1 Baire's Theorem and Its Applications

BAIRES THEOREM AND ITS APPLICATIONS
EDUARD KONTOROVICH
The completeness of Banach space is frequently exploited. It depends on the following theorem about complete metric spaces, which in itself has many applications
in several parts of mathematics. It implies two of three most important theorems
which makes a Banach spaces useful tools in analysis, the Banach-Steinhaus theorem
and the open mapping theorem. The third is the Hanh-Banach extension theorem, in
which completeness plays no role.
1. Baires Theorem
Theorem 1.1. (Baires Theorem) If X is a complete metric space, the intersection
of every countable collection of dense open subsets of X is dense in X.
Proof. Suppose V1 , V2 , V
T3 , ... are dense and open in X. Let W be any open set in X.
We have to show that Vn has a point in W if W 6= .
Let be a metric in X; let us write
B(x, r) = {y X : (x, y) < r}
r) be the closure of B(x, r).
and let B(x,
Since V1 is dense, W V1 is a nonempty set, and we can therefore find x1 and r1
such that
1 , r1 ) W V1 and 0 < r1 < 1.
B(x
If n 2 and xn1 and rn1 are chosen, the denseness of Vn shows that Vn
B(xn1 , rn1 ) is not empty, and we can therefore find xn and rn such that
n , rn ) Vn B(xn1 , rn1 ) and 0 < rn < 1 .
B(x
n
By induction, this process produce a sequence {xn } in X. If i > n and j > n, the
construction shows that xi and xj both lie in B(xn , rn ), so that (xi , xj ) < 2rn < n2 ,
and hence {xn } is a Cauchy sequence. Since X is complete, there is a point x X
such that x = lim xn .
n , rn ) if j > n, it follows that x lies in each
Since xj lies in the closed set B(X
n , rn ), and therefore x lies in each Vn and W. This completes the proof
B(X
Corollary 1.2. In a complete metric space, the intersection of any countable collection of dense G is again a dense G .
Proof. This follows from the theorem, since every G is the intersection of a countable
collection of open sets, and since the union of countably many countable sets is
countable.
Definition 1.3. Call a set E X nowhere dense if its closure E contains no

nonempty opens subset of X.
Date: May 2, 2009.
1
Any countable union of nowhere dense sets is called a set of the first category;
all other subsets of X are of the second category. Theorem 1.1 is equivalent to the
statement that no complete metric space is of the first category. To see this, just take
complements in the statement of Theorem 1.1.
It is simple to see that any subset of a set of first category is of first category. Also,
the union of any countable family of first category is of first category. However, the
closure of a set of first category is not in general of first category. In fact, the closure
of a linear set A is of first category if and only if A is nowhere dense. It is follows
that if X is a complete metric space and X = E (X/E) where E is of first category
then X/E is of second category.
Example 1.4. (Liouville numbers) A real number z is called algebraic if it satisfies
some equation of the form
a0 + a1 z + ... + an z n = 0.
Any real number that is not algebraic is called transcendental.
A real number x is called Liouville number if x is irrational and has the property
that for each positive integer n there exist integers p and q such that
p
1
| x |< n and q > 1.
q
q
k!
(n1)!
For example, x =
)
1 1/10 is a Liouville number.(Take q = 10
It is true that every Liouville number is transcendental. If we shall try to examine
the set E of Liouville numbers, then well understand that by using the Theorem 1.1
E is of second category. The full proof we can find in [2].
Example 1.5. (Kuratowski-Ulam theorem)

If E X Y and x X, the set Ex = {y : (x, y) E}
is called the x-section of E.
The theorem is: If E is a plane set of first category, then Ex is a linear set of
first category for all x except a set of first category. If E is a nowhere dense subset
of the plane X Y, then Ex is a nowhere dense subset of Y for all x except a set of
first category in X.
S
S Proof: The two statement are essentially equivalent. For if E = Ei , then Ex =
i (Ei )x . Hence the first statement follows from second. If E is nowhere dense, so
is E, and Ex is nowhere dense whenever (Ex ) is of first category. Hence the second
statement follows from the first. It is therefore sufficient to prove the second statement
for any nowhere dense closed set E.
Let Vn be a countable base for Y, and put G = (X Y ) E. Then G is dense
open subset of the plane. For each positive integer n, let Gn be the projection of
G (X Vn ) in X, that is,
Gn = {x : (x, y G) f or some y Vn }.
To more details see [2]
2. Banach-Steinhaus Theorem
Definition 2.1. Let f be a real valued function on a topological space. If
{x : f (x) > }
is a open for every real , f is said to be lower semicontinuous.

The following property of semicontinuous functions are almost immediate consequence of this definition.
The supremum of any collection of lower semicontinuous functions is
lower semicontinuous.
Theorem 2.2. (Banach-Steinhaus theorem) Suppose X is a Banach space, Y is
a normed linear space, and { } is a collection of bounded linear transformations of
X into Y, where ranges over some index set A. Then either there exists an M <
such that
k k M
for every A, or
sup k x k=
A
for all x belonging to some dense G in X.

