Fixed Point Theory and Applications
Volume 2010 (2010), Article ID 362829, 13 pages
doi:10.1155/2010/362829
Research Article

The Fixed Point Property of Unital Abelian Banach Algebras

Department of Mathematics, Faculty of Science, Chiang Mai University, Chiang Mai 50200, Thailand

Received 28 October 2009; Accepted 22 January 2010

Academic Editor: Anthony To Ming Lau

Copyright © 2010 W. Fupinwong and S. Dhompongsa. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

We give a general condition for infinite dimensional unital Abelian Banach algebras to fail the fixed point property. Examples of those algebras are given including the algebras of continuous functions on compact sets.

1. Introduction

Let 𝑋 be a Banach space. A mapping 𝑇 𝐸 𝑋 𝑋 is nonexpansive if

𝑇 𝑥 𝑇 𝑦 𝑥 𝑦 ( 1 . 1 ) for each 𝑥 , 𝑦 𝐸 . The fixed point set of 𝑇 is F i x ( 𝑇 ) = { 𝑥 𝐸 𝑇 𝑥 = 𝑥 } . We say that the space 𝑋 has the fixed point property (or weak fixed point property) if for every nonempty bounded closed convex (or weakly compact convex, resp.) subset 𝐸 of 𝑋 and every nonexpansive mapping 𝑇 𝐸 𝐸 we have F i x ( 𝑇 ) . One of the central goals in fixed point theory is to solve the problem: which Banach spaces have the (weak) fixed point property?

For weak fixed point property, Alspach [1] exhibited a weakly compact convex subset 𝐸 of the Lebesgue space 𝐿 1 [ 0 , 1 ] and an isometry 𝑇 𝐸 𝐸 without a fixed point, proving that the space 𝐿 1 [ 0 , 1 ] does not have the weak fixed point property. Lau et al. [2] proved the following results.

Theorem 1.1. Let 𝑋 be a locally compact Hausdorff space. If 𝐶 0 ( 𝑋 ) has the weak fixed point property, then X is dispersed.

Corollary. Let 𝐺 be a locally compact group. Then the 𝐶 -algebra 𝐶 0 ( 𝐺 ) has the weak fixed point property if and only if 𝐺 is discrete.

Corollary. A von Neumann algebra has the weak fixed point property if and only if is finite dimensional.

Continuing in this direction, Benavides and Pineda [3] developed the concept of 𝜔 -almost weak orthogonality in the Banach lattice 𝐶 ( 𝐾 ) and obtained the results.

Theorem. Let 𝑋 be a 𝜔 -almost weakly orthogonal closed subspace of 𝐶 ( 𝐾 ) where 𝐾 is a metrizable compact space. Then 𝑋 has the weak fixed point property.

Theorem. Let 𝐾 be a metrizable compact space. Then, the following conditions are all equivalent: ( 1 ) 𝐶 ( 𝐾 ) is 𝜔 -almost weakly orthogonal,( 2 ) 𝐶 ( 𝐾 ) is 𝜔 -weakly orthogonal,( 3 ) 𝐾 ( 𝜔 ) = .

Corollary. Let 𝐾 be a compact set with 𝐾 ( 𝜔 ) = . Then 𝐶 ( 𝐾 ) has the weak fixed point property.

As for the fixed point property, Dhompongsa et al. [4] showed that a 𝐶 -algebra has the fixed point property if and only if it is finite dimensional. In this paper, we approach the question on the fixed point property from the opposite direction by identifying unital abelian Banach algebras which fail to have the fixed point property. As consequences, we obtain results on the algebra of continuous functions 𝐶 ( 𝑆 ) , where 𝑆 is a compact set, and there is a unital abelian subalgebra of the algebra 𝑙 ( ) which does not have the fixed point property and does not contain the space 𝑐 0 .

2. Preliminaries and Lemmas

The fields of real and complex numbers are denoted by and , respectively. The symbol 𝔽 denotes a field that can be either or . The elements of 𝔽 are called scalars.

An element 𝑥 in a unital algebra 𝑋 is said to be invertible if there is an element 𝑦 in 𝑋 such that

𝑥 𝑦 = 𝑦 𝑥 = 1 . ( 2 . 1 ) In this case 𝑦 is unique and written 𝑥 1 .

