1 Introduce slack variables x3, x4, and x5to obtain
Max
z = x1 + 3x2
Subject to
-x1 - x2 + x3 = -2
x1 - x2 + x4 = -1
2x1 + x2 + x5 = 4
,
i = 1, ..., 5..
If we choose x1, and x2 as non-basic variables,
then it is not feasible since x3, and x4 are negative.
Let us find an initial feasible solution by introducing a new variable x0,
and solving the following LP problem:
Max
z = -x0
Subject to
,
i = 0, 1,2..
This can be solved starting with
Max
z = -x0
Subject to
-x1 - x2 - x0 + x3 = -2
x1 - x2 - x0 + x4 = -1
2x1 + x2 - x0 + x5 = 4
,
i = 0, ..., 5..
If we select x3, x4, and x5 as basic variables, it is not feasible as can be seen below.
x3 = -2 + x1 + x2 + x0
x4 = -1 - x1 + x2 + x0
x5= 4 - 2x1 - x2 + x0
z = -x0.
,
i = 0, ..., 5..
Select 2 for the initial value for x0 while keeping x1 and x2 0. Then the initial dictionary is
x0 = 2 - x1 - x2 + x3
x4 = 1 - 2x1 + x3
x5= 6 - 3x1 - 2x2 + x3
z = -2 + x1 + x2 - x3.
,
i = 0, ..., 5..
Interchanging x2 with x0 for a new dictionary, we obtain
x2 = 2 - x1 - x0 + x3
x4 = 1 - 2x1 + x3
x5= 2 - x1 + 2x0 - x3
z = -x0.
Thus the original problem is feasible, and an initial feasible solution for the original problem is
x0 = 0, x1 = 0, x2 = 2, x3= 0, x4 = 1, and x5 = 2.
2 a) The dual is
Max
5y1 + 8y2
Subject to
b) Draw a graph of the feasible region and also two or three
positions of z = c lines for the objective function.
By sliding up the line z = c we can see that z is maximum
at
y1 = 2, y2 = 3, and z = 34.
3 a) Let
f1(x1) = 4x1 - x12, and
f2(x2) = 4x2 - x23.
Then f's can be approximated by
f1(x11,x12,x13) = 3x11 + x12 - x13, and
f2(x21,x22,x23) = 3x21 -3x22 -15x23,
where and xij = 0 if xij-1 < 1for all i and j, i = 1,2, and j = 1,2,3.
Hence the approximation problem is
Max
z = 3x11 + x12 - x13 + 3x21 -3x22 -15x23
Subject to
for all i and j, i = 1,2, and j = 1,2,3.
b) Let
xij = 0 for all i and j, i = 1,2, and j = 1,2,3, and
let all the slack variables a basic variable.