next up previous


The results

To state our main result we need some definitions. Recall that a Banach space X has the Uniform Opial Property [P3] if for every c>0 there exists r=r(c)>0 such that tex2html_wrap_inline693 for every tex2html_wrap_inline553 with tex2html_wrap_inline697 and every weakly null sequence tex2html_wrap_inline537 in X with tex2html_wrap_inline703 .

To any Banach space X we can associate its modulus of Opial [LTX] by

displaymath707

where tex2html_wrap_inline709 and the infimum is taken over all tex2html_wrap_inline553 with tex2html_wrap_inline697 and all weakly null sequences tex2html_wrap_inline537 in X with tex2html_wrap_inline703 .

In order to simplify the statement of the following theorem, for tex2html_wrap_inline721 we define the number

displaymath723

Since the well known inequality (see [X])

displaymath725

holds for all tex2html_wrap_inline727 , then tex2html_wrap_inline729 .

theorem64

Proof

First, let us observe that we can suppose that there exists a>0 such that

displaymath765

Suppose, for a contradiction, that Y lacks the WFPP. Then there exists a nonempty, convex and weakly compact set tex2html_wrap_inline769 and a fixed point free tex2html_wrap_inline771 -nonexpansive mapping tex2html_wrap_inline773 such that K is minimal for T. Since T has no fixed point in K, then d=diam (K) >0 and we can suppose that d=1. Let tex2html_wrap_inline537 be an afps for T in K. Since every subsequence of tex2html_wrap_inline537 is again an afps for T, we can suppose that tex2html_wrap_inline537 is an afps which is weakly convergent and then, by translation of K, that K is minimal for T, tex2html_wrap_inline805 and that tex2html_wrap_inline537 is weakly null.

Consider the subset tex2html_wrap_inline809 of tex2html_wrap_inline621 defined by

displaymath813

where tex2html_wrap_inline815 .

It is easy to see that tex2html_wrap_inline809 is a nonempty, closed, convex and tex2html_wrap_inline649 -invariant set. Since tex2html_wrap_inline805 , we have, by Lemma 1, that

displaymath823

Let tex2html_wrap_inline627 be any element of tex2html_wrap_inline809 . Since K is weakly compact, there exists a subsequence tex2html_wrap_inline831 of tex2html_wrap_inline833 such that

displaymath835

and such that tex2html_wrap_inline831 converges weakly to tex2html_wrap_inline839 . By passing to subsequences, we can assume that the following limits exist:

displaymath841

Fix tex2html_wrap_inline843 . Since the sequence tex2html_wrap_inline845 converges weakly to tex2html_wrap_inline847 , then

displaymath849

so that

displaymath851

Hence, we have that

displaymath853

Since tex2html_wrap_inline855 is an afps for T, we have that tex2html_wrap_inline859 , by the Goebel-Karlovitz lemma, and then

displaymath861

displaymath863

>From this we get that

displaymath865

Since t<1, we can consider the sequence tex2html_wrap_inline869 defined by

displaymath871

Then tex2html_wrap_inline869 is weakly convergent to 0 and tex2html_wrap_inline877 . By the definition of tex2html_wrap_inline879 we have that

displaymath881

On the other hand, since tex2html_wrap_inline883 , we have that

displaymath885

Hence, we have that tex2html_wrap_inline887 and then tex2html_wrap_inline889 .

Pick any tex2html_wrap_inline891 . Since tex2html_wrap_inline893 , there exists a subsequence tex2html_wrap_inline895 of tex2html_wrap_inline831 such that tex2html_wrap_inline899 for every positive integer k.

On the other hand, we have that

displaymath903

and also that

displaymath905

>From the above facts we conclude that

displaymath907

and, since tex2html_wrap_inline891 is arbitrary,

displaymath911

Finally, we get that

displaymath913

which contradicts (1).

corollary177

Proof

Since tex2html_wrap_inline497 is an euclidean norm, we have that tex2html_wrap_inline933 for all tex2html_wrap_inline709 and then tex2html_wrap_inline937 . On the other hand,

displaymath939

for all tex2html_wrap_inline941 and then

displaymath943

By the Theorem, we only have to show that

displaymath945

i.e. that there exists tex2html_wrap_inline941 such that

displaymath949

which is equivalent to

displaymath951

The left hand side F(a) in the above inequality takes its maximum value at tex2html_wrap_inline955 . Then a solution of the above inequation is every B>1 such that

displaymath959

and this happens if tex2html_wrap_inline961 .