We define the spectrum of an element 𝑥 of a unital algebra 𝑋 over 𝔽 to be the set

𝜎 ( 𝑥 ) = { 𝜆 𝔽 𝜆 1 𝑥 i s n o t i n v e r t i b l e } . ( 2 . 2 ) The spectral radius of 𝑥 is defined to be

𝑟 ( 𝑥 ) = s u p 𝜆 𝜎 ( 𝑥 ) | | 𝜆 | | . ( 2 . 3 )

We note that a subalgbra of a normed algebra is itself a normed algebra with the norm got by restriction. The closure of a subalgebra is a subalgebra. A closed subalgebra of a Banach algebra is a Banach algebra. If ( 𝐵 𝛼 ) 𝛼 Λ is a family of subalgebras of an algebra 𝑋 , then 𝛼 Λ 𝐵 𝛼 is a subalgebra also. Hence, for any subset 𝐸 of 𝑋 , there is the smallest subalgebra 𝐴 ( 𝐸 ) of 𝑋 containing 𝐸 . This algebra is called the subalgebra of 𝑋 generated by 𝐸 . If 𝐸 is the singleton { 𝑥 } , then 𝐴 ( 𝐸 ) is the linear span of all powers 𝑥 𝑛 of 𝑥 . If 𝑋 is a normed algebra, the closed algebra 𝐵 ( 𝐸 ) generated by a set 𝐸 is the smallest closed subalgebra containing 𝐸 . We can see that 𝐵 ( 𝐸 ) = 𝐴 ( 𝐸 ) .

We denote by 𝐶 𝔽 ( 𝑆 ) the Banach algebra of continuous functions from a topological space 𝑆 to 𝔽 , with the sup-norm

𝑓 = s u p 𝑥 𝑆 | | | | . 𝑓 ( 𝑥 ) ( 2 . 4 )

The following theorems are known as the Stone-Weierstrass approximation theorem for 𝐶 ( 𝑆 ) and 𝐶 ( 𝑆 ) , respectively. For the details, the readers are referred to [5].

Theorem. Let 𝐴 be a subalgebra of 𝐶 ( 𝑆 ) such that ( 1 ) 𝐴 separates the points of 𝑆 , ( 2 ) 𝐴 annihilates no point of 𝑆 . Then 𝐴 is dense in 𝐶 ( 𝑆 ) .

Theorem. Let 𝑆 be a compact space, 𝐴 a subalgebra of 𝐶 ( 𝑆 ) such that ( 1 ) 𝐴 separates the points of 𝑆 , ( 2 ) 𝐴 annihilates no point of 𝑆 , ( 3 ) 𝑓 𝐴 implies that the conjugate 𝑓 of 𝑓 is in 𝐴 . Then 𝐴 is dense in 𝐶 ( 𝑆 ) .

A character on a unital algebra 𝑋 over 𝔽 is a nonzero homomorphism 𝜏 𝑋 𝔽 . We denote by Ω ( 𝑋 ) the set of characters on 𝑋 . Note that if 𝑋 is a unital abelian complex Banach algebra, then

𝜎 ( 𝑥 ) = { 𝜏 ( 𝑥 ) 𝜏 Ω ( 𝑋 ) } ( 2 . 5 ) for each 𝑥 𝑋 (see [6]).

Remark. It is unknown if (2.5) is valid whenever Ω ( 𝑋 ) . Equation (2.5) obviously does not hold for a space 𝑋 with Ω ( 𝑋 ) = as the following example shows.

Example 2.4. Let 𝑋 = be considered as a real unital abelian Banach algebra under ordinary complex multiplication and whose norm is the absolute value. We have Ω ( 𝑋 ) = . Indeed, assume to the contrary that there is a non-zero homomorphism 𝜏 0 on 𝑋 and 𝜏 0 ( 𝑖 ) = 𝜆 , so 𝜏 0 ( ( 1 + 𝑖 ) 𝑖 ) = 𝜏 0 𝜏 ( 𝑖 1 ) = 𝜆 1 , 0 ( 1 + 𝑖 ) 𝜏 0 ( 𝑖 ) = ( 1 + 𝜆 ) 𝜆 = 𝜆 + 𝜆 2 . ( 2 . 6 ) Thus 𝜆 1 = 𝜆 + 𝜆 2 ; so 𝜆 is not a real number, which is a contradiction.
Since Ω ( 𝑋 ) = , so { 𝜏 ( 1 ) 𝜏 Ω ( 𝑋 ) } = but 𝜎 ( 1 ) = { 1 } .

