1 1

Optimization in Operations Research 2nd

Edition Rardin Solutions Manual
Full download at link:

Solution Manual: https://testbankpack.com/p/solution-manual-for-

optimization-in-operations-research-2nd-edition-rardin-0134384555-
9780134384559/

Chapter 5 Solutions 5-2.

1
(a)
5-1. w2
(a) 4
w2

4
w (4)= (3,3)
3
w(1)= (2,3) w (4) = (3,3)
3 w (2) = (0,3)

2
2
w (3)= (2,1) w (2)= (1,1)
1
1
w (3)= (2,0)
w1
1 w (1) = (0,0) 1 2 3 4
w
1 2 3 4
(b) w (1) is both boundary and extreme
(b) w (1) is boundary and extreme because it because it forms a corner point of the feasible
forms a corner point of the feasible region. region. w(2) is an interior point of the feasible
w(2) is infeasible because it lies outside the region. w (3) is a boundary point of the
feasible region. w (3) is an interior point of the feasible region, but not extreme. w(4) is
feasible region. w (4) is a boundary point of infeasible because it lies outside the feasible
the feasible region, but not an extreme point. region. (c) Both w1 ≥ 0 and w2 ≥ 0 are
(c) Both −w1 + w2 ≤ 1 and w2 ≤ 3 are active active at extreme point w(1) . No constraints
at extreme point w(1) . No constraints are are active at interior point w(2) . Only w2 ≥ 0
active at interior point w(3) . Only w2 ≤ 3 is is active at boundary point w (3) . (d) w(1)
active at boundary point w(4) . (d) w(1) could could be optimal or uniquely optimal because
be optimal or uniquely optimal because it is it is an extreme point. Interior point w (2) can
an extreme point. Infeasible w(2) could be be neither. Boundary point w(3) can be
neither. Interior point w (3) also cannot be optimal, but not uniquely so. Infeasible points
either optimal or uniquely optimal. Boundary w(4) and w(5) can be neither.
point w (4) can be optimal, but not uniquely 5-3. (a) Introducing nonnegative slack
so because it is not an extreme point. (e) variables −x4 in the first main constraint and
w(1) basic feasible with actives w2 ≤ 3 and +x5 in the last gives standard form
−w1 + w2 ≤ 1; w (2) basic infeasible with ⎛ ⎞
1 −4 1 1 0

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 2 2

actives w1 ≥ 0 and w2 ≤ 3; all others not A =⎝ 9 0 6 0 0 ⎠,

basic.w(1) basic feasible with actives w2 ≤ 3 −5 9 0 0 −1
and −w1 + w2 ≤ 1; w (2) basic infeasible with b = (12 , 3), c = (4, 2, , 0, 0)
, 15 −33
actives w1 ≥ 0 and w2 ≤ 3; all others not (b) Introducing nonnegative slack variables
basic. −x4 in the second main constraint and +x5 in
the last
⎛ gives standard form ⎞
4 −2 9 −1 0
A = ⎝ −2 5 −1 0 0 ⎠,
1 Supplement to the 2nd edition of Optimization in 1 −1 0 0 1
Operations Research, by Ronald L. Rardin, Pearson
Higher Education, Hoboken NJ, c 2017. b = (22, 1, 5), c = (45, 0, 15, 0, 0)

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 3 3

(c) Simplifying the objective function gives 5-4.

max 15x1 + 41x2 − 11x3 . Collecting variables (a)
on the left and constants on the right makes y2
the main constraint 2x1 − x2 − x3 ≥ 0. Then (2,4)
introducing nonnegative slack variable −x4 in
that row and +x5 through +x7 in the 3 upper
bound rows gives standard form
⎛ ⎞
2 −1 −1 −1 0 0 0
⎜ 1 0 1 0 0 ⎟
A =⎜ ⎝ 0
0 0 ⎟,
1 0 0 0 1 0 ⎠
0 −11,
b = (0, 3,0 3, 3),0 c =1(15, 41, 0 0 0, 10, 0, 0) (d)
Simplifying the objective function gives max
−20x1 + 99x3 . Collecting variables on the
left and constants on the right makes the y1
first 2 constraints xj − xj+1 ≤ −1 for j = 1, 2. (-2,0) (0,0)
Then introducing nonnegative slack variables −1 1 1 0

