676
Views
0
CrossRef citations to date
0
Altmetric
APPLIED & INTERDISCIPLINARY MATHEMATICS

Lipschitz isomorphism and fixed point theorem for normed groups

ORCID Icon, ORCID Icon & ORCID Icon | (Reviewing editor)
Article: 1859673 | Received 20 Sep 2020, Accepted 01 Dec 2020, Published online: 12 Jan 2021

Abstract

This paper aims to propose normed structures for groups and to establish the Lipschitz mapping of a normed group G to itself. We also investigate some conjugate and isomorphic Lipschitz mappings to determine the equivalent norm and inverse Lipschitz mappings. Specifically, in the main result, we present a fixed point theorem for self-mappings satisfying certain contraction principles on a complete normed group.

PUBLIC INTEREST STATEMENT

In mathematical analysis, Lipschitz continuity (named after Rudolf Lipschitz) is the Picard–Lindelöf theorem’s central condition. A particular type of Lipschitz continuity, called contraction, is used in the Banach fixed-point theorem (known as the contraction mapping theorem) is an essential tool that ensures the existence and uniqueness of fixed points of certain self-mappings and provides a constructive method to find those fixed points. From previous studies, it has been shown that Banach fixed-point theorem is a useful mechanism for numerous branches of mathematical analysis, general topology, functional analysis, and economics. One of the most known applications of Banach’s fixed point theorem for economists is Bellman’s functional equations.

This article mainly focused on understanding the general properties of Lipschitz self-mappings and its connection with the existence of isomorphic group-norm. We have investigated Banach fixed point theorem for normed groups to determine the existence of unique fixed points following specific conditions.

1. Introduction and preliminaries

Normed groups are just groups associated with a right-invariant metric. The right-invariant metric performs an important role in the construction of group-norm. The fundamental metrization theorem for groups presented by Birkhoff (Birkhoff, Citation1936) and Kakutani (Kakutani, Citation1936) is usually stated that a first-countable Hausdorff group has a right-invariant metric. Klee (Klee, Citation1952) introduced invariant metrics in groups, which provided a different way to define a group-norm associated with the invariant metrics.

When G is an abelian group and also topologically complete, then it admits a metric that is both right and left-invariant (bi-invariant). See, for instance, (Bingham & Ostaszewski, Citation2010; Klee, Citation1952). Bingham et al. (Bingham & Ostaszewski, Citation2010) also made contributions for the equivalent representation of a group-norm as a right (dR) and left (dL) invariant metric. The reader may also consult (Batagelj, Citation1995; Bökamp, Citation1994; Nourouzi & Pourmoslemi, Citation2017; Sarfraz & Li, Citation2019; Zorzitto, Citation1985) to understand the notion of groupnorm and its association with right and left-invariant metric. This metric enhances the group structure, and accordingly, the group becomes a metric space with a norm-like function.

The subsequent proposition is well recognized (see (Bingham & Ostaszewski, Citation2010)).

Proposition 1.1 If is a group-norm, then d(g,h):=∥gh1 is a right-invariant metric; equivalently, d1(g1,h1):=∥g1h is the conjugate left-invariant metric on the group. Conversely, if d is a right-invariant metric, then g∥:=d(e,g)=d1(e,g) is a group-norm.

We say that group G is normed if it has a group-norm as defined below.

Definition 1.2 (Batagelj, Citation1995; Bingham & Ostaszewski, Citation2010; Sarfraz & Li, Citation2019) Let (G,e,) be a group under binary operation with identity element e, then its norm ∥:G[0,+) is said to be a group-norm if the subsequent properties hold for every g1,g2G,

(1) g1g2∥≤∥g1+g2;

(2) g1∥≥0, with g1∥=0 if and only if g1=e;

(3) g11∥=∥g1.

If property (1) holds, then it is said to be a semi-norm; if (1) and (2) satisfies with e∥=0 then one speaks of a pseudo-norm; if properties (1) and (2) holds, it is known as a pre-norm. A normed group having group-norm and group (G,e,) with identity element e is denoted by (G,,e,). Moreover, if G is complete for group-norm , then we say that (G,) is a complete normed group.

It follows from the following theorem that the cardinal norm of a generated group G (generated by set S) induces a metric given by dc(g1,g2)=∥g11g2 for every g1,g2G, called the cardinal metric, and so G becomes a metric space. See, for instance, (Suksumran, Citation2019).

Theorem 1.3. (see (Suksumran, Citation2019), Theorem 2.1) Assume that (G,S) be a generated group. The cardinal norm induced by S is a group-norm; that is, it satisfies the group-norm properties exhibited in Definition 1.2.

Fixed point theory serves as an essential mechanism for numerous branches of mathematical analysis, general topology, functional analysis, and its applications. Loosely speaking, there are three main approaches in this theory: the topological, the metric, and the order-theoretic approach, where illustrative examples of these are Brouwer’s, Banach’s (Banach, Citation1922) and Tarski’s theorems, respectively.

The Banach fixed point theorem is also recognized as contraction mapping theorem, and it is a very useful to show the existence of a unique solution to functional and differential equations. One of the most known applications of Banach’s fixed point theorem for economists is Bellman’s functional equations. For comprehensive analyses and studies concerning Banach contraction principle, readers see (Jain et al., Citation2012; CitationJleli and Samet ((2014)); Shatanawi & Nashine, Citation2012).

