Chapter2 Preference Orders and Utility Functions for countable sets

http://blog.sina.com.cn/s/blog_642075770100u0np.html

Choices, Values, and Frames

Notations:
X X X : a countable set, i.e. finite or denumerable.
≺ \prec : strict preference
∼ \sim : the absence of strict preference

Main Theorems:

Theorem 2.1. Let X X X be a countable set and ≺ \prec on X X X is a weak order, then numbers u ( x ) , u ( y ) , ⋯ u(x),u(y),\cdots u(x),u(y), can be assigned to the elements x , y , ⋯ x,y,\cdots x,y, in X X X in such a way that x ≺ y ⇔ u ( x ) &lt; u ( y ) x\prec y \Leftrightarrow u(x)&lt;u(y) xyu(x)<u(y) holds.

Theorem 2.2. Let X X X be a countable set and ≺ \prec on X X X is a strict partial order, then there is a real-valued function u u u on X X X such that x ≺ y ⇔ u ( x ) &lt; u ( y ) x\prec y \Leftrightarrow u(x)&lt;u(y) xyu(x)<u(y) holds.

2.1 Binary Relations

A binary relation on a set Y Y Y is a set of ordered pairs ( x , y ) (x,y) (x,y) with x ∈ Y x\in Y xY and y ∈ Y y\in Y yY. x R y x R y xRy means ( x , y ) ∈ R (x,y)\in R (x,y)R, where R R R is a binary relation on Y Y Y.

Some Relations Properties. A binary relation R R R on a set Y Y Y is
p 1. p1. p1. reflexive if x R x xRx xRx for all x ∈ Y x\in Y xY,
p 2. p2. p2. irreflexive if not x R x xRx xRx for all x ∈ Y x\in Y xY,
p 3. p3. p3. symmetric if x R y ⇒ y R x xRy\Rightarrow yRx xRyyRx for all x , y ∈ Y x,y\in Y x,yY,
p 4. p4. p4. asymmetric if x R y ⇒ xRy\Rightarrow xRy not y R x yRx yRx for all x , y ∈ Y x,y\in Y x,yY,
p 5. p5. p5. antisymmetric if ( x R y , y R x ) ⇒ x = y (xRy, yRx)\Rightarrow x=y (xRy,yRx)x=y for all x , y ∈ Y x,y\in Y x,yY,

p 6. p6. p6. transitive if ( x R y , y R z ) ⇒ x R z (xRy, yRz)\Rightarrow xRz (xRy,yRz)xRz for all x , y , z ∈ Y x,y,z\in Y x,y,zY,
p 7. p7. p7. negatively transitive if (not x R y xRy xRy, not y R z yRz yRz) ⇒ \Rightarrow not x R z xRz xRz for all x , y , z ∈ Y x,y,z\in Y x,y,zY,
p 8. p8. p8. connected or complete if x R y xRy xRy or y R x yRx yRx (possibly both) for all x , y ∈ Y x,y\in Y x,yY,
p 9. p9. p9. weakly connected if x ≠ y ⇒ x\neq y\Rightarrow x̸=y ( x R y xRy xRy or y R x yRx yRx) throughout Y Y Y. \qquad

Lemma 2.1. R R R is negatively transitive if and only if for all x , y , z ∈ Y x,y,z\in Y x,y,zY, x R y ⇒ ( x R z &ThickSpace; o r &ThickSpace; z R y ) x R y \Rightarrow (xRz\;or \;zRy) xRy(xRzorzRy).

2.2 PREFERENCE AS A WEAK ORDER

DEFINITION 2.1 A binary relation R R R on a set Y Y Y is
(1) a weak order ⇔ R \Leftrightarrow R R on Y Y Y is asymmetric and negatively transitive;
(2) a strict order ⇔ R \Leftrightarrow R R on Y Y Y is weakly connected weak order;
(3) an equivalence ⇔ R \Leftrightarrow R R on Y Y Y is reflexive, symmetric, transitive.

Example: the relation < on real numbers is a weak order and also a strict order, = is an equivalence.

Let R ( x ) = { y : y ∈ Y &ThickSpace; and &ThickSpace; y R x } R(x)=\{y:y\in Y\;\text{and}\; yRx\} R(x)={y:yYandyRx}If R R R is an equivalence, then R ( x ) R(x) R(x) is the equivalence classes generated by x x x. In this case R ( x ) = R ( y ) R(x)=R(y) R(x)=R(y) if and only if x = y x=y x=y. We denote the equivalence classes on Y Y Y under R R R by Y / R Y/R Y/R.