We consider throughout this paper on Banach algebras 𝑋 for which Ω ( 𝑋 ) and satisfy (2.5).

If 𝑋 is a unital abelian Banach algebra, it follows from Proposition 2.5 that Ω ( 𝑋 ) is contained in the closed unit ball of 𝑋 . We endow Ω ( 𝑋 ) with the relative weak* topology and call the topological space Ω ( 𝑋 ) the character space of 𝑋 .

Detailed proofs of the following propositions can be found in [6].

Proposition. Let 𝑋 be a unital abelian Banach algebra. If 𝜏 Ω ( 𝑋 ) , then 𝜏 = 1 .

Proposition. If 𝑋 is a unital Banach algebra, then Ω ( 𝑋 ) is compact.

If 𝑋 is a unital abelian Banach algebra, and 𝑥 𝑋 , we define a continuous function ̂ 𝑥 by

̂ 𝑥 Ω ( 𝑋 ) 𝔽 , 𝜏 𝜏 ( 𝑥 ) . ( 2 . 7 ) We call ̂ 𝑥 the Gelfand transform of 𝑥 , and the homomorphism

𝜑 𝑋 𝐶 𝔽 ( Ω ( 𝑋 ) ) , 𝑥 ̂ 𝑥 ( 2 . 8 ) is called the Gelfand representation.

The following two lemmas, Lemmas 2.7 and 2.10, will be used to prove our main theorem.

Lemma. Let 𝑋 be a unital abelian real Banach algebra with i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } > 0 . ( 2 . 9 ) Then one has the following: (i)the Gelfand representation 𝜑 is a bounded isomorphism,(ii)the inverse 𝜑 1 is also a bounded isomorphism.

Proof. (i) 𝜑 is injective since i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } > 0 implies k e r ( 𝜑 ) = 0 . It is easily checked that 𝜑 is a bounded homomorphism, and 𝜑 ( 𝑋 ) is a subalgebra of 𝐶 ( Ω ( 𝑋 ) ) separating the points of Ω ( 𝑋 ) , and having the property that for any 𝜏 Ω ( 𝑋 ) there is an element 𝑥 𝑋 such that ̂ 𝑥 ( 𝜏 ) 0 . In order to use the Stone-Weierstrass theorem to show that 𝜑 ( 𝑋 ) = 𝐶 ( Ω ( 𝑋 ) ) , we shall show that 𝜑 ( 𝑋 ) is closed. We show that 𝜑 ( 𝑋 ) is closed by showing that 𝜑 ( 𝑋 ) is complete. Let { 𝑥 𝑛 } be a Cauchy sequence in 𝜑 ( 𝑋 ) . First, we show that the sequence { 𝑥 𝑛 } = { 𝜑 1 ( 𝑥 𝑛 ) } is Cauchy. Assume on the contrary that { 𝑥 𝑛 } is not Cauchy. Thus there exists 𝜀 0 > 0 and subsequences { 𝑧 𝑛 } and { 𝑧 𝑛 } of { 𝑥 𝑛 } such that 𝑧 𝑛 𝑧 𝑛 𝜀 0 ( 2 . 1 0 ) for each 𝑛 . Write 𝑦 𝑛 = ( 𝑧 𝑛 𝑧 𝑛 ) / 𝜀 0 , then 𝑦 𝑛 1 , for each 𝑛 . But { 𝑥 𝑛 } is Cauchy, and so we have 𝑦 𝑛 0 . Thus 0 < i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } i n f 𝑛 𝑟 𝑦 𝑛 𝑦 𝑛 = i n f 𝑛 𝑦 𝑛 𝑦 𝑛 = 0 , ( 2 . 1 1 ) which is a contradiction. Hence { 𝑥 𝑛 } must be Cauchy and so 𝑥 𝑛 𝑥 0 , for some 𝑥 0 𝑋 . Since ̂ 𝑥 = 𝜑 ( 𝑥 ) 𝑥 , for each 𝑥 𝑋 , so 𝑥 𝑛 𝑥 0 . Thus 𝜑 ( 𝑋 ) is complete. The Stone-Weierstrass theorem can be applied to conclude that 𝜑 is surjective.
(ii) follows from the open mapping theorem.