In this paper, the fixed point theorem plays a significant role to determine the fixed points of η in Banach group G. This study’s main purpose is to investigate the existence of a fixed point for self-mappings η defined on a complete normed group (Banach group) G that satisfies certain conditions. We recommend the readers to see (Debnath, Citation2020; Debnath et al., Citation2016, Citation2020; Debnath & Sen, Citation2020; Debnath & Srivastava, Citation2020, Citation2020; Merryfield et al., Citation2002; Neog et al., Citation2019; Pata, Citation2011) and the references cited therein to obtain comprehensive results relating to the fixed points. Fixed point theorems concern maps η of a set G into itself that, under certain conditions, admit a fixed point gG such that η(g)=g.

Apart from this introduction, the paper is arranged into two sections. In Section 2, we determine the Lipschitz mapping η of a normed group G (may or may not be complete) to itself and also introduce some isomorphic and continuous inverse Lipschitz mappings. This section also describes an equivalent norm and the construction of a conjugate Lipschitz map from a normed group G to itself.

Section 3 deals with employing the fixed point theorem to determine the unique fixed points of η from a Banach group G to itself. Moreover, we also find the fixed points of η from a normed cyclic group G to itself whenever η is an expansive mapping.

In our considerations the following definition will play an important role.

Definition 1.4. (Debnath & Sen, Citation2020; Hussain et al., Citation2015) Assume that G be a normed group defined over a norm . A mapping η:GG is said to be a Lipschitz continuous if there exists γ0 such that

η(g1)[η(g2)]1∥≤γg1g21foranyg1,g2G,

whenever g1g2. The smallest value of γ for which the above inequality holds is said to be a Lipschitz constant of η. If γ=1, then η is termed as non-expansive mapping; if γ<1, then η is said to be a contraction.

Definition 1.5. (Bingham & Ostaszewski, Citation2010; Sarfraz & Li, Citation2019) A group-norm ∥:G[0,+) is stated as the abelian group-norm if

g1g2∥=∥g2g1 for all g1,g2G.

Thus the metric is bi-invariant if and only if the group-norm is abelian.

Subadditivity intimates that e∥≥0, and this collectively with symmetry implies that g∥≥0, since e∥=∥gg1∥≤2g. Thus, g∥≥0 for any gG. Subadditivity also refers that gn∥≤ng for every nN.

2. Lipschitz mapping and Banach groups

Assume that (G,) be a normed group, we say it a complete normed group or Banach group if for any gG with gmgn1∥→0 when m,n, then there exists g0G such that gng01∥→0.

Theorem 2.1. Suppose that (G,) be a normed finite group, then (G,) is a Banach group.

Proof. Let {gn} be a Cauchy sequence in G, noted ε=min{g1g21|g1,g2G,g1g2}, since G is a finite group, we can get ε>0 and also take ε0=ε2>0, then there exists N such that for any m>n>N, we have gmgn1∥<ε0. Thus gmgn1∥=0 for any m>n>N. Hence {gn} converges to gN, therefore, (G,) is a complete normed group.

The following corollary concerning the finite subgroup can be easily proved:

Corollary 2.2. Let (G,) be a normed group and H is a finite subgroup of G, then (H,) is a closed subgroup of (G,).

When we consider non-trivial cyclic or non-cyclic normed subgroup G, then we have following theorem:

Theorem 2.3. Assume that (G,,+) be a non-trivial normed subgroup of an additive normed group (R,,+). Then

(a) G is non-cyclic and dense in R if and only if G does not contain minimal positive element.

(b) G is cyclic and closed in R if and only if G contains minimal positive element.

Proof. Define a norm ∥:R[0,+) by

x∥=|x|,(xR).

Then, it can be seen that R is an additive normed group. Assume that G be a non-trivial normed subgroup of R.

(a). Define B={g∥:gG{0}} and β=inf{B}. Suppose that G does not contain minimal positive element. Therefore β=0. Let xR and . Since β=0, therefore, there exists an element g(0,)G. Also, assume that x0 and nN, then

ngx<(n+1)g,

yielding that ngG and ng>0, therefore

0xng<(n+1)gng
0xng<g.

Also, which implies that

0xng<∈.

xng<∈and∈>0.

which also infers that x is within distance from some elements ngG. Since this result holds for every ∈>0 and also for all x; therefore, we can conclude G=R. Conversely, it is easily seen that the condition β=inf{B}=0 provided that G=R.

(b). Assume that G contains minimal positive element β under the assumption that β0. So, by division algorithm, for any gG and for any positive integer n, we have

g∥=nβ+α,0α<β.

Since G is a subgroup, then α=[gnβ]G. But α<β and β is minimum positive element,so α=0. Which concludes that g∥=nβ. Hence G=βZ

If possible, let α>0 and utilize the definition of infimum, choose for every ∈>0 there exist an element g1G such that ng(g1,g1+).Take ∈=β+α

β=α<nβ+g1+α.

β=∥g+g1+α.

Since (g+g1)<0 and (g+g1)G, therefore,

β=∥g+g1<α<β
∥∥g+g1∥<β,

which is a contradiction because β is the minimal positive element. Thus, α=0 and G=βZ is closed and cyclic. Conversely, it is also obvious that when G is cyclic and closed, then there exists positive minimal element β0.

Example 2.4. The group of real numbers R (complex numbers) is a complete normed group with respect to the group-norm ||.

The notion of bi-Lipschitz equivalent inspires us to derive the following result:

Theorem 2.5. Assume that (G,.) be a normed group and

(a) η:GG is bi-Lipschitz equivalent, then

(i) η is bijective mapping.

