In W(S), let usview functions as vectors and an arbitrary continuousfunction on any bounded set. Now
energy as the square of length ; this makes B(S) into a given f€&(S), we form
Hilbert space. Let d?'(S) be the subspace of g ( S ) consisting
of those functions which vanish at all the points { t k } . This
subspace is closed, since, by ( 1 5), convergence in energy of
functions of to a function h(t) implies uniform con- 1 r

vergence as well, so that h(t)also must vanish on { t k } . The

reason for introducing So(S)is.that it offers the possibility
of reducing the energy of interpolating functions. Specif- whence by Schwarz's inequality
ically, starting with one f(t) in W ( S )which interpolates to
a prescribed set of values {ak} at { t k } , we can form the col-
lection { f ( t ) + g ( t ) } with g(t)E&O(S); all of these functions
ls(t)I2Ic fIt-yI < d / 2
IfCv)12 dY, (19)
continue to interpolate, and we may ask for that member
of the collection which has least energy or length. This with c = ( 2 7 c ) - ' flh(t)12 d t . We observe that the Fourier
problem has as its solution the function which represents transforms F and G o f f and g, respectively, are related by
the projection off ontothe orthogonal complement in B(S) G = F H . It follows that g E W ( S ) , so that by the definition of
of d?O(S), a subspace which we denote by &(S). Thus inter- 'k}

polation can be performed in &@), whereupon the inter- m

polating function is uniquely determined and has minimum (dt)12 dt I 1l d t k ) I 2 , (20)

zyxwvutsrq
-00
energy.
Now let us consider the transformation T which assigns and by Parseval's theorem
to a square-summable sequence { a t } that function of 8 ( S )
r m r r
which interpolates to ak at t = t k . By hypothesis, T is de-
fined on the whole space 1, of square-summable sequences,
maps 1, into the Hilbert space d?(S), and, in view of ( 1 5 ) ,
has a closed graph. Then by the closed graph theorem (see
Loomi~[~], p. 1 8 ) T is bounded, i.e., there exists a constant
= Im -m
Ig(t)12d t .
1706

zyxwvzyxw
zyxwvutsrq
zyxwvutsrq OF THE IEEE, OCTOBER
PROCEEDINGS 1967

--a:
zyxw
zyxwvutsrqpo
Now subjecting f to the n(Z +) orthogonality conditions

f ( y ) h ( t k - y ) d y = 0, tk E I+ 9

ensures that g ( t k ) = O , t k E I + , whereupon the sum in (20)

is extended only over t k $ I + .For those t k , the corresponding
with some fixed constant K . For each t k , we let & ( t ) E & ( S )
be the function whose value is 1 at t k and 0 at every Z j ,
j # k ; these functions are linearly independent.
We let h be the same as in the proof of Lemma 1, and
construct functions + k E a ( S ) such that
1 “
integrals in (19) are taken over non-overlapping intervals,
4k(t) =
271 - m
7
Il/kb)h(t - y )d y ; (24)
each contained outside I ; thus combining (21),(20), and

zyxwvutsr
(19) we find to see that these exist, we need only take the Fourier trans-
form of (24) and note that I H ( o ) l 2 1 for OES.The $’s are
likewise linearly independent, so that, letting C be the sub-
space of W ( S ) spanned by the functions +k with t k in I - ,
we obtain a subspace of dimension n(I -).
or Now given ~ E Cwe , form the function g of (18). In view
of (24), g is a linear combination of the $ k , hence lies in
j
zyxwvutsrq
zyxwvu
If (t)12 d t &(S) and satisfies (21).However, since the only q i k figuring
I < l - - =1y < l . in g are those with t k in I - , the values of g necessarily
jm If @)I2 d t
--a:
Kc vanish at every other t k , so that
I-m

The constant y depends on S and ( t k } since c and K do, but

J -m
Ig(t)I2 dt 5 K 1 Ig(tk)(2.
I-
(25)
not on I. Applied to (17), this proves Lemma 1.
The relations (19) and (21) between f and g continue to
Lemma 2 hold, and when the points t k are in I - the corresponding

zyxwvu
zyxwvu
Let S be a bounded set and { t k } be a set of interpolation integrals in (19) are extended over non-overlapping in-
for W(S),whose points are separated by at least d >O. Let tervals, each contained in I , so that
I be any compact set, I - be the set of points whose distance
to the complement of I exceeds d/2, and n(Z-) be the num-
ber of points of {t,} contained in I - . Then A,,,- ) - l(Z, S )
2 6 >0, where 6 depends on S and {t,} but not on I .
Proof: We again begin with the Weyl-Courant lemma in
the form
I-

J-m

with ck any subspace of dimension k. Any proof of (17) can which, applied to (22), proves Lemma 2. As in Lemma 1,
be modified to yield(22), but the following independent the constant 6 does not depend on I.
argument serves as well. Let L k - l be the subspace of B(S)
spanned by &,,. . . ,dk-2,the first k - 1 eigenfunctions. REFERENCES
Since the dimension of Lk-lis lower than that of ck, any [‘I A. Beurling and P. Malliavin, “On the closure of characters and
linear transformation of C, into & - I , in particular the zeros of entire functions,” Acta Mathematica, vol. 118, pp. 79-93, May
orthogonal projection onto L,-,, must annihilate some 1967.
H . J. Landau, “A sparse regular sequence of exponentials closed
nonzero element of ck. Thus Ckcontains a function ff0 on large sets,” Bull. Am. Math. Soc.,vol. 70, pp. 5 6 5 6 9 , July 1964.
orthogonal to Lk- and (22) follows from the definition of -, “Necessary density conditions for sampling and interpolation
of certainentire functions,” ActaMathematica, vol. 117, pp. 37-52,
‘ k - 1. February 1967.
We have shown in Proposition 1 that it is possible to [41 H. J. Landau andH. 0.Pollak, “Prolate spheroidal wave functions,
perform the interpolation in a subspace &(S) of a(S),in Fourier analysis and uncertainty.111,” Bell Sys. Tech.J., vol. 41, pp. 1295-
which every function g ( t ) satisfies 1336, July 1962.
c5] L. H. Loomis, An Introduction toAbstract Harmonic Analysb.
m New York: Van Nostrand, 1953.
F. Riesz and B. S. Nagy, Functional Analysis. New York: Ungar,
(23) 1955.
-m