Remark 2.8. (i) Lemma 2.7 tells us that if 𝑋 is a unital abelian real Banach algebra with property i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } > 0 , ( 2 . 1 2 ) then 𝑋 and 𝐶 ( Ω ( 𝑋 ) ) are homeomorphic and isomorphic under 𝜑 . Hence if we would like to consider the convergence of a sequence { 𝑥 𝑛 } in 𝑋 , we could look at the convergence of the corresponding sequence { 𝑥 𝑛 } .
(ii) Property (2.12) clearly implies the semisimplicity property ( 𝑟 ( 𝑥 ) 𝑥 = 0 ) but the following example shows that it is stronger.

Example. Let 𝑙 1 ( ) denote the Banach algebra of complex-valued absolutely summable functions on the group of integers under convolution regarded as a real Banach algebra and let 𝑋 be the real subalgebra of 𝑙 1 ( ) consisting of those functions that satisfy 𝑓 ( 𝑛 ) = 𝑓 ( 𝑛 ) , 𝑛 . Then the maximal ideal space of 𝑋 equals 𝑇 = 𝑅 / 𝑍 and the Gelfand transform is precisely the Fourier transform which maps 𝑋 into the real Banach algebra 𝐶 ( 𝑇 ) of continuous real-valued functions on 𝐶 ( 𝑇 ) under pointwise multiplication and maximum norm. Although the image of the Fourier transform is dense, it is clearly not all of 𝐶 ( 𝑇 ) since it is simply the real-valued functions in the Wiener space which consists of complex-valued functions whose Fourier series are absolutely summable. Therefore Lemma 2.7 shows that 𝑋 does not have Property (2.12).

Lemma. Let 𝑋 be an infinite dimensional unital abelian real Banach algebra with i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } > 0 . ( 2 . 1 3 ) Then one has the following: (i) Ω ( 𝑋 ) is an infinite set, (ii)if there exists a bounded sequence { 𝑥 𝑛 } in 𝑋 which contains no convergent subsequences and such that { 𝜏 ( 𝑥 𝑛 ) 𝜏 Ω ( 𝑋 ) } is finite for each 𝑛 , then there is an element 𝑥 0 𝑋 with 𝜏 𝑥 0 = 1 𝜏 Ω ( 𝑋 ) 1 , 2 , 2 3 , 3 4 , , ( 2 . 1 4 ) (iii)there is an element 𝑥 0 𝑋 such that { 𝜏 ( 𝑥 0 ) 𝜏 Ω ( 𝑋 ) } is an infinite set,(iv) there exists a sequence { 𝑥 𝑛 } in 𝑋 such that { 𝜏 ( 𝑥 𝑛 ) 𝜏 Ω ( 𝑋 ) } [ 0 , 1 ] , for each 𝑛 , and { ( 𝑥 𝑛 ) 1 { 1 } } is a sequence of nonempty pairwise disjoint subsets of Ω ( 𝑋 ) .