Proof. Put
(x) = supA k x k, (x X)
and let
Vn = {x : (x) > n}, (n = 1, 2, 3, ...).
Since each is continuous and since the norm of Y is a continuous function on Y
(an immediate consequence of the triangle inequality) each function x k x k is
continuous on X. Hence is lower semicontinuous, and each Vn is open.
If one of this sets, say VN fails to be dense in X, then there exist an x0 X and
r > 0 such that k x k< r implies x0 + x 6 VN ; this means that (x0 + x) N, or
k (x0 + x) k N
for all A and all x with k x k r. Since x = (x0 + x) x0 , we then have
k x kk (x0 + x) k + k x0 k 2N,
and it follows that M = 2N/r.
T
The other possibility is that every Vn is dense inTX. In that case, Vn is a dense
G in X, by 1.1; and since (x) = for every x Vn , the proof is complete.
The geometric meaning of following theorem is: Either there is a ball B in Y (with
radius M and center 0) such that every maps the unit ball of X into B, or there
exists x X (in fact, a whole dense G of them) such that no ball in Y contains x
for all .
Example 2.3. (Fourier series convergence problem) Is it true for every f
C(T ) that the Fourier series of f convergence to f (x) at every point x?
Let us recall that the n-th partial sum of the Fourier series of f at the point x is
given by
H
1
sn (f ; x) = 2
f (t)Dn (x t)dt, (n = 0, 1, 2, ...),
where
Dn (t) =
n
X
eikt .
k=n
The problem is to determine whether

lim sn (f ; x) = f (x)
for every f C(T ) and for every real x. It is true that the partial sums do converge
to f in the L2 norm, and therefore Theorem ?? implies that each f L2 (T) [hence
also each f C(T )] is the pointwise limit a.e. of some subsequence of the full sequence
of the partial sums. But the does not answer the present question.
The Banach-Steinhaus theorem answers the question negatively. The full proof
exists in [3]
Example 2.4. Some important standard applications of Banach-Steinhaus theorem.
For example the following: Let C be a weak-star compact subset in the dual V of a
Banach space V . Prove that C is norm-bounded.
3. The Open Mapping Theorem
Theorem 3.1. (open mapping theorem) Let U and V be the open unit balls of the
Banach space X and Y. To every bounded linear transformation of X onto Y there
corresponds a > 0 so that
(U ) V.
The symbol V stands for the set {y : y Y }, i.e., the set of all y Y with
k y k< .
It follows from theorem conditions and the linearity of that the image of every
open ball in X, with center x0 , say, contains an open ball in Y with center at x0 .
Hence the image of every open set is open. This explain the name of the theorem.
Proof. Given y Y, there exists an x X such that x = y; if k x k< k, it follows
that y (kU ). Hence Y is the union of the sets (kU ), for k = 1, 2, 3, ... Since Y
is complete, the Baire theorem implies that there is a nonempty open set W in the
closure of some (kU ). This means that every point of W is the limit of a sequence
{xi }, where xi kU ; from now on, k and W are fixed.
Choose y0 W, and choose > 0 so that y0 + y W if k y k< . For any such y
0
00
there are sequence {xi }, {xi } in kU such that
0
00
(2) xi y0 , xi y0 + y, (i ).
0
00
Setting xi = xi xi , we have k xi k< 2k and xi y. Since this holds for every y

with k y k< , the linearity of shows that the following is true, if = \2k :
To each y Y and to each > 0 there corresponds an x Xsuch that
(3) k x k 1 k y k and k y x k< .
This is almost the desired conclusion, as stated just before the start os the proof,
expect that there had = 0.
Fix y V, and fix > 0. By (3) there exists an x1 with k x k< 1 and
(4) k y x1 k< 12 .
Suppose x1 , ..., xn are chosen so that

(5) k y x1 ... xn k< 2n .
Use (3), with y replaced by the vector on the left side of (5), to obtain an xn+1 so
that (5) holds with n + 1 in place of n, and
(6) k xn+1 k< 2n , (n = 1, 2, 3, ...).
If we set sn = x1 + ... + xn , (6) shows that {sn } is a Cauchy sequence in X. Since X is
complete, there exists an x X so that sn x. The inequality k x1 k< 1, together
with (6), shows that k x k< 1 + . Since is continuous, sn x. By (5), sn y.
Hence x = y.
We have now proved that
((1 + )U ) V,
or
(7) (U ) (1 + )1 V,
for every > 0. The union of the sets on the rights of (7), taken over all > 0, is
V. This proved the theorem.
References
1.
2.
3.
4.
5.
J. Kelly General Topology, Nauka, 1961.

John C. Oxtoby, Mrasure and Category, Springer, 1971.
W. Rudin, Real and complex Analysis, McGraw-Hill series in higher mathematics, 1974
A. N. Kolmogorov, S. B. Fomin, Functional Analysis, Nauja, 1972
H. L. Royden, Real Analysis, 1963, no.4, 353-355.
Department of Mathematics, Bar-Ilan University, 52900 Ramat-Gan, Israel

E-mail address: [email protected]
URL: http://www.math.biu.ac.il/ kantore

1 Baire's Theorem and Its Applications

Uploaded by

Copyright:

Available Formats

1 Baire's Theorem and Its Applications

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 Baire's Theorem and Its Applications

Uploaded by

Copyright:

Available Formats

BAIRES THEOREM AND ITS APPLICATIONS

Definition 1.3. Call a set E X nowhere dense if its closure E contains no

Example 1.5. (Kuratowski-Ulam theorem)

is a open for every real , f is said to be lower semicontinuous.

for all x belonging to some dense G in X.

The problem is to determine whether

Setting xi = xi xi , we have k xi k< 2k and xi y. Since this holds for every y

Suppose x1 , ..., xn are chosen so that

J. Kelly General Topology, Nauka, 1961.

Department of Mathematics, Bar-Ilan University, 52900 Ramat-Gan, Israel

You might also like