Professional Documents
Culture Documents
N n {1, 2, , n}
R ( X 1 , X 2 , , X n ) X 1 X 2 X n
i.e., | |
0 otherwise
○ Example 5.1 :
R ( X Y ) Z )Y1Y2Y3Y4 Z 2 Z1Z 3 Z 4 Z 5
R ({ X 1 , X 3} , X i J N nY X
1 1 1 1 0.8 0.8 0.8 0.8
{ X i | j J N n } R2
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,$), (Y ,*), (Y , s)}
( x) R ( y ) [ R { X i }](Y ) max R( x) Rij
x y
Y j | j J X X j
jJ
Y1 Dollar 1 0 1 0 0 Dollar 0 0 1
0 0
Y2 Pound 0 0 0 1 0 Pound 0 0 0
0 0
Y3 Franc 0 0 0 0 0 Franc 0 1 0
0 0
Y4 Mark 0 0 0 0 0 Mark 0 0 0 0
0
US Fran Can Brit Ger US
Fran Can Brit Ger
Z1 Z2 Z3 Z4 Z5 Z1 Z2 Z3 Z4
Z5
English
Franch
X1 X2
· Fuzzy Relations
tuples : ( x1 , x2 ,..., xn )
membership grade :
0 R ( x1 , x2 ,..., xn ) 1
· set family X = { X
i | i Nn}
Let X = xi | i N n X X i
jJ
Let Y = Yj | j J X X j
jJ
Where J Nn ,|J|=r
Y a subsequence of X , YX
iff Y j X j j J
Y= { X
i | j J Nn}
· Example 5.3 X1
={X,Y}, X ={a,b}, X ={*,$}
2
3
Let R = [ R { X , X }] ,
ij
i
j Ri [ R { X i }]
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 }
1
2
3
And R= R 1,2 y { X 1 , X 3}
∴ X-Y =
X3
= {*,$}
0.9 0.4 1 0.8
From example 5.3 R1,2
X , a X ,b Y, a Y ,b
∴[R X Y ] [R
1,2 { X 3}]
=
0.9 0.9 0.4 0.4 1 1 0.8 0.8
X , a,* X , a,$ 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 ] = [ R 2 { X 1 , X 3}]
y { X 2 }, X Y { X 1 , X 3} {(*,*), ( x,$), (Y ,*), (Y , s)}
{x,y} {x,$}
1 0.8
R= R 2
a b
∴[R
2 { 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
X Y :bipartite
graph
X Y :directed
graph
‧ Representations
i, matrices R [rij ] , where rij R ( xi , y j )
x1
.9 1 0 0 0
x2 0 .4 0 0 0
x3 0 0 1 .2 0
x4 0 0 0 0 .4
x5 0 0 0 0 .5
0 0 0 0 .2
x6
ii)
5-9
‧ Domain:dom R
Crisp – dom R = {x X | ( x, y) R, y Y }
Fuzzy – dom R(x) = max R ( x, y )
yY
‧ Range:ran R
Crisp – ran R = { y Y | ( x, y) R, x X }
Fuzzy – ran R(y) = max R( x, y )
xX
R 1 ( y, x ) R ( x, y )
R 1 R T , ( R 1 ) 1 R
0.3 0.2
e.g. R
0 1
0.6 0.4
0.3 0 0.6
R 1 R T
0.2 1 0.4
‧ 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
P。Q Q。P
Properties: 1 1
( P。Q ) Q 。P
( P。Q。
1
) R P。(Q。R )
Matric form: [r ] [ p ij ik ] [ q kj ]
。
max-product composition
matrix form [ rij ] [ pik 。
] [ q kj ]
‧ Example
0 .3 0 .5 0 .8 0 .9 0 .5 0 .7 0 .7
0 .0 0 .7 1 .0 0 .9
。0.3 0 .2 0 .0
0 .4 0 .6 0 .5
1.0 0 .0 0 .5 0 .5
0 .8 0 .3 0.5 0 .5
1.0 0.7
Max-min = 0.2 0.5
0 .5
0.4 0.5 0 .6
irrflexive
antiflexive
ii, symmetric
asymmetric
antisymmetric ( x, y ) R, ( y, x) R x y
strictly antisymmetric x y, ( x, y ) R or ( y, x) R
iii, transitive
nontransitive
antitransitive
◎ Fuzzy Relations
i, reflexive --- x , R ( x, x ) 1
irreflexive --- x , R ( x, x ) 1
antiflexive --- x , R ( x, x ) 1
-reflexive --- x , R ( x, x )
5-14
ii, symmetric -- x, y, R ( x, x) R( y, x)
asymmetric -- x , y , R ( x, x ) R ( y , x )
R ( x, x ) 0
antisymmetric --
R ( y , x) 0
x y
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
Antireflexive
Symmetric
Antisymmetric
Transitive
Reflexive
Crisp:
equivalence;
Fuzzy:
similarity
Quasi-
equivalence
Compatibilit
y or
Tolerance
Partial
ordering
Preordering
or
Quasi-
ordering
Strict
ordering
Figure3.6 Some important types of binary relation
R(X,X)