Proof. Let 𝑋 be an infinite dimensional unital abelian real Banach algebra with i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } > 0 . ( 2 . 1 5 ) (i)If suffices to show that if Ω ( 𝑋 ) is a finite set, for then the closed unit ball 𝐵 𝑋 of 𝑋 is compact, and this will lead to us a contradiction. Let Ω ( 𝑋 ) be a finite set, say { 𝜏 1 , 𝜏 2 , , 𝜏 𝑚 } , and let { 𝑥 𝑛 } be a sequence in 𝐵 𝑋 .
Since the sequences { 𝜏 𝑝 ( 𝑥 𝑛 ) } , 𝑝 = 1 , 2 , , 𝑚 , are bounded, we can choose a subsequence { 𝑥 𝑛 𝑚 } of { 𝑥 𝑛 } such that 𝜏 𝑝 ( 𝑥 𝑛 𝑚 ) 𝜆 𝑝 , for each 𝑝 = 1 , 2 , , 𝑚 .
Define 𝜓 Ω ( 𝑋 ) by 𝜓 ( 𝜏 𝑝 ) = 𝜆 𝑝 . Thus there exists 𝑥 𝑋 such that 𝜓 = ̂ 𝑥 , and consequently, 𝑥 𝑛 𝑚 ̂ 𝑥 since
𝑥 𝑛 𝑚 ̂ 𝑥 = s u p 𝜏 Ω ( 𝑋 ) | | 𝑥 𝑛 𝑚 | | ( 𝜏 ) ̂ 𝑥 ( 𝜏 ) = m a x 1 𝑝 𝑚 | | 𝑥 𝑛 𝑚 𝜏 𝑝 𝜏 ̂ 𝑥 𝑝 | | , ( 2 . 1 6 ) and 𝜏 𝑝 ( 𝑥 𝑛 𝑚 ) ̂ 𝑥 ( 𝜏 𝑝 ) , for each 𝑝 = 1 , 2 , , 𝑚 .
So { 𝑥 𝑛 𝑚 } is a subsequence of { 𝑥 𝑛 } such that 𝑥 𝑛 𝑚 ̂ 𝑥 . By Remark 2.8, 𝑥 𝑛 𝑚 𝑥 , where 𝑥 = 𝜑 1 ( ̂ 𝑥 ) . Thus 𝐵 𝑋 is compact.
(ii) Let { 𝑥 𝑛 } be a bounded sequence in 𝑋 which has no convergent subsequences and suppose that the set { 𝜏 ( 𝑥 𝑛 ) 𝜏 Ω ( 𝑋 ) } is finite for each 𝑛 . By Remark 2.8, we will consider { 𝑥 𝑛 } as a sequence of Gelfand transforms { 𝑥 𝑛 } .
First, we show that we can write
Ω ( 𝑋 ) = 𝑛 𝐺 𝑛 𝐹 , ( 2 . 1 7 ) where 𝐹 is closed, 𝐺 𝑛 are all closed and open, and { 𝐹 , 𝐺 1 , 𝐺 2 , } is a partition of Ω ( 𝑋 ) . For each 𝑛 , write 𝜏 𝑥 𝑛 = 𝜆 𝜏 Ω ( 𝑋 ) ( 𝑛 , 𝑖 ) 𝑖 = 1 , 2 , , 𝑚 𝑛 , 𝑛 = 𝑥 𝑛 1 𝜆 ( 𝑛 , 𝑖 ) 𝑖 = 1 , 2 , , 𝑚 𝑛 . ( 2 . 1 8 ) Define = 𝑘 𝐴 𝑘 𝐴 𝑘 𝑘 { } . ( 2 . 1 9 ) Note that ( 𝑥 𝑛 ) 1 { 𝜆 ( 𝑛 , 𝑖 ) } are all closed and open. Since 𝑛 is a partition of Ω ( 𝑋 ) for each 𝑛 , is a partition of Ω ( 𝑋 ) . There are two cases to be considered. Case 1 ( is infinite). Thus there exists 𝑖 1 such that 𝑥 1 1 𝜆 ( 1 , 𝑖 1 ) 𝑘 𝐴 𝑘 𝐴 𝑘 𝑘 , 𝑘 2 ( 2 . 2 0 ) is an infinite set. Similarly, there exists 𝑖 2 such that 𝑥 1 1 𝜆 ( 1 , 𝑖 1 ) 𝑥 2 1 𝜆 ( 2 , 𝑖 2 ) 𝑘 𝐴 𝑘 𝐴 𝑘 𝑘 , 𝑘 3 ( 2 . 2 1 ) is an infinite set. Continuing in this process we obtain a sequence of the sets ( 𝑥 𝑛 ) 1 { 𝜆 ( 𝑛 , 𝑖 𝑛 ) } 𝑛 such that 𝑛 𝑗 = 1 𝑥 𝑗 1 𝜆 ( 𝑗 , 𝑖 𝑗 ) 𝑘 𝐴 𝑘 𝐴 𝑘 𝑘 , 𝑘 𝑛 + 1 ( 2 . 2 2 ) is an infinite set, for each 𝑛 .