+x4 and +x5 in those 2 constraints gives (b) A = , b = (2, 10) (c)
5 0 0 1
⎛ form
standard ⎞ The columns will be a basis if they are a
1 −1 0 1 0
maximal set of linearly independent vectors,
A =⎝ 0 1 −1 0 1 ⎠,
or equivalently, if the submatrix they form
1 1 1 0 0
has nonzero determinant. This is true for
b = (−1, −1, 12), c = (−20, 0, 99, 0, 0)
{ y1 , y2 } with matrix determinant =-5, for
(e) Substituting for URS variable x2 with {y3 , y4 } with determinant =+1, and for
nonnegative variable difference x2 = x − x ,
2 3 y1 , y4 } with determinant = -1. Sets {y2 , y3 }
and for nonpositive variable x3 with negative
{
and {y1 , y2 , y4 } are linearly dependent
x3 = −x4 , then introducing nonnegative slack because the column of one member can be
variable +x5 in the first main constraint gives
expressed as a weighted sum of the others.
standard form
1 −1 1 5 1 Set {y3 } has too few members. (d) We obtain
A = , b = (10, −6), basic solutions by fixing nonbasic variables =0
0 3 −3 −9 0
c = (2, 1, −1, 4, 0) and solving Ay = b for the basics. For
(f ) Substituting for nonpositive x2 with {y1 , y2 } this gives y = (2, 4, 0, 0) which is
negative x2 = −x2, and for URS variable x1 feasible because basic components are
with nonnegative variable difference nonnegative. For {y3 , y4 } the result is
x1 = x − x , then introducing nonnegative y = (0, 0, 2, 10) which is also feasible. For
1 1
{y1 , y4 } the basic solution is y = (−2, 0, 0, 20)
slack variable +x4 in the second main
Δx , x , x , x , x ) which is infeasible because its first component
constraint gives for x= 3 4
1 1 2
is negative. (e) Corresponding points on the
standard form
− 4 4 1 7 0 plot are the first two components of each
A = ,
−1 1 1 3 1 basic solution, i.e. y = (2, 4), y = (0, 0), and
b = (9, 14), c = (4, −4, 1, 0, 0, 0) y = (−2, 0). Basic feasible solutions are
extreme points, but basic infeasible solutions
just represent cases where enough constraints
are active to define a point.

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 4 4

5-5. active nonnegativity gives Δw2 ≥ 0.

5-7. (a) Setting nonbasics x2 = x4 = 0 and
(a)
y2
solving Ax = b yields x = (1, 0, 3, 0). (b) For
x2 we fix Δx2 = 1, Δx4 = 0, and solve
4
(0,3)
A Δx = 0 to obtain Δx = (−3, 1, −1, 0). For
x4 we fix Δx4 = 1, Δx2 = 0, and solve
(0,2) (2,2) A Δx = 0 to obtain Δx = (1, 0, −5, 1) for x4 .
(c) Each was constructed to satisfy
A Δx = 0, and have Δx2 ≥ 0, Δx4 ≥ 0 for
y1 active nonnegativity constraints. (d)
2 4 6 8
Computing c̄ = c · Δx gives -29 for the
1 2 1 0 direction increasing x2 and 10 for the
(b) A = , b = (6, 2) (c) The
0 1 0 1 direction increasing x4 . Thus only the second
columns will be a basis if they are a maximal improves for this maximize problem. (e) The
set of linearly independent vectors, or maximum step size λ is the maximum move
equivalently, if the submatrix they form has before some decreasing basic variable reaches
nonzero determinant. This is true for set its lower limit of zero. For x2 λ =
{y1 , y2 } with determinant =+1, for {y2 , y3 } min{1/3, 3/1} = 1/3; x2 replaces x1 in the
with determinant =-1, and for {y2 , y4 } with basis to give {x2 , x3 }. For x4 λ = min{3/5} =
determinant =+2. Sets {y2 , y3 , y4 } and 3/5, {x1 , x4 }
{y1 , y3 } are linearly dependent because the 5-8. (a) Setting nonbasic x1 = x2 = 0 and
column of one can be expressed as a weighted solving Ax = b yields x = (0, 0, 2, 1). (b) For
combination of the others. Set {y1 } has too x1 we fix Δx1 = 1, Δx2 = 0, and solve
few members. (d) We obtain basic solutions A Δx = 0 to obtain Δx = (1, 0, −3, −4). For
by fixing nonbasic variables =0 and solving x2 we fix Δx2 = 1, Δx1 = 0, and solve
Ay = b for the basics. For {y1 , y2 } this gives A Δx = 0 to obtain Δx = (0, 1, −1, 1). (c)
y = (2, 2, 0, 0) which is feasible because basic Each was constructed to satisfy A Δx = 0,
components are nonnegative. For {y2 , y3 } the and have Δx1 ≥ 0, Δx2 ≥ 0 for active
results is y = (0, 2, 2, 0) which is also feasible. nonnegativity constraints. (d) Computing
Set {y2 , y4 } yields y = (0, 3, 0, −1) which is c̄ = c · Δx gives 4 for the direction increasing
infeasible because its last component is x1 and -4 for the direction increasing x2 .
negative. (e) Corresponding points on the Thus only the second improves for this
plot are the first two components of each minimize problem. (e) The maximum step
basic solution, i.e. y = (2, 2), y = (0, 2), and size λ is the maximum move before some
y = (0, 3). Basic feasible solutions are decreasing basic variable reaches its lower
extreme points, but basic infeasible solutions limit of zero. For x1 λ = min{2/3, 1/4} =
just represent cases where enough constraints 1/4; x1 replaces x4 in the basis to give
are active to define a point. {x1 , x3 }. For x2 λ = min{2/1} = 2; x2
replaces x3 in the basis to give {x2 , x4 }.
5-6. (a) The equality constraints imply
5 Δw1 + 1 Δw2 − 1 Δw3 = 0 and
3 Δw1 − 4 Δw2 + 8 Δw3 = 0, and the one
active nonnegativity gives Δw2 ≥ 0. (b) The
equality constraints imply
4 Δw1 − 2 Δw2 + 5 Δw3 = 0 and
4 Δw1 + 2 Δw2 − 3 Δw3 = 0, and the one

