Chapter 5 : Fuzzy Relation

5.1 Crisp and Fuzzy Relation

Crisp relation:
+ degrees or strengths of relation
Fuzzy relation
Cartesian product :
X i {( x1 ,
, xn ) | xi

X i , i N
n }

iN n

N n {1, 2,
, n}

n-ary relation: a subset of iN X i

n

R( X 1 , X 2 ,
, Xn)
i.e.,
$
|

a set

X 1 X 2

Xn
$
|

the universal set

Characteristic function:

R ( x1 , x1 ,
xn )

xn ) R

1 if ( x1 , x1 ,

otherwise
0

Binary, Ternary, Quaternary, Quinary, n-ary

relations

 Representation of a relation

R ( X 1 ,..., X n )
(ri1 ,i2 ,..,in )
ri1 ,i2 ,..,in

: n-D membership array

= 1

iff

( x1 ,..., xn ) R

otherwise

Example 5.1 :

R ( X Y ) Z )Y1Y2Y3Y4 Z 2 Z1Z 3 Z 4 Z 5

R ({ X 1 , X 3}
, Xi J

N nY

{ X i | j J N n } R2

1
1
1
1
0.8
0.8
0.8
0.8

X , a,* X , a,$ Y , b,* Y , a,$ X , b,* X , b,$ Y , b,* Y , b,$

y { X 1 , X 3}
Y j X j j J

[ R X Y ]:[ R2 { X 1 , X 3}]

y { X 2 }, X Y { X 1 , X 3 } {(*,*), ( x,$), (Y ,*), (Y , s)}

( x) R ( y ) [ R { X i }](Y ) max R( x) Rij

xf y

Y j | j J X X j
jJ

1 0.7 0.4 0.8

.
a,* a,$ b,* b,$
0.9
0.4
1
0.7
0.8

X , a,* X , b,* Y , a,* Y , a,$ Y , b,$

0.9
0.4
1
0.8
0.9
0
1
0.8
R1,2

R1,3

X , a X ,b Y , a Y ,b
X ,* X ,$ Y ,* Y ,$
1
0
0.6
0.9
0.7
0 R ( x1 , x2 ,..., xn ) 1

( NY , Beijing ) ( NY , NY ) ( NY , London) ( Paris, Beijing ) ( Paris, NY ) ( Paris

R2,3

={English , French} , Y ={ dollar , pound , franc ,

mark}
Z={US , France , Canada , Britain , Germary}

R( X Y Z )

={(English , dollar , US) , (French , franc ,

France)
(English , dollar , Canada) , (French ,
dollar , Canada)
(English , pound , Britain)}
Y1

Dollar 1 0 1 0 0

Dollar

0 1

Pound

0 0

0 0
Y2

Pound 0 0 0 1 0

0 0
Y3

Franc

0 0 0 0 0

Franc

1 0

0 0 0 0 0

Mark

0 0 0 0

0 0
Y4

Mark

0
US Fran Can Brit Ger

US

Fran Can Brit Ger

Z1

Z2

Z3

Z4

Z5

Z1

Z2

Z3

Z4

Z5

English
Franch
X1

X2

 Fuzzy Relations
Cartesian Product :
tuples :

X 1 X 2
Xn

( x1 , x2 ,..., xn )

membership grade :
0 R ( x1 , x2 ,..., xn ) 1

Example 5.2: Birary relation R : represents the

concept very far

= { New York , Paris}

={Beijing , New York , London}

Relation in list notation

R( X Y )

1
0
0.6
0.9
0.7
0.3

( NY , Beijing ) ( NY , NY ) ( NY , London) ( Paris, Beijing ) ( Paris, NY ) ( Paris, Londom)

Relation in membership array

NY
Beijing
NY
London

Paris
1

0.9
0.7

0.6

0.3

 Ordinary fuzzy relation

with valuation set [0,1]
L-fuzzy relation
With ordered valuation set L
5.2 Projection and Cyclindric Extensions

set family X = { X

| i Nn}

Let X =

xi | i N n X X i

Let Y =

Yj | j J X X j

Where

jJ

jJ

J Nn

,|J|=r

Y a subsequence of X ,
iff Y

YX

X j j J

Projection : [ R y] the projection of R on Y

R ( X 1 , X 2 ,
, Xn)

Y= { X

[ R y]

: a relation

| j J Nn}

: a fuzzy relation (set)

[ R y ](Y ) max R ( x)
xf y

max can be generalized by other t-conorms

Example 5.3

R( X 1 , X 2 , X 3 )

X1

0.9

0.4

0.7

0.8

= X , a,* X , b,* Y , a,* Y , a,$ Y , b,$

Let R = [ R { X , X }] ,
ij

R1,2

={X,Y}, X ={a,b}, X ={*,$}

Ri [ R { X i }]