Write
𝐻 1 = 𝑖 𝑖 1 𝑥 1 1 𝜆 ( 1 , 𝑖 ) , 𝐻 2 = 𝑥 1 1 𝜆 ( 1 , 𝑖 1 ) 𝑖 𝑖 2 𝑥 2 1 𝜆 ( 2 , 𝑖 ) , 𝐻 3 = 𝑥 1 1 𝜆 ( 1 , 𝑖 1 ) 𝑥 2 1 𝜆 ( 2 , 𝑖 2 ) 𝑖 𝑖 3 𝑥 3 1 𝜆 ( 3 , 𝑖 ) , . ( 2 . 2 3 ) Thus 𝐻 𝑛 are all closed and open, and Ω ( 𝑋 ) = 𝑛 𝐻 𝑛 Ω ( 𝑋 ) 𝑛 𝐻 𝑛 , ( 2 . 2 4 ) where Ω ( 𝑋 ) 𝑛 𝐻 𝑛 is a nonempty closed set since Ω ( 𝑋 ) is compact. And since has infinite elements, we can see that there exists a subsequence { 𝐺 𝑛 } of { 𝐻 𝑛 } such that 𝑛 𝐺 𝑛 = 𝑛 𝐻 𝑛 and 𝐺 𝑛 , for each 𝑛 .
Hence we have
Ω ( 𝑋 ) = 𝑛 𝐺 𝑛 Ω ( 𝑋 ) 𝑛 𝐺 𝑛 , ( 2 . 2 5 ) and { ( Ω ( 𝑋 ) 𝑛 𝐺 𝑛 ) , 𝐺 1 , 𝐺 2 , } is a partition of Ω ( 𝑋 ) .
Case 2 ( = { 𝐿 𝑖 𝑖 = 1 , 2 , , 𝑚 } ). To show that this case leads to a contradiction, we first observe that if 𝜏 , 𝜏 are in the same 𝐿 𝑖 , then 𝜏 𝑥 𝑛 𝑥 = 𝜏 𝑛 ( 2 . 2 6 ) for each 𝑛 . Write 𝛼 ( 𝑛 , 𝑖 ) = 𝜏 ( 𝑥 𝑛 ) , if 𝜏 𝐿 𝑖 . There exists a subsequence { ( 𝛼 ( 𝑛 , 1 ) , 𝛼 ( 𝑛 , 2 ) , , 𝛼 ( 𝑛 , 𝑚 ) ) } 𝑛 of { ( 𝛼 ( 𝑛 , 1 ) , 𝛼 ( 𝑛 , 2 ) , , 𝛼 ( 𝑛 , 𝑚 ) ) } 𝑛 such that for each 𝑖 = 1 , 2 , , 𝑚 , 𝛼 ( 𝑛 , 𝑖 ) 𝛼 𝑖 ( 2 . 2 7 ) for some ( 𝛼 1 , 𝛼 2 , , 𝛼 𝑚 ) 𝑚 . Define a Gelfand transform ̂ 𝑥 Ω ( 𝑋 ) by ̂ 𝑥 ( 𝜏 ) = 𝛼 𝑖 , if 𝜏 𝐿 𝑖 . Since 𝑥 𝑛 ̂ 𝑥 = s u p 𝜏 Ω ( 𝑋 ) | | 𝑥 𝑛 | | ( 𝜏 ) ̂ 𝑥 ( 𝜏 ) = m a x 1 𝑖 𝑚 | | 𝛼 ( 𝑛 , 𝑖 ) 𝛼 𝑖 | | , ( 2 . 2 8 ) so 𝑥 𝑛 ̂ 𝑥 , which is a contradiction.
Now we conclude that
Ω ( 𝑋 ) = 𝑛 𝐺 𝑛 𝐹 , ( 2 . 2 9 ) where 𝐹 is closed, 𝐺 𝑛 is closed and open for each 𝑛 , and { 𝐹 , 𝐺 1 , 𝐺 2 , } is a partition of Ω ( 𝑋 ) . Define a map 𝜓 Ω ( 𝑋 ) by 𝑛 𝜓 ( 𝜏 ) = 𝑛 + 1 i f 𝜏 𝐺 𝑛 , 1 i f 𝜏 𝐹 . ( 2 . 3 0 ) We can check that the inverse image of each closed set in 𝜓 ( Ω ( 𝑋 ) ) is closed. Therefore, 𝜑 1 ( 𝜓 ) is an element in 𝑋 , say 𝑥 0 , with 𝜏 𝑥 0 = 1 𝜏 Ω ( 𝑋 ) 1 , 2 , 2 3 , 3 4 . , ( 2 . 3 1 )