≺ \prec : strict preference, x ≺ y x \prec y xy read as " x x x is less preferred than y y y, or y y y is preferred to x x x"
∼ \sim : the absence of strict preference, x ∼ y ⇔ ( not &ThickSpace; x ≺ y , not &ThickSpace; y ≺ x ) x\sim y \Leftrightarrow (\text{not}\;x\prec y, \text{not}\;y\prec x ) xy(notxy,notyx)
⪯ \preceq : preference-indifference, x ⪯ y ⇔ x ≺ y &ThickSpace; or &ThickSpace; x ∼ y x\preceq y\Leftrightarrow x\prec y\;\text{or}\;x\sim y xyxyorxy

THEOREM 2.1. Suppose ≺ \prec on X X X is a weak order, being asymmetric and negatively transitive. Then
(1) exactly one of x ≺ y , y ≺ x , x ∼ y x\prec y, y\prec x, x\sim y xy,yx,xy holds for each x , y ∈ X x,y\in X x,yX;
(2) ≺ \prec is transitive;
(3) ∼ \sim is an equivalence (reflexive, symmetric, transitive);
(4) ( x ≺ y , y ∼ z ) ⇒ x ≺ z (x\prec y, y\sim z)\Rightarrow x\prec z (xy,yz)xz, and ( x ∼ y , y ≺ z ) ⇒ x ≺ z (x\sim y, y\prec z)\Rightarrow x\prec z (xy,yz)xz;
(5) with ≺ ′ \prec&#x27; on X / ∼ X/\sim X/ (the equivalence classes on X X X under ∼ \sim ) defined by
a ≺ ′ b ⇔ x ≺ y &ThickSpace;&ThickSpace; for some &ThickSpace; x ∈ a &ThickSpace; and &ThickSpace; y ∈ b ( 2.4 ) a\prec&#x27;b \Leftrightarrow x\prec y\;\;\text{for some}\;x\in a\;\text{and}\;y\in b\qquad\qquad\qquad(2.4) abxyfor somexaandyb(2.4) ≺ ′ \prec&#x27; on X / ∼ X/\sim X/ is a strict order.

THEOREM 2.2. If ≺ \prec on X X X is a weak order and X / ∼ X/\sim X/ is countable then there is a real-valued function u u u on X X X such that
x ≺ y ⇔ u ( x ) &lt; u ( y ) &ThickSpace;&ThickSpace; for all &ThickSpace; x , y ∈ X ( 2.5 ) x\prec y\Leftrightarrow u(x)&lt;u(y)\;\;\text{for all}\;x,y\in X\qquad\qquad\qquad (2.5) xyu(x)<u(y)for allx,yX(2.5)Outline of the proof. Only consider the case X / ≺ X/{\prec} X/ is denumerable, and the elements be enumerated as a 1 , a 2 , ⋯ a_1,a_2,\cdots a1,a2,. First we define a real-valued function u u u on X / ≺ X/{\prec} X/ by induction.
&ThickSpace;&ThickSpace;&ThickSpace; \;\;\; Set u ( a 1 ) = 1 u(a_1)=1 u(a1)=1, set u ( a m ) u(a_m) u(am) as follows:
∙ \bullet If a i ≺ ′ a m a_i\prec&#x27;a_m aiam for all i &lt; m i&lt;m i<m, set u ( a m ) = max ⁡ i &lt; m { u ( a i ) } + 1 u(a_m)=\max_{i&lt;m}\{u(a_i)\}+1 u(am)=maxi<m{u(ai)}+1
∙ \bullet If a m ≺ ′ a i a_m\prec&#x27;a_i amai for all i &lt; m i&lt;m i<m, set u ( a m ) = min ⁡ i &lt; m { u ( a i ) } − 1 u(a_m)=\min_{i&lt;m}\{u(a_i)\}-1 u(am)=mini<m{u(ai)}1
∙ \bullet If a i ≺ ′ a m ≺ ′ a j a_i\prec&#x27; a_m \prec&#x27; a_j aiamaj for some i , j &lt; m i,j&lt;m i,j<m and not ( a i ≺ ′ a h ≺ ′ a j ) (a_i\prec&#x27; a_h \prec&#x27; a_j) (aiahaj) for every positive integer h h h that is less than m m m and differs from i i i and j j j. In this case, set u ( a m ) = u ( a i ) + u ( a j ) 2 u(a_m)=\frac{u(a_i)+u(a_j)}{2} u(am)=2u(ai)+u(aj).
This formulation can be easily checked by the fact that ≺ ′ \prec&#x27; is a strict order on X / ≺ X/{\prec} X/, and can be extended to X X X naturally. □ \qquad\qquad\Box