◎ transitive closure:
RT ( X )
1. R / R ( R R)
2. If R/ R , Let R/ R , go to step1
3. Stop, RT R /
Step2: R /
R , Let R R/
repeat step1
0.7 0.5 0.5 0.5
0.0 0.4 0.8 0.4
RR
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) R/
0.0 0.4 0.4 0.4
0.0 0.4 0.8 0.0
Step3: R /
R , Let R R/
repeat step1
0.7 0.5 0.5 0.5
0.0 0.4 0.8 1.0
R
'
R
0.0 0.4 0.4 0.4
0.0 0.4 0.8 0.4
Step4: Stop
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, ,10
R( X X )
{ ( x, y) | x, y have the same remainder when
divided by 3}
defined on X .
(2.1)(2.2))
If R: Similarity relation,
R : equivalence relation
( R) ( R) | 0,1
( R) : nested, i.e.,
( R) : a redefinement of ( R) iff
Prove that :A fuzzy relation R: X X is a similarity
relation, then
R is a equivalent relation
Pf : ∵ R : a similar relation
∴ R : reflexive, i.e., x X , R ( x, x ) 1
i,
R : reflexive
x X , R ( x, x) 1 [0,1], ( x, x) R
R : reflexive
ii,
R : symmetric
∵ R : symmetric
x, y Z , R ( x, y ) R ( y , x)
Let R ( x, y ) R ( y , x )
Then or
a, if => ( x, y ), ( y, x) R
b, if => ( x, y ), ( y, x) R
iii,
R : transitive
∵ R : transitive
x, z X 2 , R ( x, z ) max[ R( x, y ), R( y, z )]
yY
Let R ( x, y ) 1 , R ( y, z ) 2
Assume 1 2
Then 1 2 , 1 2 , or 1 2
R ( x , z ) , ( x, z ) R --- (B)
(A) , (B) =>
R : transitive
b. if 1 2
( x, y ) R, ( y , z ) R , don’t care ( x, z )
c. if 1 2
( x, y ) R, ( y , z ) R , don’t care ( x, z )
Crisp case :
Maximal compatibility classes – not properly
contained within any other compatibility class
Complete cover – all the maximal compatibility
classes
Fuzzy case :
α-compatibility class ---- a subset A of X ,
s.t.
x, y A, if R ( x, y ) R( x, y ) , R : fuzzy compatibility
relation
maximalα ---- compatibility classes
completeα-cover
0.5
0.7 0.9
b d
a 1 0.7 0 1 0.7
b 0 1 0 0.9 0
c 0.5 0.7 1 1 0.8
d 0 0 0 1 0
e 0 0.1 0 0.9 1
1 1 0 0 0 0 0 0 0
1 1 0 0 0 0 0 0 0
0 0 1 1 1 0 0 0 0
e.q. 0.4
0 0 1 1 1 1 1 0 0
0.5
R 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
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 ) R[ x ]y
xA
1 1 0 0 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
0.4
R 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
e.q. 0.5
1 1 0 0 00 0 0 0
1 1 0 0 00
0 0 0
0 0 1 1 10 0 0 0
0 0 1 1 10 1 1 0
0.5
R 0 0 1 1 10 1 1 1
0 0 0 1 10 1 0 0
0 0 0 1 10 0 1 0
0 0 0 0 10 0 0 1
0 0 0 0 0 0 0 0 1
X Y : X : predecessor
precedes Y : successor
˙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 the first member
partial ordering
5-28
˙ A X
If x X , and y A , x y ,
x: lower bound of A on X
If ……… , x y
x : upper bound of A on X
s.t. 1, 2,
U ( R, A)( x) >0 R(x,y) > 0 ,
y support [ U(R,A) ]
˙ Example 5.13
a b c d e
a 1 0.7 0 1 0.7
b 0 1 0 0.9 0
Fuzzy partial ordering R: c 0.5 0.7 1 1 0.8
d 0 0 0 1 0
e 0 0.1 0 0.9 1
4, LUB(A) =b
5-32
5. Crisp ordering captured by the fuzzy ordering
e.g. 0.5
1 1 0 1 1
0 1 0 1 0
R 1 1 1 1 1
0 0 0 1 0
0 0 0 1 1
# is → 2 3 1 5 3
※The ordering become weaken with the increasing
α
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
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 )]
※ It’s possible that a relation ( h( x1 ), h( x2 )) Q which
( x1 , x2 ) R .