(ii) η1 is also Lipschitz mapping.

(b) Every isomorphic mapping η:GG is the bi-Lipschitz equivalent, but the converse may or may not be true.

Proof. (a). Suppose η is bi-Lipschitz equivalent, then for any γ1,γ2R+ and g1,g2G we have

γ1g1g21∥≤∥η(g1)[η(g2)]1∥≤γ2g1g21.

Let η(g1)=η(g2) then we can evaluate

γ1g1g21∥≤∥η(g1)[η(g2)]1γ1g1g21∥=0g1=g2,

which concludes that η is injective. Also, η is surjective. Therefore, η1 exists. Consequently, for any g1,g2G, the following computation shows that η1 is 1γ1-Lipschitz:

γ1η1(g1)[η1(g2)]1≤∥η(η1(g1))[η(η1(g2))]1γ1η1(g1)[η1(g2)]1γ2η1(g1)[η1(g2)]1γ1η1(g1)[η1(g2)]1≤∥g1g21η1(g1)[η1(g2)]11γ1g1g21.

(b). Since η is isomorphic, then for every g1G, we have g1∥≤∥η(g1)∥≤∥g1. Then for any g1,g2G, we can determine a bi-Lipschitz equivalent mapping as

g1g21≤∥η(g1g21)∥≤∥g1g21g1g21≤∥η(g1)[η(g2)]1∥≤∥g1g21,

where γ1=γ2=1. Conversely, bi-Lipschitz mapping η can only be isomorphic for particular values γ1=γ2=1 and can not be isomorphic for γ1,γ2>1.

The following corollary, which admittedly is unmotivated here, will be needed later.

Corollary 2.6. Let (G,.) be a normed group, then every isomorphic mapping η:GG is 1-Lipschitz.

Definition 2.7. (Bingham & Ostaszewski, Citation2010) Let (G,.) be a normed group. For some gG, there is self-mapping ηg:GG; then g-conjugate norm is defined as

xg=∥ηg(x)∥=∥gxg1forallxG.

Moreover, Left-shift norm (Lgnorm) and Right-shift norm (Rgnorm) can also be defined as

Lg(x)∥=∥gx,Rg(x)∥=∥xgforallxG.

Definition 2.8. (Klee, Citation1952) Let (G,.) be a normed group, then Klee property of a normed group G is defined as

g1xyg2∥≤∥g1g2+xyforallg1,g2,x,yG.

If a normed group G is abelian, then we can see that it satisfies the Klee property.

Definition 2.9. (Bingham & Ostaszewski, Citation2010) Let (G,.) be a normed group and for some gG, Lipschitz constant for g-conjugacy (γg), Left-shift (L(γg)) and Right-shift (R(γg)) of a normed group G, can be defined as

γg=supxeηg(x)x,lipL(γg)=supxegxx,lipR(γg)=supxexgx.

The following theorem gives a complete description of the conjugate mapping and Klee property (Klee, Citation1952) on abelian normed group.

Theorem 2.10. Let (G,) be a normed group. A conjugacy self-mapping ηg:GG is 1-Lipschitz if and only if G is an abelian normed group.

Proof. For some gG, let us define g-conjugacy self-mapping as ηg(x)=gxg1, for all xG. Assume that G is an abelian normed group. Since G is abelian then Klee property holds, then g-conjugate norm yields that

ηg(x)=∥gxg1∥=∥gxeg1ηg(x)≤∥gg1+xeηg(x)≤∥x.

Hence ηg is a 1-Lipschitz.

Conversely, suppose that ηg is a 1-Lipschitz, which concludes that γg=1 for any gG, then we can compute

(1) gxg1∥≤∥x.(1)

Since ηg is a 1-Lipschitz, so applying (1), we have

gx∥=∥g(xg)g1∥≤∥xg
(2) gx∥≤∥xg.(2)

Since ηg1 also exists and is 1-Lipschitz, which intimates that γg1=1 for any gG, therefore, we can deduce

xg∥=∥g1(gx)g∥≤∥gx
(3) xg∥≤∥gx.(3)

Inequalities (2) and (3) yields that xg∥=∥gx for any x,gG. Hence, G is an abelian normed group.

In the proof of Theorem 2.13, we use the following technical result about conjugate γg-Lipschitz and equivalent group-norm satisfying certain conditions.

Theorem 2.11. Let (G,.) be a normed group and ηg:GG be a conjugacy γg-Lipschitz, then

(a) ηg1=ηg1 also exists and is γg1-Lipschitz.

(b) For γg1=γg=1, ηg:GG has a isomorphic norm.

(c) γe=1 and γg1 for any gG.

(d) γg1g2γg1γg2 for g1,g2G.

(e) Conjugate norm is an equivalent as 1γg1x∥≤∥ηg(x)∥≤γgx.

Proof. (a). From definition of conjugacy norm and also ηg is a γg-Lipschitz, we can deduce ηg(x)∥=∥gxg1∥≤γgx. It is obvious that the mapping is surjective evenly for every xG, there exists g1xgG such that

ηg(g1xg)=gg1xgg1=x.

Then there exists an inverse mapping of ηg indicated as ηg1=ηg1 such that ηg1(x)=g1xg, so we have

ηg1(x1)[ηg1(x2)]1∥=∥g1x1g[g1x2g]1
ηg1(x1)[ηg1(x2)]1∥=∥g1x1x21g
ηg1(x1)[ηg1(x2)]1∥≤γg1x1x21,

which signifies that inverse mapping is also γg1-Lipschitz.