0.9
0.4
1
0.8

X , a X ,b Y , a Y,b

R1,3

0.9
0
1
0.8

X ,* X ,$ Y ,* Y ,$

R2,3

1 0.7 0.4 0.8

a,* a,$ b,* b,$

R1

0.9 1

*
y

R2

1 0.8

a b

1 0.8
R3
* $

Cyclindric Extension

[R X Y ]

the CE of R into

X-Y
X-Y : sets

Xi

that are in X but are not in Y

[ R X Y ]( x) R ( y )

R: a relation defined on Y

Example 5.4 ( Refer to example 5.3)

Let X = { X , X , X }

And R= R

y { X 1 , X 3}

1,2

X-Y =

= {*,$}

X3

From example 5.3

[R X Y ] [R

1,2

{ X 3}]

0.9
0.4
1
0.8

X , a X ,b Y , a Y,b

R1,2

0.9
0.9
0.4
0.4
1
1
0.8
0.8

X , a,* X , a,$ X , b,* X , b,$ Y , a,$ Y , a,$ Y , b,* Y , b,$

[ R X Y ]:[ R12 { X 3 }] [ R13 { X 2 }] [ R23 { X 1}] [ R1 { X 2 , X 2 }]

[ R2 { X 1 , X 3}] [ R3 { X 1 , X 2 }]

Consider

[R X Y ]

[ R2 { X 1 , X 3}]

y { X 2 }, X Y { X 1 , X 3} {(*,*), ( x,$), (Y ,*), (Y , s)}

{x,y} {x,$}
R= R

1 0.8

a b

[ R2 { X 1 , X 3}]

1
1
1
1
0.8
0.8
0.8
0.8

X , a,* X , a,$ Y , b,* Y , a,$ X , b,* X , b,$ Y , b,* Y , b,$

5-7
Cyclindric closure
-A relation may be exactly reconstructed from
several of its projections by taking the set
intersection of their cyclindric extensions

Pi | i I :a

set of projections of a relation on X

cyl{Pi }( X ) min[ Pi X Yi ]( X ) R
iI

Yi

The family of sets on which

Pi

is defined.

Example

0.9 0.4 1 0.7 0.4 0.8

Cyl 1,2 1,3 RRR 2,3},,{
x, a x, b y , a y , a $ y , b y, b
Refer to the original relation

R( X , X , X )

in example

5.3.
It is not fully reconstructable from its projections
become of ignoramus of

R1

, R ,and
2

R3

5-8

5.3. Binary Relations R( X , Y )

X Y bipartite

graph
graph

X Y directed

Representations
i, matrices

R [ rij ]

, where

ii, sagittal diagrams

Examples
i)

y1

y3

y4

y5
0
0

x1
.9
x2 0

1
.4

0
0

0
0

x3

.2

x4
x5

0
0

0
0

0
0

0
0

.4
.5

.2

x6

ii)

y2

rij R ( xi , y j )

5-9
Domaindom R
Crisp dom R = {x X | ( x, y) R, y Y }
Fuzzy dom R(x) =

max R ( x, y )
yY

The domain of a fuzzy relation R(x,y) is a fuzzy set

on X; dom R(x) is its membership function.
e.g. dom R( X ) = max(0.9, 1) = 1
1

Rangeran R
Crisp ran R = { y Y | ( x, y) R, x X }
Fuzzy ran R(y) =

max R ( x, y )
xX

e.g. ran R( y ) = max(0.4, 0.5, 0.2) = 0.5

5

Height h( R) max max R( x, y)

yY

e.g.,

h( R ) 1

xX

normal fuzzy relation

5-10
Inverse R

(Y , X )

R 1 ( y, x ) R ( x, y )
R 1 R T , ( R 1 ) 1 R

e.g.

0.3
R 0
0.6

0.3
R 1 R T
0.2

0.2
1
0.4

0.6
0.4

0
1

Composition R( X , Z ) P( X , Y ) Q(Y , Z )
R ( x, z ) [ P Q]( x, z ) max min[ P ( x, y ), Q ( y , z )]
yY

Max-min composition
PQ QP

Properties

1
( PQ ) 1 Q
P 1
) R P(QR )
( PQ

Matric form [r ] [ p
ij

Where [r ] max min( p

ij

ik

ik

] [ qkj ]

, qkj )

5-11
R ( x, z ) [ PQ]( x, z ) max[ P ( x, y
) Q ( y , z )]
yY

max-product composition
matrix form

[ rij ] [ pik
] [ qkj ]

Where [r ] max( p
ij

ik

, qkj )

Example
0.3
0.0

0.4

0.5
0.7
0.6

0.8 0.9
1.0 0.3
0.5 1.0

Max-min =
Max-prod =

0.8
1 .0

0.5
0.8
1.0

0.5

0.5
0.2

0.7
0.0

0.0

0.5

0.3
0.2

0.5
0.5

0.4

0.5

0.7
0.9
0.5

0 .5
0.7
0.6

0.15
0.14

0.4
0.5

0.2

0.28

0.45
0.63

0.54

Relational join R( X , Y , Z ) P( X , Y ) * Q(Y , Z )

R ( x, y, z ) [ P * Q ]( x, y , z ) min[ P ( x, y ), Q ( y , z )]

The max-min composition can be obtained by

aggregating appropriate elements of the
corresponding join.

