Skip to main content

Full text of "On Multi-Metric Spaces"

See other formats


On Multi-Metric Spaces 



Linfan Mao 

(Academy of Mathematics and System Sciences^ Chinese Academy of Sciences, Beijing 100080) 



Abstract: A Smarandache multi-space is a union of n spaces Ai, A2, • ■ ■ , An 
with some additional conditions holding. Combining Smarandache multi- 
spaces with classical metric spaces, the conception of multi-metric space is 
introduced. Some characteristics of a multi-metric space are obtained and 
Banach's fixed-point theorem is generalized in this paper. 

Key words: metric, multi-space, multi-metric space, ideal subspace chain. 

Classification: AMS(2000) 51D99,51K05,51M05 

1. Introduction 

The notion of multi-spaces is introduced by Smarandache in [6] under his idea 
of hybrid mathematics: combining different fields into a unifying field{[7]), which is 
defined as follows. 

Definition 1.1 For any integer i,l < i < n let Ai be a set with ensemble of 
law Li, and the intersection of k sets Ai^^Ai^^- ■ ■, of them constrains the law 
I{Ai^^ Ai^, ■ ■ ■ , Afc)- Then the union of Ai, 1 < i < n 



A = [JA, 



1=1 



is called a multi-space. 



As we known, a set M associative a function p : M x M ^ = {x \ x & R, x > 
0} is called a metric space if for \lx,y,z G M, the following conditions for the metric 
function p hold: 

{l){definiteness) p{x, y) = if and only ii x = y; 

{ii) {symmetry) p{x,y) = p{y,x); 

(Hi) {triangle inequality) p{x, y) + p{y, z) > p{x, z). 

By combining Smarandache multi-spaces with classical metric spaces, a new kind 
of space called multi-ring space is found, which is defined in the following. 

— m 

Definition 1.2 A multi-metric space is a union M = {J such that each Mi is a 
space with metric pi for \/i,l < i < m. 

— m 

When we say a multi-metric space M = [j Mi, it means that a multi- metric 

i=l 



1 



space with metrics Pi, P2, • " • > Pm such that {M^, pi) is a metric space for any integer 

— m — 

< i < m. For a multi-metric space M = \J Mi, x E M and a positive number 

i=l 

R, a R-disk B{x, R) in M is defined by 

B{x, R) = {y \ there exists an integer k,l < k < m such that pk{y, x) < R,y G M} 

The subject of this paper is to find some characteristics of a multi-metric space. 
For terminology and notation not defined here can be seen in [1] — [2], [4] for termi- 
nologies in metric space and in [3] , [5] — [9] for multi-spaces and logics. 

2. Characteristics of a mult i- metric space 

For metrics on a space, we have the following result. 

Theorem 2.1 Let pi, p2, ■ ■ ■ , Pm be m metrics on a space M and F a function on 
such that the following conditions hold: 

(i) F{xi,X2, - • ■ ,Xm) > F{yi,y2, - ■ • ,ym) if for yi, I < i < m, Xi > yf, 
{a) F{xi, X2,- - ■ , Xm) = only if xi ^ X2 ^ ■ ■ ■ ^ Xm ^ 0] 
(Hi) For two m-tuples {xi,X2, - ■■ , Xm) and (t/i, ^2, • • • , Z/m), 

X2, • • • , Xm) + F{yi,y2, ■■■ ,ym)> F{xi + yi, X2 + y2, ■ ■ ■ ,Xm + ym)- 

Then F{pi, p2, ■ ■ ■ , Pm) is also a metric on M. 

Proof We only need to prove that F{pi, p2, • ■ ■ , Pm) satisfies the metric conditions 

for Wx, y, z E M. 

By (ii), F{pi{x, y), p2{x,y),- ■ ■ , pm{x, y)) = Q only if for any integer i, pi{x, y) = 
0. Since pi is a metric on M, we know that x — y. 

For any integer i,l < i < m, since pi is a metric on M, we know that pi{x,y) — 
Pi{y,x). Whence, 

F(pi{x, y),P2(x, y),---, Pm(x, y)) = F(pi(y, x),p2(y, x), • • • , pm(y, x)). 
Now by (i) and {Hi), we get that 

Fipiix, y),p2ix, y), ■ ■ ■ , pmix, y)) + F{pi{y, 2), p2(y, z),---, pm{y, z)) 

> F{pi{x, y) + pi{y, z),p2{x, y) + p2{y, 2), • • • , pm{x, y) + Pm{y, z)) 

> F{pi{x, Z),P2{X,Z),---, Pm{x, Z)). 