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 5 5

5-9. 5-10.
(a)
(a)
z 2 z2

4 4

z (1) = (0,3)
3 3
z (2) = z* = (4,3)
z (2) = z* = (4,2)

2 2

1 1
z (1) = (4,0)
z1 z1
z (0) = (0,0) 1 2 3 4 5
z (0) = (0,0) 1 2 3 4

⎛ ⎞ ⎛ ⎞
−2 1 1 0 0 3 2 1 0 0

(b) A = ⎝ 1 1 0 1 0 ⎠, b = (2, 6, 4), (b) A = ⎝ 1 0 0 1 0 ⎠, b = (18, 5, 3),

1 0 0 0 1 0 1 0 0 1

c = (3, 1, 0, 0, 0) c = (2, 5, 0, 0, 0)
(c) (c)

z1 z2 z3 z4 z5
max c 3 1 0 0 0 b
A -2 1 1 0 0 2
1 1 0 1 0 6 z1 z2 z3 z4 z5
1 0 0 0 1 4
max c 2 5 0 0 0 b
t=0 N N B B B A 3 2 1 0 0 18
z(0) 0 0 2 6 4 0 1 0 0 1 0 5
Δz, z1 1 0 2 -1 -1 c̄1 = 3 0 1 0 0 1 3
Δz, z2 0 1 -1 -1 0 c̄2 =1 t=0 N N B B B

6 4 z(0) 0 0 18 5 3 0
- - - λ=4

1 1 Δz, z1 1 0 -3 -1 0 c̄1 =2

t=1 B N B B N Δz, z2 0 1 -2 0 -1 c̄2 = 5

z(1) 4 0 10 2 0 12 18 3
- - - λ=3
Δz, z2 0 1 -1 -1 0 c̄2 = 1 2 1

Δz, z5 -1 0 -2 1 1 c̄5 =-3 t=1 N B B B N

- -
10 2
- λ=2 z(1) 0 3 12 5 0 15

1 1 Δz, z1 1 0 -3 -1 0 c̄1 = 2

t=2 B B B N N Δz, z5 0 -1 2 0 1 c̄5 =-5

z(2) 4 2 8 0 0 14 12 5
- - - λ=4
Δz, z4 0 -1 1 1 0 c̄4 =-1 3 1

Δz, z5 -1 1 -3 0 1 c̄5 =-2

t=2 B B N B N
z(2) 4 3 0 1 0 23
Δz, z3 -0.33 0 1 0.33 0 c̄3 =-0.67
Δz, z5 0.67 -1 0 -0.67 1 c̄5 =-3.67

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 6 6

5-11.