5-12
Example
[ PQ]( x, z ) max[ P * Q]( x, y, z )
yY

5.4Binary Relation on a Simple Set

Representations

5-14
characteristic Properties (Crisp case)
i,

reflexive
irrflexive

antiflexive
ii,

symmetric
asymmetric

antisymmetric

( x, y ) R, ( y, x) R x y

strictly antisymmetric
iii, transitive
nontransitive

x y, ( x, y ) R

or

( y, x) R

antitransitive

Fuzzy Relations
i, reflexive ---

irreflexive ---

R ( x, x ) 1

antiflexive ---

R ( x, x ) 1

-reflexive ---

R ( x, x )

R ( x, x ) 1

5-14
ii, symmetric
asymmetric

---

x, y , R ( x, x ) R ( y, x)

x , y ,

antisymmetric --

R ( x, x ) R ( y , x )

R ( x, x ) 0

R ( y, x) 0

iii, max-min transitive --

x y

x, z

R ( x, z ) max
min R( x, y ), R( y, z )

yY

max-product transitive --

x , z

R ( x, z ) max
min R( x, y ) R( y, z )

yY

nontransitive --

x , z

R ( x, z ) max
min R( x, y ), R( y, z )

yY

antitransitive --

x, z

R ( x, z ) max
min R( x, y ), R( y, z )

yY

Example 5.7 R:very near

reflexive, symmetric, nontransitive

5-15

Crisp:
equivalence;
Fuzzy:
similarity
Quasiequivalence
Compatibilit
y or
Tolerance
Partial
ordering
Preordering
or

Transitive

Antisymmetric

Symmetric

Antireflexive

Reflexive

Summary

Quasiordering
Strict
ordering
Figure3.6 Some important types of binary relation
R(X,X)
transitive closure:

RT ( X )

Algorithm for computing

1.

RT

R / R ( R oR )

2. If

R/ R

, Let

R/ R

, go to step1

3. Stop,

RT R /

Where

component-wise max

5-16
Example 5.8
0.7
0.0
R
0.0

0.0

Step1:

R oR

R ( R oR )

Step2: Q R

0.7
0.0
0.0
0.0
/

0.5
0.0
0.4
0.0
0.7
0.0
0.0
0.0
0.5
0.0
0.4
0.4

0.0
0.0
0.0
0.8
0.5
0.0
0.0
0.4
0.0
0.8
0.0
0.8

, Let

0.0
1.0
0.0

0.0
0.0
0.8
0.0
0.0
0.5
1.0
0.4

0.0

0.5
0.0
0.4

0.0
R/

R R/

repeat step1
0.7 0.5 0.5 0.5
0.0 0.4 0.8 0.4

R oR
0.0 0.4 0.4 0.4

0.0 0.4 0.4 0.4

0.7 0.5 0.5 0.5
0.0 0.4 0.8 1.0
R/
R ( R oR )
0.0 0.4 0.4 0.4

0.0 0.4 0.8 0.0

Step3: Q R

, Let

R R/

repeat step1
0.7
0.0
'
R
0.0

0.0

0.5
0.4
0.4
0.4

Step4: Stop

0.5
0.8
0.4
0.8

0.5
1.0
R
0.4

0.4

RT R /

5-17
5.5 Fuzzy Equivalence Relation
Crisp binary relation
equivalence: reflexive, symmetric, and
transitive
equivalence classes
partition: X/R
Example 5.9:

X 1, 2,L ,10

R( X X )

{ ( x, y) | x, y have the same remainder when

divided by 3}

R: reflexive, symmetric, transitive

equivalence
partition

X / R (1, 4, 7,10), (2,5,8), (3, 6,9)

5-18
Fuzzy Binary Relation
Fuzzy
Similarity relation

equivalence relation

Similarity classes

equivalence classes

2 Interpretations of a similarity relation:

1.Group similar elements into crisp classes
whose members are similar to each other
to some specified degree.

2.

x X ,

associate a fuzzy set

Ax

defined on X .

a fuzzy relation

U g R (Theorem 2.5, Eqs.

[0,1]

(2.1)(2.2))
If R: Similarity relation,
R :

Let

equivalence relation
( R) :

the partition of

w.r.t.