※ 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
B {b1 , b2 , , bn } h
R,Q:fuzzy relations
h : strong homomorphism
max ( R (ai , b j )) Q ( y1 , y 2 )
iff i, j
y1 h(ai )ai A
where y 2 h(b j )b j B
5-35
‧Example 5.14
R(X,X)
0 0 .5 0 0
0 0 0.9 0
R
1 0 0 0.5
0 0 .6 0 0
Q(Y,Y)
0.5 0.9 0
Q
1 0 0.9
1
0.9 0
But Q ( , ) 0.9 R ( d , c) 0
Q(Y,Y)
0 .7 0 0 .9
0 .4 0 .8 0
1
0 1
Endomorphism : (subgraph)
h:X→Y, 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
:sup-i composition
i
P o Q( X Z )
i
[ P o Q]( X , Z ) sup i[ P( x, y ), Q ( y, z )]
yY
‧ Properties
1.
i i i i
( P o Q ) o R P o( Q o R )
i i
2. P o( Q j ) ( P o Q j )
j j
i i
3. P o( Q j ) ( P o Q j )
j j
i i
4. ( Pj ) o Q ( Pj o Q)
j j
i i
5. ( Pj ) o Q ( Pj o Q)
j j
6.
i i
( P o Q) 1 Q 1 o R 1
5-39
i i
Where and are fuzzy relations.
P( X , Y ) Q(Y , Z )
i
pf. From Eq.(5.13), i.e., P Q ( x, z ) sup i P( x, y ), Q( y, z )
yY
i
P ( Q j ) ( x, z ) sup i P( x, y ), Q j ( y, z )
jJ yY
jJ
Q Q j
Let jJ
Q Q1 , Q Q2 , , Q Q J
i.e., ( y, z ), Q( y, z ) Q1 ( y, z ), , Q( y, z ) Q J ( y, z )
i is monotonically increasing
i[ P ( x, y ), Q j ( y , z )] i[ P( x, y ), Q1 ( y, z )]
jJ
........... ( x, y )
i[ P ( x, y ), Q ( y , z )] i[ P( x, y ), Q ( y, z )]
j J
jJ
i[ P( x, y ), ( Q j )( y, z )] i[ P( x, y ), Q j ( y, z )]
jJ jJ
sup i[ P( x, y ), Q j ( y, z )], ( x, y ), ( y, z )
jJ yY
i i
P ( Q j ) ( x, z ) ( P Q j ) ( x, z ), ( x, z )
jJ jJ
i i
i,e., P ( Q j ) ( P Q j )
jJ jJ
5-40
。Sup-i composition monotonically increases
i i
P Q1 P Q2 (5.20)
i.e., i i if Q1 Q2
Q1 P Q2 P (5.21)
。Identity of i
1 0
1, x y
E ( x, y )
0, x y 0
1
i.e., i
EP PE P
i
2
。Relation R on X
: i-transitive
iff R ( x, z ) i R ( x, y ), R( y, z ) , x, y, z X
i
RR R
。i-transitive closure RT ( i )
proof:
i (n ) i (m )
(n ) i (m )
i, RT ( i ) RT ( i ) R R
n 1 m 1 n 1
m 1
R R R
n ,m 1
(nm)
R ( k ) R ( k ) RT (i )
k 2 k 1
i i
R (2) R R S S S
mathematical
i-transitive induction
If R(n) S ,
i i
R ( n 1) R R ( n ) S S S
R ( k ) S , k
RT (i ) R ( k ) S
k 1
i.e., RT ( i ) : smallest
R ( m ) R m 1
RT ( i ) R ( n 1) n n 1
m
R R
5-42
proof : i, R : reflexive,
i i
E R, R E R R R R (2)
proof: If x y , R ( n 1) ( x, x) 1
reflexive
If x y, i
Extension of definition