In order to obtain a similar result for tex2html_wrap_inline963 when tex2html_wrap_inline569 and tex2html_wrap_inline967 we will need to find the maximum a of a positive real function

displaymath969

defined for tex2html_wrap_inline971 .

It is easy to see that

displaymath973

On the other hand, for nonnegative a

displaymath977

Since the term tex2html_wrap_inline979 is positive, the sign of the derivative F'(a) is the same that the sign of the continuous function

displaymath983

We have

displaymath985

Moreover, for p>2

displaymath989

displaymath991

Therefore, there exists tex2html_wrap_inline993 in the open interval tex2html_wrap_inline995 , for which tex2html_wrap_inline997 . By arguments from elementary Calculus, it is straightforward to see that tex2html_wrap_inline993 is the unique maximum of tex2html_wrap_inline1001 . One can search also an interval for this maximum in the case 1<p<2, but we do not repeat this for shake of brevity.

corollary258

Proof

Since (see [X]) tex2html_wrap_inline1025 for all tex2html_wrap_inline709 , then tex2html_wrap_inline1029 . On the other hand, (see [Do]), for all tex2html_wrap_inline941

displaymath1033

By the Theorem, we only have to show that there exists tex2html_wrap_inline941 such that

displaymath1037

i.e.

displaymath1039

which is equivalent to

displaymath1041

The left hand side tex2html_wrap_inline1043 in the above inequality takes its maximum value at some tex2html_wrap_inline1045 . Then a solution of the above inequation is every B>1 such that

displaymath1049

and this happens if tex2html_wrap_inline1051 .

remark294

The effective computation of the constant tex2html_wrap_inline1053 in the above corollary seems to be hard, spcecially if p is not a natural number. Nevertheles, it is clear that

displaymath1057

The right hand side in this inequality fournishes bounds of stability in tex2html_wrap_inline963 which are greater than tex2html_wrap_inline1061 , at least for p<6.

remark294

Let X be a Banach space with Opial modulus tex2html_wrap_inline879 . In [LTX] the authors proved that tex2html_wrap_inline879 is continuous on tex2html_wrap_inline1071 and that, for tex2html_wrap_inline1073 ,

displaymath1075

Then, if tex2html_wrap_inline879 satisfies the condition tex2html_wrap_inline1079 , for tex2html_wrap_inline1081 , we have

displaymath1083

Therefore, for the Banach spaces X with tex2html_wrap_inline1079 , tex2html_wrap_inline1089 , ( tex2html_wrap_inline721 ).

It is obvious that, for such spaces, the stability bound given by our theorem is M(X).

Spaces verifying tex2html_wrap_inline1079 are, for example, tex2html_wrap_inline1097 (see [LTX], Remark 3.1.), and the Bynum spaces tex2html_wrap_inline1099 ( tex2html_wrap_inline569 ).

As a counterpart of the above remark we can state the following result.

corollary316

Proof

Since tex2html_wrap_inline879 is a continuous function, there exists tex2html_wrap_inline1111 such that tex2html_wrap_inline1113 . By definition we obtain

displaymath1115

Now, we can apply the theorem for B=1. Since for every a> 0 one has tex2html_wrap_inline1121 , we have

displaymath1123

displaymath1125

Notice that in the Theorem 3.3. of [Do] has been proved that tex2html_wrap_inline1127 implies M(X)>1, for reflexive Banach space, in some sense, a dual version of last Corollary.

Acknowledgement. The authors would like to thank Professor T. Domínguez Benavides for his comments on the first version of this paper, mainly for the result of Corollary 3.


next up previous

Fixed Point Theory on the Web
Fri Oct 25 17:49:18 MDT 1996