( R ) ( R) | 0,1
( R) :

nested, i.e.,

( R) :

a redefinement of

( R)

iff

Prove that :A fuzzy relation

relation, then

R: X X

is a similarity

is a equivalent relation

Pf : R : a similar relation
R : reflexive, i.e.,
symmetric, i.e.,
transitive, i.e.,
i,

x X , R ( x, x) 1
x, y X , R ( x, y ) R ( y, x )

x, z X 2 , R( x, z ) max[ R ( x, y ), R ( y, z )]

: reflexive

Q x X , R ( x, x) 1 [0,1], ( x, x) R

ii,

: reflexive

: symmetric

R : symmetric
x, y Z , R ( x, y ) R ( y, x)

Let

R ( x, y ) R ( y , x )

Then

iii,

or

a, if

=>

( x, y ), ( y, x) R

b, if

=>

( x, y ), ( y, x) R

: transitive

R : transitive

yY

x, z X 2 , R( x, z ) max[ R ( x, y ), R ( y, z )]
yY

Let

R ( x, y ) 1

Assume

R ( y, z ) 2

1 2

1 2

Then

a. if

=>

, or

1 2

( x, y ) R , ( y , z ) R

--- (A)

Q R ( x, z ) max[ R ( x, y ), R ( y, z )] min[ 1 , 2 ] 1
yY

R ( x , z ) , ( x, z ) R

(A) , (B) =>

b. if

--- (B)

R : transitive

1 2

( x, y ) R, ( y, z ) R

c. if

, dont care

( x, z )

, dont care

( x, z )

1 2

( x, y ) R, ( y, z ) R

Example 5.10 :

R( X , X )

: a fuzzy relation

R : reflexive , symmetric , transitive (

Q R ' R ( R oR ) R

level set :

R {0.0, 0.4, 0.5, 0.8, 0.9,1.0}

There are five nested partition

's

The similarity class for each element is a fuzzy

set defined by the row of the membership matrix
corresponds to that element
Example : see Example 5.10
For c :

0 0 1 0 1 0.9 0.5

a b c d e
f
g

For e :

0 0 1 0 1 0.9 0.5

a b c d e
f
g

c and e are similar at any level

5.6 Compatibility Relations ---- reflexive ,

symmetric
compatibility
Alternatives :

tolerance

proximity
Crisp case :

relation

Maximal compatibility classes not properly

contained within any other compatibility class
Complete cover all the maximal compatibility
classes
Fuzzy case :
-compatibility class ---- a subset A of X ,
s.t.
x, y A, if R ( x, y ) R ( x, y ) , R :

fuzzy compatibility

relation
maximal ---- compatibility classes
complete-cover
Example 5.11 :

R( X , X )

: a fuzzy relation

R : reflexive , symmetric
a compatibility relation

{0.0, 0.4, 0.5, 0.7, 0.8,1.0}

=> the complete-covers

5-25
0.5
0.7 0.9

b
d
a 1 0.7
b 0
1
c 0.5 0.7

d 0
0
e 0 0.1

e.q.

1
1

0
0.5
R 0

0
0

0
0

1
1
0
0
0
0
0
0
0

0 1 0.7
0 0.9 0
1 1 0.8

0 1
0
0 0.9 1
0 0 0 0
0 0 0 0
1 1 1 0
1 1 1 1
1 1 1 1
0 1 1 1
0 1 1 0
0 0 1 0
0 0 0 0

0
0
0
1
1
0
1
0
0

0
0
0
0
1
0
0
1
0

0
0
0

0
0

0
0

0
1

x y xx yy xx X ( x, y ) XS {x1 , x2 } Xy Ax y
(x,y)y X ( x y, or y xAx XA Xx yR[ x ] ( y ) R ( y , x )
x U ( R, A)( x ) I R[ x ]y
xA

1
1

0
0.4
R 0

0
0

0
0

1
1
0
0
0
0
0
0
0

0
0
1
1
1
0
0
0
0

0
0
1
1
1
1
1
0
0

0
0
1
1
1
1
1
1
0

0
0
0
1
1
1
1
0
0

0
0
0
1
1
1
1
0
0

0
0
0
0
1
0
0
1
0

0
0
0

0
0

0
0

0
1

Look for complete subgraphs

0.4

(1,2) , (3,4,5),(4,5,6,7),(5,8),(9)
(34,),(4,5,6),(4,5,7),(3,5),(5,6)
(4,5),(5,6,7),(4,6,7),(4,6),(6,7)

maximal compatible classes (the complete 0.4-

cover):
(1,2),(3,4,5),(4,5,6,7),(5,8),(9)
These do not partition X.

5-26
e.q.