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 7 7

(a) (b)
x1 x2 x3 x4 x5
max c 5 -10 0 0 0 b
A 1 -1 2 0 4 2
1 1 0 2 1 8
t=0 N N B B N
x(0) 0 0 1 4 0 0
Δx, x1 1 0 -.5 -.5 0 c̄1 = 5
Δx, x2 0 1 .5 -.5 0 c̄2 =-10
Δx, x5 0 0 -2 -.5 1 c̄5 =0
2 8
t=1 B N N B N

x(1) 2 0 0 3 0 10
Δx, x2 1 1 0 -1 0 c̄2 =0
Δx, x3 -2 0 1 0 0 c̄3 =-10
Δx, x5 -4 0 0 1.5 1 c̄5 =-20

x(1) is optimal because all nonbasic c̄j ≤ 0 in

(b) a max problem.
5-14. (a) x(0) = (6, 9, 0, 0, 0), appropriate
y1 y2 y3 y4
max c 10 1 0 0 b because basic feasible. (b)
A 3 2 -1 0 6
x1 x2 x3 x4 x5
2 4 0 1 8
min c 2 4 6 10 7 b

(c) Slack variable values are the differences A 1 0 0 1 0 6

0 1 1 -1 2 9
between the original LHS and the RHS. If an t=0 B B N N N
original constraint is active, the corresponding x(0) 6 9 0 0 0 48

Δx, x4 -1 1 0 1 0 c̄4 =12

constraint active. Δx, x5 0 -2 0 0 1 c̄5 = -1
5-12. (a) x(0) = (2, 0, 8, 5, 0), appropriate 4.5
because basic feasible. (b) t=1 B N N N B

x(1) 6 0 0 0 4.5 43.5

x1 x2 x3 x4 x5
Δx, x2 0 1 0 0 -.5 c̄2 =.5
min c 0 1 0 1 1 b Δx, x3 0 0 1 0 -.5 c̄3 =2.5
A -2 1 0 2 0 7 Δx, x4 -1 0 0 1 1/2 c̄4 =11.5
0 0 0 1 1 5
0 1 1 -1 0 3 (1)

t=0 B N B B N
x is optimal because all nonbasic c̄j ≥ 0 in
x(0) 3/2 0 8 5 0 5 a min problem.

Δx, x2 .5 1 -1 0 0 c̄2 =1 5-15. (a) Yes, because this is a sequence of

Δx, x5 1 0 -1 -1 1 c̄5 =0
adjacent extreme points. (b) No, because

x(0) is optimal because all nonbasic c̄j ≥ 0 in starting point P 2 is not an extreme point. (c)
a min problem. No, because interior point P 3 could not be
5-13. (a) x(0) = (0, 0, 1, 4, 0), appropriate part of a simplex path. (d) Yes, because this
because basic feasible. is a sequence of adjacent extreme points. (e)
No, even though all these points are extreme
points, because the move from P 1 to P 7 goes
through the interior. (f ) No, because simplex
would never take a nonimproving step like the
one from P 4 to P 1.
5-16. (a) z = 10 − 29x2 + 10x3 ,
x1 = 1 − (3x2 − 1x4 ), x3 = 3 − (1x2 + 5x4 )

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 8 8

(b) z = 1 + 4x1 − 4x2 , x3 = 2 − (3x1 + 1x2 ), (c)

x4 = 1 − (4x1 − 1x2 )
y1 y2 y3 y4
5-17. This minimizing LP will be unbounded
max c 4 5 0 0 b
if the indicated simplex direction improves the A -1 1 1 0 4
objective function and has no negative 1 -1 0 1 10
components (decreasing basic variables). (a) t=0 N N B B
No, because there is a negative component. y(0) 0 0 4 10 0
(b) Yes, because there are no negative Δy, y1 1 0 1 -1 c̄1 =4
components, and (3, 11, −8, 0) · (1, 3, 10, 1) Δy, y2 0 1 -1 1 c̄2 = 5

= −44 < 0. (c) No, because (3, 11, −8, 0) · - -

4
- λ=4
1
(1, 3, 0, 1) = 36 > 0. (d) No, because there
t=1 N B N B
are negative components.
y(1) 0 4 0 14 20
5-18.
Δy, y1 1 1 0 0 c̄1 = 9
(a) Δy, y3 0 -1 1 -1 c̄3 =-5
y2 - - - - unbounded

8
5-19.
6
(a)

y2
4 y (1) = (0,4)

5
2
y (0) = (0,0) 4
y1
2 4 6 8 10

−1 1 1 0 3
(b) A = , b = (4, 10),
1 −1 0 1
c = (4, 5, 0, 0) 2

1
y (1) = (8/3,0)
y
1
y (0) = (0,0) 1 2 3 4

−5 3 1 0
(b) A = , b = (15, 8),
3 −5 0 1

c = (−10, 1, 0, 0)