Therefore, F{pi, p2, • • • , Pm) is a metric on M. \ 



2 



Corollary 2.1 li pi, p2, ■ ■ ■ , Pm are m metrics on a space M, then pi+ P2 + - ■ ■ + Pm 
and — h -TT^ — I h 7^2!^ are also metrics on M. 

l+pi 1+P2 1+Pm 

— m 

A sequence {xn} in a multi- metric space M = U Mj is said convergence to 

a pomt X, X G M if for any positive number e > 0, there exist numbers N and 
i,l < i < m such that if n > then 

Pi{xn,x) < e. 

For {xn} convergence to a point x,x & M, we denote it by limx„ = x. 

We have a characteristic for convergent sequences in a multi-metric space. 

— m 

Theorem 2.2 A sequence in a multi-metric space M — \J Mi is convergent 

i=l 

if and only if there exist integers N and k,l < k < m, such that the subsequence 
{xn\n > N} is a convergent sequence in {Mk,pk). 

Proof If there exist integers and k,l < k < m, such that {a;„|n > A^} 
is a convergent subsequence in {Mk.pk), then for any positive number e > 0, by 
definition there exists a positive integer P and a point x,x G such that 

Pk{xn,x) < e 

if n > max{N, P}. 

Now if {xn} is a convergent sequence in the multi-space M, by definition for any 
positive number e > 0, there exist a point x,x E M and natural numbers N(e) and 
k,l < k < m, such that if n > N{e), then 

Pk{xn,x) < e, 

that is, {xn\n > N{e)} C and {xn\n > N{e)} is a convergent sequence in 
{Mk,Pk). \\ 

— m 

Theorem 2.3 Let M — \J Mi be a multi-metric space. For two sequences {xn}, {Un} 

1=1 

in M, if\\m.Xn — xq, limy„ = vq and there is an integer p such that xo,yo £ Mp, 

n n 

then ]im. pp{xn,yn) = Pp{xo,yo). 

Proof According to Theorem 2.2, there exist integers A^i and A^2 such that if 
n > max{Ni, N2}, then Xn, yn £ Mp. Whence, we have that 

Pp{Xn, Vn) < Pp{Xn, Xq) + pp{xQ, yo) + pp{yn, Vo) 

and 

Pp{xo, yo) < pp{Xn, Xo) + Pp{Xn, yn) + PpiVn, Vo)- 



3 



Therefore, 



\pp{Xn,yn) - Pp{xQ,yo)\ < pp{Xn, Xq) + Pp{yn, Vo) ■ 

For any positive number e > 0, since lim.T,„ = Xq and limy^i = y^. there exist 

n n 

numbers A'^i(e), A''i(e) > A^i and N2{e),N2{e) > N2 such that pp{xn,XQ) < | if 
n > Ni{e) and Pp{yn,yo) < f if > ^2(e). Whence, if n > max{Ni[e),N2{e)}, 



then 



\pp{Xn,yn) - Pp{xo,yQ)\ < C t] 



Whether a convergent sequence can has more than one hmit point? The following 
result answers this question. 

— m 

Theorem 2.4 is a convergent sequence in a multi-metric space M = U Mj, 

then {xn} has only one limit point. 



i=l 



Proof According to Theorem 2.2, there exist integers N and z, 1 < z < m such 
that a;„ e Mj if n > N. Now if 

lima;^ = xi and limx„ = 0:2, 

n n 

and n > A^, by definition, 

< Piixi,X2) < Pi{Xn,Xi) + Pi{Xn,X2). 

Whence, we get that Pi{xi, X2) — 0. Therefore, xi = X2. \\ 

Theorem 2.5 Any convergent sequence in a multi-metric space is a bounded points 
set. 

Proof According to Theorem 2.4, we obtain this result immediately. t] 

— m 

A sequence {xn\ in a multi-metric space M — (J Mj is called a Cauchy sequence 

1=1 

if for any positive number e > 0, there exist integers A^(e) and s, 1 < s < m such 
that for any integers m,n> N{e), Ps{xm, Xn) < e- 

— m 

Theorem 2.6 A Cauchy sequence in a multi-metric space M = \J Mi is 

1=1 

convergent if and only if for Wk,l < k < m, |{a;„}n-^fc| is finite or infinite but 
{xn} n is convergent in {Mk, pk) ■ 

Proof The necessity of conditions is by Theorem 2.2. 