0.5

1
1

0
0.5
R 0

0
0

0
0

1
1
0
0
0
0
0
0
0

0
0
1
1
1
0
0
0
0

0
0
1
1
1
1
1
0
0

0
0
1
1
1
1
1
1
0

0
0
0
1
1
1
0
0
0

0
0
0
1
1
0
1
0
0

0
0
0
0
1
0
0
1
0

0
0
0

0
0

0
0

0
1

maximal compatible classes

( The complete 0.5-cover)

(1,2),(3,4,5),(4,5,6),(4,5,7),(5,8),(9)

5-27
5.7. Ordering Relations
partial ordering: reflexive , antisummetric ,
transitive

X Y

: X : predecessor

precedes Y : successor
if exist

First member : if

x y y X

(minimum)

unique

Last member : if

y x y X

(maximam)

may not Minimal member : if

yx x y

be unique Maximal member : if

x y x y

properties :
1, if , at most one first member
if , at most one last member
2, There may be several maximal and minimal
member

3, if a first member X ,

only one minimal

member Y exists and x=y

4, if a last member x ,

only one maximal

member Y exists and x=y.

5, partial

the first member

the last member

inverse
ordering

the last member

partial ordering

the first member

5-28
In a partial ordering , it does not guarantee that
,

(x,y)

If ,

( x y, or y x

).

(x,y) : comparable (total ordering)

Otherwise (x,y) : non comparable

A X

If

x X

, and

x: lower bound of A on X

If ,

y A

x y

x y

x : upper bound of A on X

greatest lower bound ( or infimum ) GLB

- a lower bound which succeeds every other lower
bound
Least upper bound ( or supermum ) LUB
- a upper bound which preceeds every other upper
bound

Lattice A partial ordering on X contains GLB

and LUB ,

S {x1 , x2 } X

5-29
Connected a partial ordering is said to be
connected
iff

x, y X

x y

x<y or y>x

Linear ordering (total ordering , simple ordering ,

complete ordering )
- when a partial ordering is connected , then

( x, y )

comparable
Hasse diagrams representing partial orderings in
which

indicates

Example 5.12 : Crisp partial orderings

5-30
Fuzzy partial ordering
- reflexive , antisymmetric , and transitive under
some form of transitivity.
any fuzzy partial ordering can be resolved into a
series of crisp partial ordering .
i.e. taking a series of cut that produce increasing
levels of refinement
In a fuzzy partial ordering , R
x X
R[ x ]

, two fuzzy sets are associated with

: dominating class
R[ x ] ( y ) R ( x, y )

R[ x ]

: dominated class
R[ x ] ( y ) R ( y , x )

5-31
x undominated iff R(x,y) = 0

y x

X undominating iff R(y,x) = 0

Fuzzy upper bound for

A X

y x

is a fuzzy set

U ( R, A) I R[ x ]
xA

If a least upper bound of A exists , it is the

unique element
s.t. 1,

2,

U ( R, A)( x)
y

x U ( R, A)

>0

R(x,y) > 0 ,

support [ U(R,A) ]

Example 5.13
a b c d
Fuzzy partial ordering

R:

e
a
b
c

d
e

1 0.7 0 1 0.7
0
1 0 0.9 0
0.5 0.7 1 1 0.8

0
0 0 1
0
0 0.1 0 0.9 1

1. row : dominating class for each element

column : dominated class for each element

2. d : undominated , C : undominating
3. For A = {a,b} , U(R,A) = the intersection of
The dominating classes of a and b =
4, LUB(A) =b

0.7 0.9

b
d

5-32
5. Crisp ordering captured by the fuzzy ordering
e.g. 0.5
1
0

R 1

0
0

1
1
1
0
0

0
0
1
0
0

1 1
1 0
1 1

1 0
1 1

# is 2 3 1 5 3
The ordering become weaken with the increasing

5-33
Fuzzy preordering reflexive and transitive
Fuzzy weak ordering
i, an ordering satisfying the proportion of a fuzzy
total ordering except antisymmetry.
ii, a fuzzy preordering in which

x y

, either

R(x,y)>0 or R(y,x)>0
Fuzzy strict ordering
Antireflexive
Antisymmetric
Transitive
5.8. Morphisms
Crisp homomorphism h from (X,R) to (Y,Q)
Where R(X,X), Q(Y,Y)binary relations
( x1 , x2 ) R ( h( x1 ), h( x2 )) Q

Fuzzy homomorphism h
If R(X,X), Q(Y,Y)Fuzzy binary relations
And

R ( x1 , x2 ) Q[h( x1 ), h( x2 )]

 Its possible that a relation

( x1 , x2 ) R

( h( x1 ), h( x2 )) Q

which

If this is never the case h is called a strong

homomorphism.