2.3 PREFERENCES AS A STRICT PARTIAL ORDER

DEFINITION 2.2 A binary relation R R R on a set Y Y Y is a strict partial order if and only if it is irreflexive and transitive.
Note that for strict partial order, ∼ \sim is not an equivalence. So we define x ≈ y ⇔ ( x ∼ z ⇔ y ∼ z &ThickSpace;&ThinSpace; for all &ThickSpace; z ∈ X ) ( 2.6 ) x\approx y\Leftrightarrow (x\sim z \Leftrightarrow y\sim z\;\,\text{for all}\;z\in X )\qquad\qquad\qquad(2.6) xy(xzyzfor allzX)(2.6)

THEOREM 2.3. Suppose ≺ \prec on X X X is a strict partial order, being irreflexive and transitive. Then
(1) exactly one of x ≺ y , y ≺ x , x ≈ y , ( x ∼ y , not &ThickSpace; x ≈ y ) x\prec y, y\prec x, x\approx y, (x\sim y, \text{not}\;x\approx y) xy,yx,xy,(xy,notxy) holds for each x , y ∈ X x,y\in X x,yX;
(2) ≈ \approx is an equivalence;
(3) x ≈ y ⇔ ( x ≺ z ⇔ y ≺ z &ThickSpace; and &ThickSpace; z ≺ x ⇔ z ≺ x &ThickSpace;&ThinSpace; for all &ThickSpace; z ∈ X ) x\approx y\Leftrightarrow (x\prec z \Leftrightarrow y\prec z\;\text{and}\;z\prec x \Leftrightarrow z\prec x\;\,\text{for all}\;z\in X ) xy(xzyzandzxzxfor allzX)
(4) ( x ≺ y , y ≈ z ) ⇒ x ≺ z (x\prec y, y\approx z)\Rightarrow x\prec z (xy,yz)xz, and ( x ≈ y , y ≺ z ) ⇒ x ≺ z (x\approx y, y\prec z)\Rightarrow x\prec z (xy,yz)xz;
(5) with ≺ ∗ \prec^* on X / ≈ X/{\approx} X/ (the equivalence classes on X X X under ≈ \approx ) defined by
a ≺ ∗ b ⇔ x ≺ y &ThickSpace;&ThickSpace; for some &ThickSpace; x ∈ a &ThickSpace; and &ThickSpace; y ∈ b ( 2.7 ) a\prec^*b \Leftrightarrow x\prec y\;\;\text{for some}\;x\in a\;\text{and}\;y\in b\qquad\qquad\qquad(2.7) abxyfor somexaandyb(2.7) ≺ ∗ \prec^* on X / ≈ X/{\approx} X/ is a strict partial order.

ZORN’s LEMMA. Suppose P P P on Y Y Y a strict partial order and, for any subset Z Z Z of Y Y Y on which P P P is a strict order, there is a y ∈ Y y\in Y yY such that z P y zPy zPy for all z ∈ Z z\in Z zZ. Then there is a y ∗ ∈ Y y^*\in Y yY such that y ∗ P x y^*Px yPx for no x ∈ Y x\in Y xY.

THEOREM 2.4 (Szpilrajn, 1930). If ≺ ∗ \prec^* is a strict partial order on a set Y Y Y then there is a strict order ≺ 0 \prec^0 0 on Y Y Y that induces ≺ ∗ \prec^* , so that
x ≺ ∗ y ⇒ x ≺ 0 y &ThickSpace;&ThickSpace;&ThickSpace;&ThickSpace; for all &ThickSpace; x , y ∈ Y ( 2.8 ) x\prec^* y\Rightarrow x\prec^0 y\;\;\;\;\text{for all}\;x,y\in Y\qquad\qquad\qquad(2.8) xyx0yfor allx,yY(2.8)

THEOREM 2.5. If ≺ \prec on X X X is a strict partial order and X / ≈ X/{\approx} X/ is countable then there is a real-valued function u u u on X X X such that for all &ThickSpace; x , y ∈ X \;x,y\in X x,yX, x ≺ y ⇒ u ( x ) &lt; u ( y ) ( 2.10 ) x\prec y\Rightarrow u(x)&lt;u(y)\qquad\qquad\qquad (2.10) xyu(x)<u(y)(2.10) x ≈ y ⇒ u ( x ) = u ( y ) ( 2.11 ) x\approx y\Rightarrow u(x)=u(y)\qquad\qquad\qquad (2.11) xyu(x)=u(y)(2.11)