(b). Suppose that γg=γg1=1 for every gG. when γg=1, then we can obtain

(4) gxg1∥≤∥xforanyg,xG.(4)

Also γg1=1, then we can compute

x∥=∥g1(gxg1)g
γg1gxg1
(5) x∥≤∥gxg1.(5)

Inequalities (4) and (5) yields that gxg1∥=∥x, which implies that ηg(x)∥=∥x for all xG. Moreover, mapping ηg is also homomorphic and bijective; hence, a normed group G has an isomorphic group norm.

(c). It is obvious that γe=1 whenever g=e. Let ge, then we can get

ηg(g)=ggg1=g.
g∥=∥ηg(g)∥≤γgg.

Then g∥≤γgg, which implies that γg1.

(d). For g1,g2G, we compute that

ηg1g2(x)∥=∥g1g2x(g1g2)1∥=∥g1(g2xg21)g11
ηg1g2(x)∥≤γg1g2xg21
ηg1g2(x)∥≤γg1g2xg21
ηg1g2(x)∥≤γg1γg2x.

Since γg1,γg21 whenever xe, then we can get γg1g2γg1γg2 for any g1,g2G.

(e). As we know that

(6) ηg(x)∥≤γgx,(6)

then from part (a) ηg1 also exists and is γg1-Lipschitz. Then, for all xG, we have

ηg1(x)∥≤γg1x.

Also,

ηg1(ηg(x))∥≤γg1ηg(x)
x∥≤γg1ηg(x)
(7) 1γg1x∥≤∥ηg(x).(7)

From inequalities (6) and (7), we conclude that conjugate norm is an equivalent because 1γg1x∥≤∥ηg(x)∥≤γgx.

Corollary 2.12. A group G is an abelian normed group if and only if γg=1 for all gG.

In the following results, left-shift (Lg) and left-shift (Rg) mappings plays an important role in the existence of equivalent norms and 1-Lipschitz mappings.

Theorem 2.13. Let (G,.) be a normed group. Also Lg, Rg, and ηg be self-mappings from G to G and also lipL(γg), lipR(γg), γg-Lipschitz respectively, then

(a) Lg under left-invariant metric dL is 1-Lipschitz and is γg-Lipschitz concerning dR.

(b) Rg under right-invariant metric dR is 1-Lipschitz and is γg1-Lipschitz concerning dL.

(c) lipL(γe)=lipR(γe)=1 and lipL(γg)lipL(γg1)1, lipR(γg)lipR(γg1)1 for any gG.

(d) lipL(γg1g2)lipL(γg1)lipL(γg2) and lipR(γg1g2)lipR(γg1)lipR(γg2) for all g1,g2G.

(e) If lipL(γg)=lipR(γg)=1 for all gG, then G is an abelian normed group.

(f) Lg and Rg have equivalent norms.

Proof. (a). By left-invariant metric dL(x1,x2)=∥x11x2 for x1,x2G, then proof of (a) consists of the following simple computation:

[Lg(x1)]1Lg(x2)∥=dL(Lg(x1),Lg(x2))
=∥(gx1)1gx2
=∥x11g1gx2
=∥x11x2
[Lg(x1)]1Lg(x2)∥=∥x11x2.

Also, by right-invariant metric dR(x1,x2)=∥x1x21 for x1,x2G, we can get

Lg(x1)[Lg(x2)]1∥=dR(Lg(x1),Lg(x2))
=∥(gx1)(gx2)1
=∥gx1x21g1
γgx1x21
Lg(x1)[Lg(x2)]1∥≤γgx1x21,

which is required proof.

(b). Assume that x1,x2G, then required proof consists of the following simple computation:

Rg(x1)[Rg(x2)]1∥=dR(Rg(x1),Rg(x2))
=∥(x1g)(x2g)1
=∥x1gg1x21
=∥x1x21
Rg(x1)[Rg(x2)]1∥=∥x1x21.

Also,

[Rg(x1)]1Rg(x2)∥=dL(Rg(x1),Rg(x2))
=∥(x1g)1(x2g)
=∥g1x11x2g
γg1x11x2
[Rg(x1)]1Rg(x2)∥≤γg1x11x2,

as required.

(c). Let g=e, then we can conclude that lipL(γe)=1=lipR(γe). Assume that ge, then we can compute

x∥=∥g(g1x)
lipL(γg)g1x
x∥≤lipL(γg)lipL(γg1)x.

Then x∥≤lipL(γg)lipL(γg1)x, which implies that lipL(γg)lipL(γg1)1. Similarly, for right shift (Lg) mapping we can also calculate lipR(γg)lipR(γg1)1.

(d). Let g1,g2,xG, then

Lg1g2(x)∥=∥g1(g2x)
Lg1g2(x)∥≤lipL(γg1)g2x
Lg1g2(x)∥≤lipL(γg1)lipL(γg2)x.

By taking suprimum whenever xe, we can obtain that lipL(γg1g2)lipL(γg1)lipL(γg2) for g1,g2G. The conclusion is easily seen to hold lipR(γg1g2)lipR(γg1)lipR(γg2) for right shift mapping (Rg).