Now we prove the sufficiency. By definition, there exist integers s, 1 < s < 
m and A^i such that Xn & Mg SI n > Ni. Whence, if |{a;„}n-^fe| is infinite 
and lim{a;„} fl = x, then there must be A; = s. Denoted by {xn} fl — 



4 



For any positive number e > 0, there exists an integer N2, N2 > Ni such that 
Pk{xm,Xn) < I and pk{xkn,x) < I if m, n > N2. According to Theorem 4.7, we 
get that 

Pk{Xn, X) < Pk{Xn, Xkn) + Pk{Xkn, x) < 6 
if n > A^2- Whence, hm Xyi Xt U 

n 

A multi-metric space M is said completed if every Cauchy sequence in this space 
is convergent. For a completed multi- metric space, we obtain two important results 
similar to metric space theory in classical mathematics. 

— m 

Theorem 2.7 Let M = U he a completed multi-metric space. For a e-disk 

1=1 

sequence {-B(e„, Xn)}, where > for n = 1, 2, 3, ■ ■ the following conditions hold: 

(i) B{ei,xi) D B{e2, X2) D B{e3, Xa) D • • • D B{en, Xn) ^ ■ ■ •; 

(ii) lim e„ = 0. 

+00 

Then fj B{en,Xn) only has one point. 

n=l 

Proof First, we prove that the sequence {xn} is a Cauchy sequence in M. By the 
condition (i), we know that if m > n, then Xm & -B(e^,a;^) C B{en,Xn). Whence, 
for Vi, 1 < i < m, Pi{xm, x^) < e„ if 

Xrm Xn 

For any positive number e, since lim — 0, there exists an integer N(e) such 

n— >+oo 

that if n > N{e), then e„ < e. Therefore, if Xn G Mi. then lim,T„j = .x„. Whence, 
there exists an integer such that if m > A, then Xm G Mi by Theorem 2.2. Take 
integers m,n> max{N, N{e)}. We know that 

Pi{xmi Xfi) < < e. 

So {xn} is a Cauchy sequence. 

By the assumption, M is completed. We know that the sequence is conver- 
gence to a point xq, xq G M. By conditions (i) and {ii), we have that pi{xo, Xn) < Cn 

+00 

if we take m — > -|-oo. Whence, xq E f] B{en,Xn)- 

n=l 

+00 

Now if there a point ye f] B{en,Xn), then there must he y e Mi. We get that 

n=l 

< Pi{y, xo) = lim Pi (y, Xn) < lim e„ = 

n n— »+oo 

by Theorem 2.3. Therefore, pi{y,xo) — 0. By definition of a metric on a space, we 
get that y = xq. t] 

Let Ml and M2 be two multi-metric spaces and / : Mi M2 a mapping, 
Xo G Ml, f{xo) = I/O- For Ve > 0, if there exists a number 6 such that for forallx G 
B{6,Xo), fix) = ye B{e,yo) C M2, i.e., 

/(B(5,a:o))c5(e,yo), 



5 



then we say / is continuous at point Xq. If / is connected at every point of Mi, then 

/ is said a continuous mapping from Mi to M2. 

For a continuous mapping / from Mi to M2 and a convergent sequence {xn} in 
Ml, limxn = Xq, we can prove that 

n 

\imf{xn) = f{xo). 

— m — — 

For a multi-metric space M = \J M^ and a mapping T : M ^ M, if there is a 

i=l 

point X* G M such that Tx* = x*, then x* is called a /ixec? |)omt of T. Denoted by 
*$(T) the number of all fixed points of a mapping T in M. If there are a constant 
a, 1 < a < 1 and integers i,j, I < i, j < m such that for \/x,y e Mj, Tx,Ty e M,- 
and 

Pj(Tx,Ty) < api{x,y), 
then T is called a contraction on M. 

— m 

Theorem 2.8 Let M = [j Mi be a completed multi-metric space andT a contraction 
_ 1=1 

on M. Then 

1 <* $(T) < m. 
Proof Choose arbitrary points xo,yo £ Mi and define recursively 

•^n+l TXyi, l/n+1 TXn 

for 77, = 1, 2, 3, ■ ■ ■. By definition, we know that for any integer n,n > 1, there exists 
an integer i,l < i < m such that Xn,yn G -/Wj. Whence, we inductively get that 

< Pi{xn,yn) < a"-pi{xo,yo). 
Notice that < a < 1, we know that lim a"' — 0. Therefore, there exists an 

n— >+oo 

integer i^ such that 

Pio(limx„,lim^/„) = 0. 