2.4 ORDERED INDIFFERENCE INTERVALS

Some Other Properties.
p 10. &ThickSpace; ( x ≺ y , z ≺ w ) ⇒ ( x ≺ w &ThickSpace; or &ThickSpace; z ≺ y ) &ThickSpace;&ThickSpace; for all &ThickSpace; x , y , z , w ∈ X p10. \;(x\prec y, z\prec w)\Rightarrow (x\prec w\;\text{or}\;z\prec y)\;\;\text{for all}\;x,y,z,w\in X p10.(xy,zw)(xworzy)for allx,y,z,wX.
p 11. &ThickSpace; ( x ≺ y , y ≺ z ) ⇒ ( x ≺ w &ThickSpace; or &ThickSpace; w ≺ z ) &ThickSpace;&ThickSpace; for all &ThickSpace; x , y , z , w ∈ X p11. \;(x\prec y, y\prec z)\Rightarrow (x\prec w\;\text{or}\;w\prec z)\;\;\text{for all}\;x,y,z,w\in X p11.(xy,yz)(xworwz)for allx,y,z,wX.

DEFINITION 2.3. A binary relation is an interval order if it is irreflexive and satisfies p 10 p10 p10, and a semiorder if it is irreflexive and satisfies p 10 p10 p10 and p 11 p11 p11.

For interval orders ( p 2 , p 10 ) (p2, p10) (p2,p10) we shall use the following:

THEOREM 2.6. If ≺ \prec on X X X is an interval order then each of ≺ 1 \prec^1 1 and ≺ 2 \prec^2 2 is a weak order, and x ≈ y ⇔ ( x = 1 y , x = 2 y ) x\approx y\Leftrightarrow (x=^1y,x=^2y) xy(x=1y,x=2y), where x = j y ⇔ ( not &ThickSpace; x ≺ j y , not &ThickSpace; y ≺ j x ) x=^j y\Leftrightarrow (\text{not}\;x\prec^j y, \text{not}\;y\prec^j x) x=jy(notxjy,notyjx).

THEOREM 2.7. If ≺ \prec on X X X is an interval order and X / ≈ X/{\approx} X/ is countable then there are real-valued functions u u u and σ \sigma σ on X X X with σ ( x ) &gt; 0 \sigma(x)&gt;0 σ(x)>0 for all x ∈ X x\in X xX such that x ≺ y ⇔ u ( x ) + σ ( x ) &lt; u ( y ) for all  x , y ∈ X ( 2.14 ) x\prec y\Leftrightarrow u(x)+\sigma(x)&lt;u(y)\qquad\text{for all }x,y\in X\qquad\qquad(2.14) xyu(x)+σ(x)<u(y)for all x,yX(2.14)

THEOREM 2.8. If ≺ \prec on X X X is a semiorder and we define
x &lt; 1 y ⇔ ( x ∼ z , z ≺ y ) for some  z ∈ X ( 2.12 ) x&lt;^1y\Leftrightarrow (x\sim z, z\prec y)\qquad\text{for some }z\in X\qquad\qquad(2.12) x<1y(xz,zy)for some zX(2.12) x &lt; 2 y ⇔ ( x ≺ z , z ∼ y ) for some  z ∈ X ( 2.13 ) x&lt;^2y\Leftrightarrow (x\prec z, z\sim y)\qquad\text{for some }z\in X\qquad\qquad(2.13) x<2y(xz,zy)for some zX(2.13) x &lt; 0 y ⇔ x &lt; 1 y &ThickSpace; or &ThickSpace; x &lt; 2 y for all  x , y ∈ X ( 2.19 ) x&lt;^0y\Leftrightarrow x&lt;^1y\;\text{or}\;x&lt;^2y\qquad\text{for all }x,y\in X\qquad\qquad(2.19) x<0yx<1yorx<2yfor all x,yX(2.19)Then ≺ 0 \prec^0 0 on X X X is a weak order.

THEOREM 2.9. Suppose ≺ \prec on X X X is a semiorder and X / ≈ X/{\approx} X/ is finite. Then there is a real-valued function u u u on X X X such that x ≺ y ⇔ u ( x ) + 1 &lt; u ( y ) for all  x , y ∈ X ( 2.20 ) x\prec y\Leftrightarrow u(x)+1&lt;u(y)\qquad\text{for all }x,y\in X\qquad\qquad(2.20) xyu(x)+1<u(y)for all x,yX(2.20)

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

zte10096334

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值