(e). Assume that lipL(γg)=lipR(γg)=1 for all gG. Then, it is obvious that Lg and Rg are 1-Lipschitz. Consider gxg1 and using Lg and Rg as a 1-Lipschitz, we have

gxg1∥≤lipL(γg)xg1
lipL(γg)lipR(γg1)x
gxg1∥≤∥x,

which holds for all gG, and implies that ηg is a 1-Lipschitz. Using the theorem that a conjugacy self-mapping ηg:GG is 1-Lipschitz, then G is an abelian normed group. Hence, G is an abelian normed group.

(f). As we know that for some fix element gG and for every xG, we have

(8) Lg(x)∥≤lipL(γg)x.(8)

Also, it can be seen that left shit Lg is surjective and inverse mapping Lg1=Lg1 also exists, and is lipL(γg1)-Lipschitz, then we have

Lg1(x)∥≤lipL(γg1)x.

Also,

Lg1(Lg(x))∥≤lipL(γg1)Lg(x)
x∥≤lipL(γg1)Lg(x)
(9) 1lipL(γg1)x∥≤∥Lg(x).(9)

Inequalities (8) and (9) implies that Lg norm is an equivalent because 1lipL(γg1)x∥≤∥ηg(x)∥≤lipL(γg)x. Similarly, the Rg norm also has an equivalent norm.

The following example gives a complete description of the conjugate mappings on additive normed group.

Example 2.14. Let (G,,+) be a normed group of Complex or Real numbers. Then for some gG all mappings ηg:GG are 1-Lipschitz.

Using continuous mappings, we prove the following theorem for isomorphic norm.

Theorem 2.15. A continuous self-mapping on a normed finite group G has an isomorphic norm.

Proof. Assume that η:GG be a continuous mapping. Since η is continuous and G is a normed finite group, then there exists a finite nN such that ηn=identity mapping, as ηn(g)=g for all gG. So, we have an automorphism mapping ηn:GG satisfies ηn(g)=g for all gG. Therefore, ηn(g)∥=∥g holds for all gG, and hence group norm is isomorphic.

Corollary 2.16. For a normed group G, isomorphic self-mapping is 1-Lipschitz.

Proof. Assume that η:GG is an isomorphic self-mapping, then η(g)∥=∥g holds for all gG. Consequently, η(g1)[η(g2)]1∥=∥η(g1)η(g21)∥=∥η(g1g21)∥=∥g1g21. Hence this self-mapping is 1-Lipschitz.

3. The fixed point theorem in Banach groups

A contraction mapping theorem is an important tool in the theory of normed groups; it guarantees the existence and uniqueness of fixed points of certain self-maps and provides a constructive method to find those fixed points. The following example gives the completion of a normed group G.

Example 3.1. Let G be a Banach space (it can be seen that a Banach space is an additive commutative group). Then g∥=d(e,g) defines a group-norm on G and is also complete normed group.

Our concern is to find sufficient conditions on normed group G and self-mapping η to ensure a fixed point of η in G. We are also interested in uniqueness and procedures for the calculation of fixed points.

For this purpose, the first main result is giving by the following theorem.

Theorem 3.2. Let η be a contraction on a complete normed group G. Then η has a unique fixed point g0G such that η(g0)=g0.

Proof. Notice first that if g1,g2G are fixed points of η, then

g1g21∥=∥η(g1)η(g2)1∥≤γg1g21,

which implies that g1=g2.

Choose now any g1G, and define the iterate sequence gn+1=η(gn). By induction on n>1,

gn+1gn1∥≤γnη(g1)g11.

If nN and m1,

gm+ngn1∥=∥gm+ngm+n11gm+n1gm+n21gm+n2gn+1gn1
≤∥gm+ngm+n11++gn+1gn1
(γm+n+γm+n1++γn)η(g1)g11
gm+ngn1∥≤γn1γη(g1)g11.

Hence {gn} is a Cauchy sequence and admits a limit g0G because G is complete. Since η is continuous, we have η(g0)=limη(gn)=limgn+1=g0. Hence η has a unique fixed point g0G such that η(g0)=g0.

The following corollary is easily seen to hold if normed group G is finite.

Corollary 3.3. Let η be a contraction mapping of finite order from a normed finite group (G,.) into itself. Then η has a fixed point in G.

Let η be a homomorphic mapping of a normed group (G,.) into itself. Then, η must have a fixed point eG such that η(e)=e.

The following theorem provides sufficient conditions for the existence of inverse Lipschitz mapping and isomorphic mapping whenever we have contraction map.

Theorem 3.4. Assume that (G,.) is an abelian Banach group and η be a homomorphic γηLipschitz mapping of a Banach group G into itself such that γη<1. Then

(a) There exists an isomorphic γLipschitz mapping p:GG,

(b) p1 also exists and is (1γη)1-Lipschitz,

where

γη=min{γR+:ηisγLipschitz}.

Proof. (a). For any uG, a self-mapping ηu:GG is defined by ηu(g)=uη(g). We proceed in a few steps. In the first step, we show that the defined mapping ηu is also contraction mapping concerning η. Since G is abelian and η be a homomorphic γηLipschitz then for every g1,g2G, we have

ηu(g1)[ηu(g2)]1∥=∥uη(g1)[uη(g2)]1
=∥η(g1)[η(g2)]1
ηu(g1)[ηu(g2)]1∥≤γηg1g21,

which infers that ηu is also a contraction mapping.

Since G is a Banach group and ηu is a contraction mapping, according to the contraction mapping theorem, there exists a unique element gG as follows

ηu(g)=uη(g).

Define a mapping p:GG by

p(g)=g[η(g)]1.