5-34
Crisp strong homomorphism h
If

( x1 , x2 ) R (h( x1 ), h( x2 )) Q

And

( y1 , y 2 ) Q (h 1 ( y1 ), h 1 ( y 2 )) R

where h : many to one

h 1 ( y )

Xs
Fuzzy strong homomorphism h
H imposes a partition

on X

Let
A {a1 , a 2 , , a n }
B {b1 , b2 , , bn } h

R,Qfuzzy relations
h : strong homomorphism
iff

max ( R( ai , b j )) Q( y1 , y 2 )
i, j

y1 h( ai )ai A
y 2 h(b j )b j B

where

contains a set of

5-35
Example 5.14
R(X,X)
0
0
R
1

0 .5
0
0
0 .6

0
0 .9
0
0

0
0
0 .5

Q(Y,Y)
0.5
Q 1
1

0.9
0
0.9

0
0.9
0

hordinary fuzzy homomorphism (one way)

R ( x1 , x2 ) Q (h( x1 ), h( x2 )) strong

But
i,e,

Q ( , ) 0.9 R ( d , c ) 0
( , ) Q

while

(d , c) R

where

h ( d ) , h (c )

5-36
R(X,X)
0.8
0

0
0

0 .4

0.5
0
0.5
0
0

0
0.3
0
0
0

0 .7
0
0
1
0

0
0
0 .9
0
1

0
0
0

0.5
0

0.8

Q(Y,Y)
0.7
0.4

0
0.8
0

0.9
0
1

hstrong fuzzy homomorphism (two way)

5-37
Q represents a simplification of R
Isomorphism : (congruence)
h:1-1, onto

X Y

Endomorphism : (subgraph)
h:XY, Y X
Automorphism :
Isomorphism and End Endomorphism
i.e.m X=Y nad R=Q

5-38
5.9 SUP-i Compositions of Fuzzy Relations
Generalize max-min Composition
i : t-norm
sup : t-conorm
P(X,Y), Q(Y,Z)fuzzy relations
i

P o Q( X Z )

:sup-i composition

[ P o Q]( X , Z ) sup i[ P ( x, y ), Q ( y , z )]
yY

Properties
1.

( P o Q) o R P o( Q o R)

2.

P o ( Q j ) ( P o Q j )

3.

P o ( Q j ) ( P o Q j )

4.

( Pj ) o Q ( Pj o Q)

5.

( Pj ) o Q ( Pj o Q)

6.

( P o Q ) 1 Q 1 o R 1

5-39
3. Show Eq.(5.16), i.e.,
Where

P( X , Y )

and

P o( I Q j ) I ( P oQ j ),
jJ

Q(Y , Z )

pf. From Eq.(5.13), i.e.,

jJ

are fuzzy relations.

P
o
Q ( x, z ) sup i P ( x, y ), Q( y, z )

yY

P o( I Q j ) ( x, z ) sup i P ( x, y ), I Q j ( y, z )

jJ
jJ

yY

Let

Q I Qj
jJ

Q Q1 , Q Q2 ,L , Q Q J

i.e.,

( y, z ), Q( y, z ) Q1 ( y, z ),L , Q( y, z ) Q J ( y, z )

i is monotonically increasing

i[ P ( x, y ), I Q j ( y , z )] i[ P( x, y ), Q1 ( y , z )]

jJ

...........
i[ P ( x, y ), Q ( y , z )] i[ P( x, y ), Q ( y, z )]
IjJ j
J

( x, y )

i[ P ( x, y ), ( I Q j )( y, z )] I i[ P( x, y ), Q j ( y, z )]
jJ

jJ

sup i[ P( x, y ), ( I Q j )( y, z )] sup I i[ P( x, y), Q j ( y, z )]

jJ

yY

yY jJ

I sup i[ P( x, y ), Q j ( y, z )], ( x, y ), ( y, z )

jJ yY

P o( I Q j ) ( x, z )

jJ

i,e.,

i
( P oQ j ) ( x, z ), ( x, z )
jJ

P o( I Q j ) I ( P oQ j )
jJ

jJ

5-40
Sup-i composition monotonically increases
i

i.e.,

P oQ1 P oQ2 (5.20)

i

Q1 oP Q2 oP (5.21)

Identity of

E oP P oE P

Relation R on
iff

Q1 Q2

1, x y
E ( x, y )
0, x y

i.e.,

if

: i-transitive

R ( x, z ) i R ( x, y ), R ( y, z ) , x, y, z X

R oR R

i-transitive closure

RT (i )

--- The smallest i-transitive relation containing R

Theorem 5.1: R: any fuzzy relation

RT ( i ) U R ( n )
n 1

R ( n ) R oR ( n 1)

, where

5-41

By (5.15) (5.17)

proof:
i,

RT (i ) oRT (i )

i
U R o
n 1

(n)

k 2

k 1

(m)
R U
U
m 1
n 1

m 1

(n )

oR U R ( n m )

n , m 1
(m)