(iii) Assume to the contrary that { 𝜏 ( 𝑥 ) 𝜏 Ω ( 𝑋 ) } is finite for each 𝑥 𝑋 . Since 𝑋 is infinite dimensional, so, as 𝐵 𝑋 is noncompact, there exists a bounded sequence { 𝑥 𝑛 } in 𝑋 which has no convergent subsequences. Hence { 𝜏 ( 𝑥 𝑛 ) 𝜏 Ω ( 𝑋 ) } is finite for each 𝑛 . It follows from (ii) that
𝜏 𝑥 0 = 1 𝜏 Ω ( 𝑋 ) 1 , 2 , 2 3 , 3 4 , ( 2 . 3 2 ) for some 𝑥 0 𝑋 , which is the contradiction.
(iv) From (iii), there is an element 𝑥 1 𝑋 such that { 𝜏 ( 𝑥 1 ) 𝜏 Ω ( 𝑋 ) } is an infinite set. We can choose 𝑥 1 so that there exists a strictly decreasing sequence { 𝑎 𝑛 } such that
𝑎 𝑛 𝑥 1 [ ] ( Ω ( 𝑋 ) ) 0 , 1 , 𝑎 1 < 1 , ( 2 . 3 3 ) and 𝜏 ( 𝑥 1 ) = 1 for some 𝜏 Ω ( 𝑋 ) . Define a continuous function 𝑔 1 [ 0 , 1 ] [ 0 , 1 ] to be linear on [ 0 , 𝑎 1 ] and on [ 𝑎 1 , 1 ] joining the points ( 0 , 0 ) and ( 𝑎 1 , 1 ) , and 𝑔 ( 1 ) ( 𝑔 1 ( 𝑎 2 ) , 1 ) . Put 𝑥 2 = 𝑔 1 𝑥 1 , for some 𝑥 2 𝑋 , and define a continuous function 𝑔 2 [ 0 , 1 ] [ 0 , 1 ] similar to the way we construct 𝑔 1 . The left part of 𝑔 2 is the line joining the point ( 0 , 0 ) and ( 𝑔 1 ( 𝑎 2 ) , 1 ) and 𝑔 2 ( 1 ) ( 𝑔 2 ( 𝑔 1 ( 𝑎 2 ) ) , 1 ) . Then put 𝑥 3 = 𝑔 2 𝑥 2 , for some 𝑥 3 𝑋 . Continuing in this process we obtain a sequence of points { 𝑥 𝑛 } such that { 𝜏 ( 𝑥 𝑛 ) 𝜏 Ω ( 𝑋 ) } [ 0 , 1 ] , for each 𝑛 , and { ( 𝑥 𝑛 ) 1 { 1 } } is a sequence of nonempty pairwise disjoint subsets of Ω ( 𝑋 ) . We then obtain the required result.