Thus, ηu(g)=g if and only if p(g)=u. Assume that g1,g2G such that

p(g1)=p(g2)
g1[η(g1)]1=g2[η(g2)]1
g1g21=η(g1)[η(g2)]1
g1g21=η(g1g21).

Since η is also contraction mapping, it is possible whenever g1g21=e, so g1=g2. Consequently, p is an injective mapping. Also, ηu(g)=g if and only if p(g)=u, therefore, p is also surjective. Next, we determine that p is also a homomorphism.

p(g1g2)=g1g2[η(g1g2)]1
p(g1g2)=g1g2[η(g2)]1[η(g1)]1
p(g1g2)=g1[η(g1)]1g2[η(g2)]1
p(g1g2)=p(g1)p(g2),

which indicates that p is a homomorphism. Lastly, to check the existence of Lipschitz mapping, let g1,g2G such that

p(g1)[p(g2)]1=∥g1[η(g1)]1[g2[η(g2)]1]1=∥g1[η(g1)]1η(g2)g21=∥g1g21[η(g1)]1η(g2)≤∥g1g21+η(g1)[η(g2)]1≤∥g1g21+γηg1g21(1+γη)g1g21=γg1g21p(g1)[p(g2)]1γg1g21,

where 1+γη=γ, and so the mapping p is also γLipschitz. Hence, p is an isomorphic γLipschitz.

(b). Since p is surjective, therefore, inverse mapping p1 also exist. Now we prove that the inverse mapping is also γLipschitz. So for u1,u2G, we have

p1(u1)[p1(u2)]1=∥u1u21=∥ηu1(g1)[ηu2(g2)]1=∥u1η(g1)[u2η(g2)]1=∥u1η(g1)[η(g2)]1u21≤∥u1u21+η(g1)[η(g2)]1≤∥u1u21+γηg1g21≤∥u1u21+γηp1(u1)[p1(u2)]1(1γη)p1(u1)[p1(u2)]1≤∥u1u21p1(u1)[p1(u2)]111γηu1u21p1(u1)[p1(u2)]1(1γη)1u1u21.

Hence the inverse of p is also (1γη)1-Lipschitz.

The following theorem shows that there exists singleton set consisting a unique fixed point.

Theorem 3.5. Let η be a contraction mapping from a Banach group G to itself. Then there exists a function h:GR0 such that h1({0}) is a singleton and h(η(g))γh(g) for every gG and 0<γ<1.

Proof. Since η:GG be a contraction mapping, therefore, for every g1,g2G and 0<γ<1, we have

η(g1)[η(g2)]1∥≤γg1g21.

By contraction principle, η has a unique fixed element gG such that η(g)=g. Define a mapping h:GR0 such that

h(g)=∥g(g)1,gG.

Then h(g)=0 if and only if g=g, which implies that h1({0})={g}. Also

h(η(g))=∥η(g)[η(g)]1γg(g)1h(η(g))γh(g),

which is required proof.

Let G=[0,1) and its norm ∥:G[0,+) defined by r∥=min{r,1r} for all rG. Then, (G,,+) is a normed group. Consequently, we have the following fixed point theorems concerning the Banach group using normed group G.

Theorem 3.6. Assume that (G,.) be a Banach group and η:GG is a mapping satisfies the condition

η(g1)[η(g2)]1∥≤pg1η(g1)1+qg2η(g2)1+rg1g21

for all g1,g2G, where p,q,rG such that {t(q+r)+p}G for t1. Then η has a unique fixed point g0G such that η(g0)=g0.

Proof. Let g0G and {gn} be a sequence of elements in G such that we can define the iterate sequence gn=η(gn1), then we can obtain

gn+1η(gn)1=∥η(gn)[η(gn1)]1pgnη(gn)1+qgn1η(gn1)1+rgngn11gn+1η(gn)1pgngn+11+qgn1gn1+rgngn11(1p)gn+1η(gn)1qgngn11+rgngn11(1p)gn+1η(gn)1(q+r)gngn11gn+1η(gn)1q+r1pgngn11gn+1η(gn)1γgngn11,

where γ=q+r1pG. Continuing this process we can get

gn+1η(gn)1∥≤γng1g01.

As {t(q+r)+p}G, then γ=q+r1p<1t, which refers that γG whenever t1, therefore mapping η is a contraction mapping. Further, we will demonstrate that {gn} is a Cauchy sequence in G. Assume that m>n for m,n>0, then using triangle inequality, we can ascertain that

gngm1≤∥gngn+11+gn+1gm1≤∥gngn+11+gn+1gn+21++gm1gm1γng0g11+γn+1g0g11+γn+2g0g11++γm1g0g11γng0g11[1+γ+γ2+]gngm1γn1γg0g11.

Taking limit n when γG, then we have limngngm1∥=0. Since G is a complete normed group, then sequence gn converges to an element g0G. Subsequently, we will demonstrate that g0 is a fixed point of η. Consider

η(g0)g01≤∥η(g0)gn1gng01≤∥η(g0)gn1+gng01=∥η(gn1)η(g0)1+gng01pgn1η(gn1)1+qg0η(g0)1+rgn1g01+gng01(1q)η(g0)g01pγng0g11+rgn1g01+gng01,

then taking limit n when γG, we can obtain

limnη(g0)g01∥=0,

then η(g0)g01=e, which intimates that η(g0)=g0. Therefore g0 is a fixed point of η in G. To prove the fixed point’s uniqueness, assume that g1 and g2 be two fixed points of η such that η(g1)=g1 and η(g2)=g2, then given condition yields that

g1g21=∥η(g1)[η(g2)]1pg1η(g1)1+qg2η(g2)1+rg1g21pg1g11+qg2g21+rg1g21g1g21rg1g21,

which is a contradiction because rG, therefore g1=g2. Hence η has a unique fixed point g0G such that η(g0)=g0.

