14 Take Home LT 2
Hyunsang Oh
Riemann Soliven
John Paolo Soto
April 29, 2016
In the formulation of the problem, we need to first identify the decision variables
of the problem then express the constraints in terms of the decision variables.
We are dealing with a goal programming problem so our objective function is
in terms of priority values and deviation variables. In the problem, we let our
decision variable xi be the number of freshman applicants, where i = 1, . . . , 6.
The pair x1 and x2 represent the number of male and female applicants in Metro
Manila, respectively. We designate the same for the male-female applicants
for the provincial and international: x3 and x4 represent the male and female
number of applicants for the provincial while x5 and x6 represent the male and
female number of international applicants, respectively.
For the constraints, we first deal with the goal constraints, these are the
constraints expressed in increasing priority. For P1 , it is stated that the incoming
class is at least 1200 freshmen. Since it is a goal constraint, we add deviational
variables d+
1 and d1 , which gives us
+
x1 + x2 + x3 + x4 + x5 + x6 + d
1 d1 = 1200
(1)
constraint
+
27(x1 +x2 )+26(x3 +x4 )+23(x5 +x6 )+d
2 d2 = 25(x1 +x2 +x3 +x4 +x5 +x6 )
Equating the right hand side to zero, we have the following constraint:
+
2x1 + 2x2 + x3 + x4 2x5 2x6 + d
2 d2 = 0
(2)
since it is stated that the average scores for the Metro Manila, provincial and
international are 27, 26 and 23, respectively. By the constraint, we aim to make
d
2 as small as possible.
For P3 , it says that the main decision variable should be related to the
international students and it should be limited at least 10% of the incoming
+
x5 + x6 + d
3 d3 = 0.1(x1 + x2 + x3 + x4 + x5 + x6 ).
+
0.1x1 0.1x2 0.1x3 0.1x4 + 0.9x5 + 0.9x6 + d
3 d3 = 0.
(3)
x1 + x3 + x5
4
it follows then that it is the same as
4(x2 + x4 + x6 ) 3(x1 + x3 + x5 )
this gives us
+
4(x2 + x4 + x6 ) + d
4 d4 = 3(x1 + x3 + x5 ).
+
3x1 + 4x2 3x3 + 4x4 3x5 + 4x6 + d
4 d4 = 0.
(4)
adding deviational variables d+
5 and d5 gives us the following goal constraint
+
x3 + x4 + d
5 d5 = 0.2(x1 + x2 + x3 + x4 + x5 + x6 )
which is equivalent to
+
0.2x1 0.2x2 + 0.8x3 + 0.8x4 0.2x5 0.2x6 + d
5 d5 = 0.
(5)
x1 = x2
2x3 = 3x4
x5 = 8x6 .
The hard constraints tell us that we can reduce the number of decision variables
in the goal programming problem. Expressing the number of female in terms of
the male applicants, it follows that
x2 = x1
2
x3
3
1
x6 = x5 .
8
x4 =
(6)
(7)
(8)
Thus, we can express the constraints in terms of x1 x3 and x5 (we have expressed
the constraint in terms of the male population of applicants). It has enabled us
to reduce the number of variables in our tableau which will be constructed later
on. Using Equations (6) to (8) on Equations (1) to (5), we have the following
constraints:
5
9
+
2x1 + x3 + x5 + d
1 d1
3
8
5
9
+
4x1 + x3 x5 + d
2 d2
3
4
1
81
1
+
x1 x3 + x5 + d
3 d3
5
6
80
1
5
+
x1 x3 x5 + d
4 d4
3
2
4
9
2
+
x1 + x3 x5 + d
5 d5
5
3
40
= 1200
=0
=0
=0
=0
Min z = P1 d
1 + P2 d2 + P3 d3 + P4 d4 + P5 d5
5
9
+
2x1 + x3 + x5 + d
1 d1
3
8
5
9
+
4x1 + x3 x5 + d
2 d2
3
4
1
1
81
+
x1 x3 + x5 + d
3 d3
5
6
80
1
5
+
x1 x3 x5 + d
4 d4
3
2
2
4
9
+
x1 + x3 x5 + d
5 d5
5
3
40
Subject to
= 1200
=0
=0
=0
=0
+
d
i , di , xi 0
As we have constructed the initial tableau, we are now ready to solve it using
Microsoft Excel. The negative deviational variables form the basis of the initial
tableau. The initial tableau is shown in Figure 1.
d
2 and d4 rows have the same ratio. Thus, we use perturbation to determine
which variable leaves the basis, with = 0.1. After doing so, we see that d
4 is
the one that leaves the basis. A new iteration then is performed, given below.
the same ratio. We then again use perturtabion where = 0.1. After doing so,
we see that d
3 is the variable that leaves the basis. Performing elementary row
operations, a new iteration given below, is performed.
value is 0 which corresponds again to more than one column in the tableau so
we again cannot determine the entering variable. Finally, going to the third
49
priority level P3 , we see that its highest value is
which corresponds to d+
2,
72
therefore it enters the basis. By the minimum ratio rule, d
4 leaves the basis, as
seen in Figure 5. Performing elementary row operations, the next iteration is
given below.
In the tableau we see that the highest value in the P1 row is zero, which corresponds to more than one column in the tableau. Continuing until the fourth
priority level P4 , we see the similar situation. Thus we go the the fifth priority
8
level P5 . In that level we see that its highest value is , which corresponds to
3
x3 , so it enters the basis. By the minimum ratio rule, d
5 leaves the basis, as
seen in Figure 6. The next iteration is given below.
From the given optimal tableau,we can come out with the number of each students from different regions. Since the ratio of male-female applicants in Metro
19632
Manila is 1:1, it then follows that x2 = x1 =
. x2 and x1 again refer to
49
the number of female and male applicants from Metro Manila.
19632 19632
+
49
49
+ 26(144 + 96) + 23
1200
6912 864
+
49
7
= 26.27102
decimal places)
6912 864
+
49
49 = 0.1322
1200
which satisfies P3 and it follows that d
3 = 0. There is a surplus of the ratio,
and d+
3 is equivalent to multiplying the surplus value by the total number of
applicants. At the fourth priority, the male-female ratio of applicants is given
by
19632
+ 96 +
49
19632
+ 144 +
49
864
49 = 0.75
6912
49
which strictly satisfies P4 . That is, there is no surplus nor deficiency in the
priority, the ratio of the provincial applicants to the total number is given by
144 + 96
= 0.2
1200
optimal tableau.
10