Vous êtes sur la page 1sur 9

Operations Research I – Final Exam Paper

(Just for reference)


1. Consider the following matrix game:

0 -1 -2
1 0 -1
2 1 0

(a) Solve the game


w1 = max(min{aij }) = max{−2, −1, 0} = 0
i j

w2 = min(max{aij }) = min{2,1, 0} = 0
j i

x* = {0, 0,1}
There exist a saddle point. The optimal strategy is .
y* = {0, 0,1}

(b) If a21 is changed from 1 to -2 and a31 from 2 to -3, solve the game graphically.

0 -1 -2 0 -1 -2
1 0 -1 -2 0 -1
2 1 0 -3 1 0

w1 = max(min{aij }) = max{−2, −2, −3} = −2


i j
w1 ≠ w2 (no saddle point)
w2 = min(max{aij }) = min{0,1, 0} = 0
j i

0 -1 -2 0 -2
-2 0 -1 -2 -1 The second column is dominated by the third column.
-3 1 0 -3 0

Let y = ( y1 ,1 − y1 )t

P1’s pure strategy P2’s expected payoffs


f1 2 y1 − 2
f2 - y1 − 1
f3 -3y1
so, 2 y1 − 2 = -3 y1
5 y1 = 2
2
y1 =
5
3
y2 = 1 − y1 =
5

Let x = ( x1 , 0,1 − x1 )t

P2’s pure strategy P1’s expected payoffs


g1 3x1 − 3
g2 -2x1
so, 3 x1 − 3 = -2 x1
5 x1 = 3
3
x1 =
5
2
x3 = 1 − x1 =
5
t
* 3 2 * 2 1 3 6
x = , 0, y = , value = -2 x1 = -2 × = -
5 5 5 5 5 5

(c) Solve (b) by simplex method.

0 -1 -2 0 -2 4 2
-2 0 -1 -2 -1 Let λ = 4 2 3
-3 1 0 -3 0 1 4
Y1 Y2 s1 s2 s3 b
-1 -1 0 0 0 0
s1 4* 2 1 0 0 1 1/4
s2 2 3 0 1 0 1 1/2
s3 1 4 0 0 1 1 1

Y1 Y2 s1 s2 s3 b
0 -1/2 1/4 0 0 1/4
Y1 1 1/2 1/4 0 0 1/4 1/2
s2 0 2 -1/2 1 0 1/2 1/4
s3 0 7/2* -1/4 0 1 3/4 3/14

Y1 Y2 s1 s2 s3 b
0 0 3/14 0 1/7 5/14
Y1 1 0 2/7 0 -1/7 1/7
s2 0 0 -5/14 1 -4/7 1/14
Y2 0 1 -1/14 0 2/7 2/14

1 3 5 1 14
Y* = , Y0* = v= * =
7 14 14 Y 5
t t
14 1 3 2 3
y* = , = ,
5 7 14 5 5
t t
14 3 1 3 2
x* = , 0, = , 0,
5 14 7 5 5
14 6
value : v − λ = − 4 = -
5 5

2. By revised simplex method, solve the following Linear Programming Problem by


Big – M technique
Min x0 = 3x1 − 2 x2 + 5 x3
Subject to x1 + x2 + x3 ≥ 5
-3 x1 + x2 − x3 ≤ 4
x1 , x2 , x3 ≥ 0
Min x0 = 3 x1 − 2 x2 + 5 x3 + MR1
Subject to x1 + x2 + x3 − x4 + R1 =5
-3 x1 + x2 − x3 + x5 = 4
x1 , x2 , x3 , x4 , x5 , R1 ≥ 0

1 2 1 -1 1 0 5
A= b=
-3 1 -1 0 0 1 4
c = [3 -2 5 0 M
t
0]
xB = ( R1 , x5 )t B −1 = I 2

cBt B −1 = ( M 0) B −1 = ( M 0)
cBt B −1 N − cNt = ( M 0) [ a1 a2 a3 a4 ] − ( 3 -2 5 0 )
= (M 2M M -M ) − ( 3 -2 5 0 )
= ( M − 3 2M + 2 M − 5 -M )
x2 be the entering variable.
B −1b = (5, 4)t B −1a2 = (2,1)t
t
5
The ratio is ,4
2
R1 be the leaving variable.
1 0 1 0
E= Bˆ −1 = EB −1 =
-1/ 2 1 -1/ 2 1
xB = ( x2 , x5 ) B −1 = Bˆ −1

cBt B −1 = (-2 0) B −1 = (-2 0)


cBt B −1 N − cNt = (-2 0) a1 a3 a4 aR1 − ( 3 5 0 M )
= (-2 -2 2 -2) − ( 3 5 0 M )
= (-5 -7 2 -2 − M )
x4 be the entering variable.
1 0 5 1 0 -1
B −1b = = (5,3 / 2)t B −1a4 = = (−1,1/ 2)t
-1/ 2 1 4 -1/ 2 1 0
t
The ratio is ( -,3)
x4 be the leaving variable.
1 0 1 0
E= Bˆ −1 = EB −1 =
0 1 -1/ 2 1
xB = ( x2 , x4 ) B −1 = Bˆ −1
cBt B −1 = (-2 0) B −1 = (-2 0)
cBt B −1 N − cNt = (-2 0) a1 a3 aR1 a5 − ( 3 5 M 0)
= (-2 -2 -2 0) − ( 3 5 M 0)
= (-5 -7 -2 − M 0) ≤ 0
1 0 5
xB* = ( xx* , x4* ) = B −1b =
t
= ( 5,3 / 2 )
-1/ 2 1 4
5
x0* = cBt B −1b = ( -2, 0 ) = -10
4