Example 3.7. Let G=Q, then it is a normed group with the padic norm ||p for prime p2. Taking the sequence gn=1+p+p2++pn1, then gn is a Cauchy sequence and this sequence has a limit with respect to group-norm ||p such that limn(p)gn=11pQ, therefore (Q,||p) is a complete normed group.

Before proving our next theorem concerning fixed point theorem, we need the following lemma:

Lemma 3.8. Let η be a mapping from a Banach group G to itself and satisfies the following condition

η(g1)[η(g2)]1∥≤a1g1η(g1)1+a2g2η(g2)1+a3g1η(g2)1+a4g2η(g1)1+a5g1g21,

where i=15ai=uG and for each i, aiG. Then for every uG there exists vG such that

η(g)[η2(g)]1∥≤vg[η(g)]1.

Proof. Let g2=η(g1), then according to given condition we have,

η(g1)[η2(g1)]1a1g1η(g1)1+a2η(g1)[η2(g1)]1+a3g1[η2(g1)]1+a4η(g1)η(g1)1+a5g1g11(1a2)η(g1)[η2(g1)]1(a1+a5)g1η(g1)1+a3g1[η2(g1)]1η(g1)[η2(g1)]1a1+a51a2g1η(g1)1+a31a2g1[η2(g1)]1,

then using triangle inequality, we can derive

η(g1)[η2(g1)]1a1+a51a2g1η(g1)1+a31a2g1[η(g1)]1+a31a2η(g1)[η2(g1)]11a2a31a2η(g1)[η2(g1)]1a1+a3+a51a2g1η(g1)1η(g1)[η2(g1)]1a1+a3+a51a2a3g1η(g1)1.

By the symmetry of given condition in the theorem, write a2 instead of a1 and a4 instead of a3, then we can conclude that

η(g1)[η2(g1)]1∥≤a2+a4+a51a1a4g1η(g1)1.

Assume that v=min[a1+a3+a51a2a3,a2+a4+a51a1a4], then we obtain that vG such that

η(g1)[η2(g1)]1∥≤vg1η(g1)1.

Using Lemma 3.8, the following theorem provides sufficient conditions for the existence and uniqueness of fixed point under specified condition.

Theorem 3.9. Let η be a mapping from a Banach group G to itself and satisfies the condition

η(g1)[η(g2)]1∥≤a1g1η(g1)1+a2g2η(g2)1+a3g1η(g2)1+a4g2η(g1)1+a5g1g21

for all g1,g2G, where i=15ai=uG and for each i, aiG. Then η has a unique fixed point g0G such that η(g0)=g0.

Proof. Let g0G and {gn} be a sequence of elements in G such that we can define the iterate sequence gn=η(gn1). Also, by Lemma 3.8, it follows that for every uG there exists γG such that

gn+1gn1∥≤γng1g01.

Further, we will demonstrate that {gn} is a Cauchy sequence in G. Assume that m>n for m,n>0, then using triangle inequality, we can obtain that

gngm1≤∥gngn+11+gn+1gm1≤∥gngn+11+gn+1gn+21++gm1gm1γng0g11+γn+1g0g11+γn+2g0g11++γm1g0g11γng0g11[1+γ+γ2++γm1]γng0g11i=0γigngm1γn1γg0g11.

Taking limit n when γG, then we conclude limngngm1∥=0. Since G is a Banach group, therefore, the sequence gn converges to an element g0G. Subsequently, we will demonstrate that g0 is a fixed point of η. For this purpose consider

η(g0)g01≤∥η(g0)gn1gng01≤∥η(g0)gn1+gng01=∥η(gn1)η(g0)1+gng01a1gn1η(gn1)1+a2g0η(g0)1+a3gn1η(g0)1+a4g0η(gn1)1+a5gn1g01+gng01(1a2)η(g0)g01a1γn1g0g11+a3gn1η(g0)1+a4g0η(gn1)1+a5gn1g01+gng01,

and taking limit n whenever γG then we get

η(g0)g01∥≤(a2+a3)η(g0)g01.

Since i=15aiG, therefore, (a2+a3)G. So, which is possible only if η(g0)g01=e which implies that η(g0)=g0, therefore g0 is a fixed point of η in G. To prove the fixed point’s uniqueness, assume that g1 and g2 be two fixed points of η such that η(g1)=g1 and η(g2)=g2, then given condition yields that

g1g21=∥η(g1)[η(g2)]1a1g1η(g1)1+a2g2η(g2)1+a3g1η(g2)1+a4g2η(g1)1+a5g1g21a3g1η(g2)1+a4g2η(g1)1+a5g1g21g1g21(a3+a4+a5)g1g21,

which is a contradiction because (a3+a4+a5)G, therefore g1=g2. Hence η has a unique fixed point g0G such that η(g0)=g0.

The following corollary gives an important characterization of Banach group G that possess continuous mappings having fixed point (may not be unique).