U R ( k ) U R ( k ) RT (i )

i.e.,

RT (i )

: i-transitive (Q R

T (i )

oRT (i ) RT ( i )

(5.20)(5.21) monotonically increasing

ii, Let S: i-transitive,

i

RS

R (2) R oR S oS S

If

i-transitive

(n)

mathematical
induction

S,
i

R ( n 1) R oR ( n ) S oS S
R ( k ) S , k

RT ( i ) U R ( k ) S
k 1

i.e.,

RT (i )

: smallest

Theorem 5.2: R: reflexive fuzzy relation on

X n

R ( m ) R m 1
RT ( i ) R ( n 1) n
m
n 1
R

5-42
proof : i, Q R : reflexive,
i

E R, R E oR R oR R (2)

(By repetition)

R ( n 1) R ( n )

ii, show

R ( n 1) R ( n )

proof: If
If

x y, R ( n 1) ( x, x ) 1
reflexive

x y,
Extension of definition
i

R ( n ) ( x, y ) sup i R( x, z1 ), R( z1 , z 2 ),L , R( z n 1 , y )
Z1 ,L , Z n1

Q X n

X Z 0 , Z1 ,L , Z n y

contains

at least 2 identical element.

Say

Z r Z s (r s )

i R ( x, z1 ),L , R ( zr 1 , zr ), L , R ( zs , z s 1 ),L , R ( zn 1 , y ) R ( k ) ( x, y ), (k n 1)

x, y X , R ( n ) ( x, y ) R ( n 1) ( x, y ),

R ( n ) R ( n 1) ( B )
RT ( i ) R ( n 1)

R ( n ) R ( n 1) ( A, B )

5-43
5.10 INF- w Compositions of Fuzzy Relations
i

a b b
a b 1

w operation:
i

wi ( a, b) sup x [0,1] | i( a, x) b

where

a, b [0,1]

, i : continuous t-norm

If i : logical conjunction (i.e., , and)

wi

: logical implication (i.e.,

Theorem 5.3
1,

i ( a, b) d

iff

2,

wi ( wi (a, b), b) a

3,

wi (i ( a, b), d ) wi (a, wi (b, d ))

4,

a b, wi (a, d ) wi (b, d )

wi (a, b) b

wi (d , a ) wi (d , b)

---- i

--- ii

5,

i ( wi ( a, b), wi (b, d )) wi (a, d )

6,

wi (inf a j , b) sup wi ( a j , b)

7,

wi (sup a j , b) inf wi (a j , b)

8,

wi (b,sup a j ) sup wi (b, a j )

9,

wi (b,inf a j ) inf wi (b, a j )

, if then)

10,

i (a, wi (a, b)) b

5-44
proof: (1) i , If

i ( a, b) d , b x | i ( a, x ) d

b sup x | i (a, x) d wi (a, d )

( )

ii, If

b wi (a, d )

i: continuous monotone

i: monotone increasing

()

By (1)<=
bi (a, b) abid(a, w
i ( a, d )) i ( a,sup x | i ( a, x ) d ) sup i ( a, x ) | i ( a, x ) d d

(3)
Q i ( a, xa) wbi (b, dd )
i(b, i( a, x)) d
By (1)=>
Associativity
communitation

i (i (a, b), x ) d x wi (i ( a, b)d )

By (A)

wi (i ( a, b), d )

wi (a, wi (b, d )) sup x | i (a, x ) wi (b, d ) sup x | x wi (i( a, b), d ) wi (i( a, b), d )

(7) Let
By(4)

S sup a j
j

---(B)

a j s, j

wi ( s, b) wi (a j , b), j
wi ( s, b) inf wi (a j , b)
j

---- (C)

Q inf wi (a j , b) wi (a j0 , b), j0 J

By(1) j

i (a j0 ,inf wi (a j , b)) b, j0
j

i ( s,inf wi (a j , b)) sup i(a j0 ,inf wi (a j , b)) b

j

j0

By(1)

wi ( s, b) inf wi (a j , b)
j

By(B)(C)(D)

--- (D)

 wi (sup a j , b) wi ( s, b) inf wi ( a j , b)
j

5-45
(2)Show

wi ( wi (a, b), b) a

(Theorem 5.3 (2))

proof : Q w (a, b) Sup x [0,1] | i(a, x) b and by Theorem

i

3.10
imin (a, b) i (a, b) min( a, b)

i, If a>b
wi ( a, b) Sup x [0,1] | i (a, x ) b Sup x [0,1] | min(a, x ) b b
i ( wi ( a, b), a )