3. Main Theorem

Now we prove our main theorem.

Theorem. Let 𝑋 be an infinite dimensional unital abelian real Banach algebra satisfying each of the following: (i)if 𝑥 , 𝑦 𝑋 is such that | 𝜏 ( 𝑥 ) | | 𝜏 ( 𝑦 ) | , for each 𝜏 Ω ( 𝑋 ) , then 𝑥 𝑦 , (ii) i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } > 0 . Then 𝑋 does not have fixed point property.

Proof. Let 𝑋 be an infinite dimensional unital abelian real Banach algebra satisfying (i) and (ii). Assume to the contrary that 𝑋 has fixed point property. From Lemma 2.10(iv), there exists a sequence { 𝑥 𝑛 } in 𝑋 such that 𝜏 𝑥 𝑛 [ ] 𝜏 Ω ( 𝑋 ) 0 , 1 ( 3 . 1 ) for each 𝑛 , and { ( 𝑥 𝑛 ) 1 { 1 } } is a sequence of nonempty pairwise disjoint subsets of Ω ( 𝑋 ) . Write 𝐴 𝑛 = ( 𝑥 𝑛 ) 1 { 1 } , and define 𝑇 𝑛 𝐸 𝑛 𝐸 𝑛 by 𝑥 𝑥 𝑛 𝑥 , ( 3 . 2 ) where 𝐸 𝑛 = 𝑥 𝑋 0 𝜏 ( 𝑥 ) 1 f o r e a c h 𝜏 Ω ( 𝑋 ) , a n d 𝜏 ( 𝑥 ) = 1 i f 𝜏 𝐴 𝑛 . ( 3 . 3 ) It follows from (i) that 𝑇 𝑛 𝐸 𝑛 𝐸 𝑛 is a nonexpansive mapping on the bounded closed convex set 𝐸 𝑛 , for each 𝑛 . Indeed, 𝐸 𝑛 is bounded since 𝑥 0 < i n f { 𝑟 ( 𝑥 ) 𝑥 𝑋 , 𝑥 = 1 } 𝑟 𝑥 = s u p 𝜏 Ω ( 𝑋 ) | | | | 𝜏 𝑥 | | | | = 1 𝑥 𝑥 s u p 𝜏 Ω ( 𝑋 ) | | 𝜏 | | ( 𝑥 ) ( 3 . 4 ) for each 𝑥 𝑋 . It follows that 𝑇 𝑛 has a fixed point, say 𝑦 𝑛 , for each 𝑛 . Since 𝑦 𝑛 = 𝑥 𝑛 𝑦 𝑛 , thus 𝑦 𝑛 = 𝑥 𝑛 𝑦 𝑛 , and then 𝑦 𝑛 ( 𝜏 ) = 0 i f 𝜏 i s n o t i n 𝐴 𝑛 , 1 i f 𝜏 i s i n 𝐴 𝑛 , ( 3 . 5 ) for each 𝑛 . Since 𝐴 1 , 𝐴 2 , 𝐴 3 , are pairwise disjoint, so 𝑦 𝑚 𝑦 𝑛 = 1 , if 𝑚 𝑛 . Hence { 𝑦 𝑛 } has no convergent subsequences. From Lemma 2.7, { 𝑦 𝑛 } has no convergent subsequences too. It follows from the existence of { 𝑦 𝑛 } and Lemma 2.10 (ii) that there exists an element 𝑥 0 in 𝑋 with 𝜏 𝑥 0 = 1 𝜏 Ω ( 𝑋 ) 1 , 2 , 2 3 , 3 4 . , ( 3 . 6 ) Write 𝐴 0 = ( 𝑥 0 ) 1 { 1 } . Define 𝑇 0 𝐸 0 𝐸 0 by 𝑥 𝑥 0 𝑥 , ( 3 . 7 ) where 𝐸 0 = 𝑥 𝑋 0 𝜏 ( 𝑥 ) 1 f o r e a c h 𝜏 Ω ( 𝑋 ) , a n d 𝜏 ( 𝑥 ) = 1 i f 𝜏 𝐴 0 . ( 3 . 8 ) By (i), 𝑇 0 is a nonexpansive mapping on the bounded closed convex set 𝐸 0 . Thus 𝑇 0 has a fixed point, say 𝑦 0 , that is, 𝑦 0 = 𝑥 0 𝑦 0 . Thus 𝑦 0 = 𝑥 0 𝑦 0 . Consequently, 𝑦 0 ( 𝜏 ) = 0 , i f 𝜏 i s n o t i n 𝑥 0 1 { 1 } , 1 , i f 𝜏 i s i n 𝑥 0 1 { 1 } . ( 3 . 9 ) The set ( 𝑥 0 ) 1 { 1 } = ( 𝑦 0 ) 1 { 1 } is open in Ω ( 𝑋 ) , since 𝑦 0 is continuous. Also the set ( 𝑥 0 ) 1 { 𝑛 / ( 𝑛 + 1 ) } is open in Ω ( 𝑋 ) for each 𝑛 , since 𝑥 0 is continuous. Thus, 𝑥 0 1 𝑛 𝑛 + 1 𝑛 𝑥 0 1 { 1 } ( 3 . 1 0 ) is an open covering of Ω ( 𝑋 ) . This leads to a contradiction, since Ω ( 𝑋 ) is compact.

From the above theorem we have the following.

Corollary 3.2. Let 𝑆 be a compact Hausdorff topological space. If 𝐶 ( 𝑆 ) is infinite dimensional, then 𝐶 ( 𝑆 ) fails to have the fixed point property.

Proof. 𝐶 ( 𝑆 ) satisfies (i), (ii) in Theorem 3.1. Indeed, if 𝑥 , 𝑦 𝐶 ( 𝑆 ) is such that | 𝜏 ( 𝑥 ) | | 𝜏 ( 𝑦 ) | , for each 𝜏 Ω ( 𝐶 ( 𝑆 ) ) , then | 𝑥 ( 𝑠 ) | | 𝑦 ( 𝑠 ) | , for each 𝑠 𝑆 . Hence 𝑥 𝑦 . And since 𝑟 ( 𝑥 ) = s u p 𝜆 𝜎 ( 𝑥 ) | | 𝜆 | | = s u p 𝑠 𝑆 | | | | 𝑥 ( 𝑠 ) = 𝑥 , ( 3 . 1 1 ) so i n f { 𝑟 ( 𝑥