3. For the following maximization problem, the initial and optimal tableaux are given
below.
x1 x2 x3 x4 x5 b
-9 -8 -5 0 0 0
x4 2 3 1 1 0 4
x5 5 4 3 0 1 11

x1 x2 x3 x4 x5 b
0 2 0 2 1 19
x1 1 5 0 3 -1 1
x3 0 -7 1 -5 2 2

A New constraint -9 x2 + x3 ≤ 1 is added to the problem. Starting with the optimal


solution, find a new optimal solution.

Test: x* = (1, 0, 2, 0, 0)t , so substitute to -9 x2 + x3 ≤ 1 , then -9 x2 + x3 = -9 × 0 + 2 = 2

It is not satisfy.

x1 x2 x3 x4 x5 x6 b
0 2 0 2 1 0 19
x1 1 5 0 3 -1 0 1
x3 0 -7 1 -5 2 0 2
x6 0 -9 1 0 0 1 1

We see that the x3 column is not satisfied to be optimal tableaux.


x1 x2 x3 x4 x5 x6 b
0 2 0 2 1 0 19
x1 1 5 0 3 -1 0 1
x3 0 -7 1 -5 2 0 2
x6 0 -2 0 5 -2 1 -1

We use Dual simplex method


x1 x2 x3 x4 x5 x6 b
0 2 0 2 1 0 19
x1 1 5 0 3 -1 0 1
x3 0 -7 1 -5 2 0 2
x6 0 -2 0 5 -2 1 -1
ratio: (- 1 - - 1/2 1)

x1 x2 x3 x4 x5 x6 b
0 1 0 9/2 0 1/2 37/2
x1 1 6 0 1/2 0 -1/2 3/2
x3 0 -9 1 0 0 1 1
x5 0 1 0 -5/2 1 -1/2 1/2

t
x* = ( 3 / 2, 0,1, 0,1/ 2, 0 )
x0* = 37 / 2

4. Consider the Linear Programming Problem Max {x0 = ct x : Ax = B, x ≥ 0} . Show that

if x is a basic feasible solution, it is also an extreme point of the feasible region.


(see Blackson proof).

5. Consider the following Linear Programming Problem


Min x0 = 2 x1 − x3
Subject to x1 − x2 + x3 ≥ 2
x1 + x2 − x3 ≥ 1
x1 , x2 ≥ 0, x3 ≤ 0
(a) Find the dual problem
Max y0 = 2 y1 + y2
Subject to y1 + y2 ≤ 2
Dual problem: -y1 + y2 ≤ 0
y1 − y2 ≥ −1
y1 , y2 ≥ 0

(b) Solve the dual problem graphically.

y1* = 2, y2* = 0
y0* = 4

(c) By complementary slackness property, solve the given problem.


y1 + y2 = 2 x1* > 0
∴ x2* = 0, x3* = 0
y1 > 0 x1 − x2 + x3 = 2
x1 = 2
t
∴ x* = ( 2, 0, 0 )
t
y* = ( 2, 0 )
x0* = y0* = 4
6. Solve the following Problem.
(a) Consider the following cost and requirement table of a transportation problem.
si
3 7 6 4 5
2 4 3 2 2
4 3 8 5 3
dj 3 3 2 2
Find an optimal solution for this transportation problem.
si
3 7 6 4 3 2 5
cost: 2 4 3 2 x-table: 1 1 2
4 3 8 5 1 2 3
dj 3 3 2 2

ui + v j − cij ≤ 0
Find the most positive pivot
Let vn = 0

si dj 3 3 2 2 ui vj 0 4 3 0
5 3 2 3 0 0 0 -1
2 1- δ 1+ δ 0 -2 0 0 -2
3 δ 1- δ 2 5 1 6 0 0
δ =1
si dj 3 3 2 2 ui vj -6 -2 -3 0
5 3 2+ δ δ 9 0 0 0 5
2 0 2 6 -2 0 0 4
3 1- δ 2+ δ 5 -5 0 -6 0
δ =1
si dj 3 3 2 2 ui vj -1 3 2 0
5 3 3 1 4 0 0 0 0
2 0 2 1 -2 0 0 -1
3 3 5 0 -1 -1 0

x0* = 3c11 + 3c12 + c14 + 2c23 + 3c34 = 9 + 21 + 4 + 6 + 15 = 55


(b) Consider the problem of assignment 4 jobs to 4 machines. The assignment costs
are given below. Find an optimal way of assignment.
Machine
1 2 3 4
1 8 7 9 9
2 5 2 7 8
Job
3 6 1 4 9
4 6 3 2 6

Subtracting the smallest entry in each column, we get


3 6 7 3
0 1 5 2
1 0 2 3
1 2 0 0

Subtracting the smallest entry in each row, we get


0 3 4 0
0 1 5 2
1 0 2 3
1 2 0 0
It obvious from the last cost matrix that an optimal assignment is given by

x21 = x32 = x43 = x14 = 1


x0* = c21 + c32 + c43 + c14 = 17

- End -
Good Luck!

Vous aimerez peut-être aussi