R ( n ) ( x, y ) sup i R( x, z1 ), R( z1 , z2 ), , R( zn 1 , y )
Z1 ,, Z n1
X n
X Z 0 , Z1 , , Z n y contains
x, y X , R ( n ) ( x, y ) R ( n 1) ( x, y ),
R ( n ) R ( n 1) ( B) R ( n ) R( n 1) ( A, B)
RT ( i ) R ( n 1)
5-43
5.10 INF- w Compositions of Fuzzy Relations
i
a b b
。 w operation:
i a b 1
。Theorem 5.3
1, i ( a, b) d iff wi (a, b) b
2, wi ( wi (a, b), b) a
6, wi (inf a j , b) sup wi (a j , b)
j j
7, wi (sup a j , b) inf wi (a j , b)
j j
ii, If b wi ( a, d ) i: continuous
monotone
d
i: monotone increasing
()
(3)
i (a, x )a wi (bb, d d) Byi ((1)=>
b, i (a, x)) d
Associativity
communitation
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 )
wi ( s, b) wi (a j , b), j
wi ( s, b) inf wi (a j , b)
j
---- (C)
inf wi (a j , b) wi (a j0 , b), j0 J
j
By(1)
By(1)
wi ( s, b) inf wi ( a j , b)
j
--- (D)
By(B)(C)(D)
wi (sup a j , b) wi ( s, b) inf wi ( a j , b)
j
j
5-45
(2)Show wi ( wi (a, b), b) a (Theorem 5.3 (2))
proof : 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 ) By Axiom i2 wi (a, b) b
i (b,1) By Axiom i2 ( a 1 )
b By Axiom i1
i ( wi (a, b), a) b
i, If ab
i ( wi (a, b), a) b
ii, W (d , a) W (d , b)
i i
ii, see i
Proof : ∵ if a b Wi (a, b) b
if a b Wi (a, b) 1
Wi ( a, d ) d
Wi ( a, d ) 1
Wi ( a, d ) 1
=> 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
B, if ab
inf Wi composition
Wi inf
( P Q)( x, z ) y Y Wi ( P ( x, y ), Q ( y , z ))
Theorem 5.4 :
i Wi Wi
(1)( P Q R) (Q P 1 R) ( P (Q R 1 ) 1 )
Wi Wi i Wi
(2)( P (Q S ) ( P Q) S
Theorem 5.5 :
Wi Wi
( Pj ) Q ( Pj Q)
j j
Wi Wi
( Pj ) Q) ( Pj Q)
j j
Wi Wi
P ( Q j ) ( Pj Q j )
j j
Wi Wi
P ( Q j ) ( Pj Q j )
j j
Theorem 5.6 : if Q1 Q2 => Wi
P Q1 P Q2
Wi
Wi Wi
Q1 R Q2 R
Proof : Q1 Q2 => Q1 Q2 Q1 , Q1 Q2 Q2
∵ (P Q ) (P Q ) P (Q Q ) P Q
Wi
1
Wi
2
Wi
1 2
Wi
=> Wi
P Q1 P Q2
Wi
∵ (Q R) (Q
1
Wi
2
Wi Wi
R ) (Q1 Q2 ) R Q2 R
Wi
=> Q R Q 1
Wi
2
Wi
R
Theorem 5.7 :
1. i Wi
P 1 ( P Q) Q
2. Wi
R P ( P 1 R)
i
3. Wi
P ( P Q) Q 1
Wi
4. Wi
R ( R Q 1 ) Q
Wi
Proof :
(1) P Q ( P
Wi
1 1
) Q
Wi
---- (A)
(5.26) (5.25)
i.e., Wi i
(Q P 1 R ) ( P Q R )
let
Wi
P QQ
,
P 1 P ,
QR
i
( A) P 1 ( P Q) Q
Let P 1 P , QR , i
P 1 R R
Wi i
( B ) R P ( P 1 R )
(3) by (5.33) , [ P
i Wi
1
( P Q)]1 Q 1
---- (C)
Wi i
( P Q )1 P Q 1
Let
Wi
P Q P, P Q, Q 1 R
Wi Wi
(C ) P ( P Q) 1 Q 1