i (b, a )
i (b,1)
b

By Axiom i2 wi (a, b) b
By Axiom i2 ( a 1 )
By Axiom i1

i ( wi (a, b), a ) b

i, If

ab

wi (a, b) Sup x [0,1] | i(a, x) b Sup x [0,1] | min( a, x) b 1

i ( wi ( a, b), a ) i ( wi (a, b), b)
i (1, a)
i (b,1)
b

By Axiom i2 wi (a, b) 1
By Axiom i2
By Axiom i1

i ( wi (a, b), a ) b

By Theorem 5.3 property 1(i.e.,

wi ( a, d ) b

i ( wi ( a, b), a ) b wi ( wi (a, b), b) a

i ( a, b) d

iff

(4) prove Theorem 5.3 (4) :

ab

=> i, W (a, d ) W (b, d )

ii, W (d , a) W (d , b)
i

proof : i, W (a, d ) W (b, d )

i

Wi (a, d ) sup{x | i( a, x) d }

---- (A)

Wi (b, d ) sup{x | i (b, x) d }

---- (B)

a, if

d ab

=> (A)=d , (B)=d ,

(A)=(B) ----- (1)

b, if

ad b

=> (A)=1 , (B)=d

(A) (B) ----- (2)

c, if

abd

=> (A)=1 , (B)=1

(A)= (B) ----- (3)

(1),(2),(3) => (A) (B)
i.e., W (a, d ) W (b, d )
i

ii, see i
5. show

i (Wi (a, d ),Wi (b, d )) Wi (a, d )

Proof : if
if
A, if

a b Wi (a, b) b

a b Wi (a, b) 1
a b i (Wi (a, b),Wi (b, d )) i (b, Wi (b, d ))

=>

i (b, Wi (b, d )) i (b, d ) min(b, d ) d

Wi (a, d ) d

=>

i (b, Wi (b, d )) i (b,1) b

Wi (a, d ) 1

=>

i (b, Wi (b, d )) i (b,1) b

Wi (a, d ) 1

B, if

a b i (Wi (a, b),Wi (b, d )) i (1, Wi (b, d )) Wi (b, d )

=>

Wi (b, d ) d
Wi (a, d ) d

Wi (b, d ) d
Wi (a, d ) 1

Wi (b, d ) 1
Wi (a, d ) 1

10. show

i (a, Wi (a, b) b

Proof :

a b Wi (a, b) b

a b Wi (a, b) 1

A, if

ab

i (a, Wi ( a, b)) i (a, b) min(a, b) b

B, if

ab

i (a, Wi ( a, b)) i (a,1) a b

inf W composition
i

Wi

inf

( P oQ)( x, z ) y Y Wi ( P ( x, y ), Q( y, z ))

Theorem 5.4 :
Wi

Wi

(1)( P oQ R ) (Q P 1 o R ) ( P (Q o R 1 ) 1 )
Wi

Wi

Wi

(2)( P o(Q o S ) ( P oQ) o S

Theorem 5.5 :
Wi

Wi

( Pj ) oQ ( Pj oQ)
j

Wi

Wi

( Pj ) oQ) ( Pj oQ)
j

Wi

Wi

P o( Q j ) ( Pj oQ j )
j

Wi

Wi

P o( Q j ) ( Pj oQ j )
j

 Theorem 5.6 : if

=>

Q1 Q2
Wi

Wi

Wi

P oQ1 P oQ2

Wi

Q1 o R Q2 o R

Proof :

=>

Q1 Q2

Q1 Q2 Q1

Q1 Q2 Q2

( P oQ ) ( P oQ ) P o(Q Q ) P oQ
Wi

Wi

Wi

=>

Wi

Wi

Wi

P oQ1 P oQ2

(Q o R) (Q
Wi

Wi

Wi

Wi

Wi

oR

Theorem 5.7 :
1.

P 1 o( P oQ) Q

2.

R P o( P 1 oR)

3.

P ( P oQ) oQ 1

4.

R ( R oQ 1 ) oQ

Wi

Wi

Wi

Wi

Wi

Wi

Proof :
(1) Q P oQ ( P
Wi

1 1

Wi

) oQ

---- (A)

Q (5.26) (5.25)

i.e.,
let

r r Wi r
ri r r
(Q P 1 o R) ( P oQ R )

r
P oQ Q
Wi

Wi

o R ) (Q1 Q2 ) o R Q2 o R

=> Q o R Q

r
P 1 P

r
QR

( A) P 1 ( P oQ) Q

(2)

Q P 1 oR P 1 oR

Let

r
P 1 P

---- (B)

r
QR
Wi

i
r
P 1 oR R

( B) R P o( P 1 oR )

(3) by (5.33) , [ P
Wi

( P oQ) 1 oP Q 1

Let

Wi

o( P oQ)]1 Q 1

---- (C)

Wi
r
r
r
P oQ P, P Q, Q 1 R
Wi

Wi

(C ) P ( P oQ ) 1 oQ 1

(4) follows (3)