Therefore, there exists an integer Ni such that ?/„ e Mj(, if n > A^i. Now if 
n> Ni, we have that 

— (^Pioi.-^nj ^PioiTxYi—l,TXyi—2) 

< a^Pi^{Xn-l,Xn-2) < ■ ■ < a"'~^' Pio{xNi+l, XnJ. 



6 



and generally, ior m > n > Ni, 



< (a"*-' + a"^-' + • • • + a'')pi,{xN,+i, x^J 

a" 

< pio(xAri+i,a;ArJ ^ 0(m,n ^ +00) 

1 — a 

Therefore, {xn} is a Cauchy sequence in M. Similarly, we can prove is also a 
Cauchy sequence. 

Because M is a completed multi-metric space, we have that 

lima;„ — liniy„ = z*. 

n n 

We prove z* is a fixed point of T in M. In fact, by pio(lima;„, lim?/„) = 0, there 
exists an integer N such that 

if n > + 1 . Whence, we know that 



^< pia{z*,Tz*) < Pi^{z*,Xn) + Pio{yn,Tz*) + Pi^{Xn,yn) 

< pi^{z*,Xn) + api^iUn-l, Z*) + Pio{Xn,yn)- 

Notice 

lim pi^{z*,Xn)= lim 2;*) = lim pio{xn,y„) ^ 0. 

We get that pi^{z%Tz*) = 0, i.e., r;^* = z\ 

For other chosen points uo,vo e Mi, we can also define recursively 

Un+l = TUn, Vn+1 = 

and get the hmit points hm Un — hm Vn — w* E , Tu* e Mj^ . Since 

PkXz*,u*) = pi^XTz\Tu*) < api^{z\u*) 

and < a < 1, there must be z* = u*. 

Similar consider the points in Mj, 2 < ^ < m, we get that 

1 <# $(T) < m. \\ 

Corollary 2.2(Banach) Let M he a metric space and T a contraction on M. Then 
T has just one fixed point. 



7 



3. Open problems for a multi-metric space 

On a classical notion, only one metric maybe considered in a space to ensure the 
same on all the times and on all the situations. Essentially, this notion is based on 
an assumption that spaces are homogeneous. In fact, it is not true in general. 

Multi-Metric spaces can be used to simplify or beautify geometrical figures and 
algebraic equations. One example is shown in Fig.l, in where the left elliptic curve is 
transformed to the right circle by changing the metric along x, y-axes and an elliptic 
equation 




to equation 

2,2 2 

X + y — r 

of a circle of radius r. 



Fig.l 

Generally, in a multi-metric space we can simplify a polynomial similar to the 
approach used in projective geometry. Whether this approach can be contributed to 

mathematics with metrics? 

Problem 3.1 Choose suitable metrics to simplify the equations of surfaces and 
curves in E^. 

Problem 3.2 Choose suitable metrics to simplify the knot problem. Whether can 
it be used for classifying dimensional manifolds? 

Problem 3.3 Construct multi-metric spaces or non-linear spaces by Banach spaces. 
Simplify equations or problems to linear problems. 



8 



References 



R. Abraham, J.E.Marsden and T.Ratiu, Manifolds, Tensor Analysis and Appli- 
cations, Addison- Wesley Publishing Company, Inc., 1983. 

S.S.Chern and W.H.Chern, Lectures in Differential Geometry, Peking University 
Press, 2001. 

Daniel Deleanu, A Dictionary of Smarandache Mathematics, Buxton University 
Press, London & New York,2004. 

J.M.Lee, Riemannian Manifolds, Springer- Verlag New York, Inc., 1997. 
L.F.Mao, Automorphism Groups of Maps, Surfaces and Smarandache Geome- 
tries, American Research Press, 2005. 

F. Smarandache, Mixed noneuclidean geometries, eprint arXiv: math/001011^^ 
10/2000. 

F. Smarandache, A Unifying Field in Logics. Neutrosopy: Neturosophic Proba- 
bility, Set, and Logic, American research Press, Rehoboth, 1999. 
F. Smarandache, Neutrosophy, a new Branch of Philosophy, Multi-Valued Logic, 
Vol.8, No. 3 (2002) (special issue on Neutrosophy and Neutrosophic Logic), 297- 
384. 

F. Smarandache, A Unifying Field in Logic: Neutrosophic Field, Multi-Valued 
Logic, Vol.8, No. 3 (2002) (special issue on Neutrosophy and Neutrosophic Logic), 
385-438. 



9