Corollary 3.10. Let η be a continuous mapping of a Banach group G into itself and also holds the condition

η(g)[η2(g)]1∥≤vg[η(g)]1

for any gG such that g[η(g)]1∥< and vG. Then there exists n0 such that ηn(g)[ηn+1(g)]1∥<. And also there exists a fixed point wG such that η(w)=w.

Remarks 3.11. The mapping η satisfies the condition of Corollary 3.10 may have more than one fixed points when we consider the mapping η(g)=g for any gG.

The following corollary about normed finite group gives the existence of unique fixed point.

Corollary 3.12. Let η be a contraction mapping of a normed finite group G into itself. Then η has a unique fixed point g0G such that η(g0)=g0. In fact, for any gG,g1=η(g),,gn+1=η(gn), there exist some integer N such that for any nN, we have gn=gN.

Definition 3.13. (Hussain et al., Citation2015) Let (G,) be a normed group. A mapping η of a normed group G into itself is said to be expansive if there is γ>1 such that

η(g1)[η(g2)]1∥≥γg1g21

for any g1,g2G and g1g2.

It is easy to see that any expansive mapping must be injective; therefore, we have the following theorem to determine a fixed point for expansive mappings.

Theorem 3.14. Let η be an expansive and surjective mapping of a Banach group G into itself. Then η has a unique fixed point g0G such that η(g0)=g0.

Proof. Since any expansive mapping is injective, therefore η is bijective. Hence η1 exists and satisfies the following;

η1(g1)[η1(g2)]1∥≤1γg1g21,

where 1γ<1, thus η1 is a contraction mapping. According to the fixed point theorem, there exists g0G such that η1(g0)=g0. Hence g0 is a fixed point of η.

The following theorem gives an important characterization of finite normed cyclic group that possess expansive mapping.

Theorem 3.15. Let η be an expansive mapping of a finite normed cyclic group G into itself, then

(a) η is bi-Lipschitz equivalent.

(b) η has a unique fixed point g0G such that η(g0)=g0.

Proof. (a). Because G is a normed cyclic group, therefore we have g1g2∥=∥g2g1 for any g1,g2G. Assume that |G|=m, then define a mapping

η(g)=gn,nN,gG,

where gcd(n,m)=1 for nN.

Further, it is obvious that gn∥≤ng for nN.

Considering an expansive mapping η and γ>1, then for any g1,g2G, proof of (a) consists of the following simple computation:

γg1g21≤∥η(g1)[η(g2)]1≤∥g1n[g2n]1≤∥[g1g21]nγg1g21≤∥η(g1)[η(g2)]1∥≤ng1g21.

(b). Since η be an expansive mapping and γ>1, then for every g1,g2G, we can obtain

γg1g21∥≤∥η(g1)[η(g2)]1.

If g1n=g2n then g1ng2n=e, therefore, we have

γg1g21≤∥g1ng2nγg1g21=∥eγg1g21=0g1=g2,

which infers that η is injective. So η is bijective, then η1 exists. Also, by expansive mapping

γη1(g1)[η1(g2)]1≤∥η(η1(g1))[η(η1(g2))]1γη1(g1)[η1(g2)]1≤∥g1g21η1(g1)[η1(g2)]11γg1g21,

thus η1 is also contraction mapping for γ>1. Since G is finite, therefore it is a complete normed group. According to the fixed point theorem, there exists g0G such that η1(g0)=g0. Hence g0 is a fixed point of η.

4. Conclusion

Thus, in this paper, we have proposed and investigated the general properties of Lipschitz mapping of a normed group G to itself and concluded that every isomorphic mapping is 1-Lipschitz. Moreover, we have also demonstrated that the conjugate Lipschitz mapping ηg has an equivalent norm. It is observed that the left-shift (Lg) under left-invariant metric (dL) and right-shift (Rg) under right-invariant metric (dR) are 1-Lipschitz, and also Lg and Rg mappings have equivalent norms. Particularly, for a normed finite group G, a continuous self-mapping η has an isomorphic norm.

We have also shown that for homomorphic Lipschitz mapping of the abelian Banach group, there is contraction mapping and isomorphic mapping such that the inverse Lipschitz mapping also exists. We have formulated some fixed point theorems for contraction mappings of a Banach group G and determined unique fixed points following certain conditions. It is also proved that the contraction mapping of a normed finite group G to itself has a unique fixed point in G. We have concluded that an expansive mapping of a finite normed cyclic group G has a unique fixed point in G and is also bi-Lipschitz equivalent.

Conflicts of Interest

The author(s) declare(s) that there is no conflict of interest regarding the publication of this paper.

Acknowledgements

The authors are grateful to the referees and the editors for valuable comments and suggestions, which have improved the original manuscript greatly.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [11971493].

Notes on contributors

Yongjin Li

Dr. Yongjin Li is working as a professor at the School of Mathematics, Sun Yat-sen University. His area of specialization is Functional Analysis, Differential Equations. Under his guidance, 04 students are working for a Ph.D., and 03 post Doctors are working in his group. He has published 150 Research papers, 06 books and completed 03 projects for the National Natural Science Foundation of China.

Muhammad Sarfraz is currently a Ph.D. Research Scholar under the supervision of Professor Dr. Yongjin Li in School of Mathematics, Sun Yat-sen University. His area of specialization is Functional Analysis and Cryptography. He has published 7 Research papers.

Fawad Ali is currently a Ph.D. Research Scholar with the School of Mathematics and Statistics Xi’an Jiaotong University, Xi’an 710049, P. R. China. He has been working in algebraic graph theory since 2017.

References