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 9 9

(c) variable sum to obtain: max

5w1 + 18w2 − M (w6 + w7 + w8 ) s.t. same
y1 y2 y3 y4 constraints and starting basis as Exercise
min c -10 1 0 0 b 5-20(b).
A -5 3 1 0 15
3 -5 0 1 8
5-22.
t=0 N N B B
y(0) 0 0 15 8 0
(a)
y2
Δy, y1 1 0 5 -3 c̄1 = -10
Δy, y2 0 1 -3 5 c̄2 =1 4
8
- - - λ = 2.67
3
t=1 B N B N 3
y (1) 2.67 0 28.33 0 -26.67
Δy, y2 1.67 1 5.33 0 c̄2 = -15.67
Δy, y4 -0.33 0 -1.67 1 c̄4 =3.33 2

- - - - unbounded

5-20. (a) We first introduce slack variables 1

+w4 and −w5 in the two main inequalities,
then introduce artificial variables −w6 and
+w7 in the last two main constraints to y1
1 2 3 4
provide an (artificially) starting feasible basis.
−2 1 −1 0 1
The objective minimizes the sum of the (b) A = , b = (2, 1),
artificials. The result is: min w6 + w7 , s.t. 0 1 0 1 0

d = (0, 0, 0, 0, 1)
w1 + w2 + w4 = 18, −2w1 + w3 − w6 = −2, (c)
3w2 + 5w3 − w5 + w7 = 15, w1 , . . . , w7 ≥ 0;
starting basis {w4 , w6 , w7 } (b) We first y1 y2 y3 y4 y5
introduce slack variables −w3 , −w4 and +w5 min d 0 0 0 0 1 b
in the three main inequalities, then introduce A -2 1 -1 0 1 2
0 1 0 1 0 1
artificial variables +w6 , +w7 and +w8 in first
t=0 N N N B B
three main constraints to provide an
y (0) 0 0 0 1 2 2
(artificially) starting feasible basis. The Δy, y1 1 0 0 0 2 c̄1 =2
objective minimizes the sum of the artificials. Δy, y2 0 1 0 -1 -1 c̄2 = -1
The result is: min w6 + w7 + w8 , s.t. Δy, y3 0 0 1 0 1 c̄3 =1

2w1 + 4w2 + w6 = 128, 1 2

- - - λ=1
7w1 + w2 − w3 + w7 = 11, 1 1

6w1 + 16w2 − w4 + w8 = 39, t=1 N B N N B

w1 + 3w2 + w5 = 222, w1 , . . . , w8 ≥ 0; starting y (1) 0 1 0 0 1 1

basis {w5 , w6 , w7 , w8 } Δy, y1 1 0 0 0 2 c̄1 =2

Δy, y3 0 0 1 0 1 c̄3 =1
5-21. (a) For this maximize model we Δy, y4 0 -1 0 1 1 c̄4 =1
subtract a large multiple of the artificial

variable sum to obtain: max Optimality with a positive artificial variable

2w1 + w2 + 9w3 − M (w6 + w7 ) s.t. same sum indicates infeasibility.
constraints and starting basis as Exercise
5-20(a). (b) For this maximize model we
subtract a large multiple of the artificial

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 10 10

5-23. constraints the bound is (4!) / (2!2!) = 6 (c)

(a) With 2340 variables and 1150 constraints the
bound is (2340!) / (1150!1190!) (d) With
7200 variables and 211 constraints the bound
is (7200!) / (211!6989!)
5-25. In the context of rudimentary simplex
Algorithm 5A, variables have only lower
bounds xj ≥ 0, and a solution is degenerate if
any basic component is at its lower bound.
(a) Yes, because basic x2 = 0. (b) Yes,
because basic x4 = 0. (c) No, because all
basic values are positive. (d) Yes, becauseall
basics x1 = 0, x2 = 0 and x4 = 0.
5-26.
(a)
1 1 1 0 0
(b) A = , b = (5, 6),
0 1 0 −1 1

d = (0, 0, 0, 0, 1)
(c)
y1 y2 y3 y4 y5
min d 0 0 0 0 1 b
A 1 1 1 0 0 5
0 1 0 -1 1 6
t=0 N N B N B
y(0) 0 0 5 0 6 6
Δy, y1 1 0 -1 0 0 c̄1 =0
Δy, y2 0 1 -1 0 -1 c̄2 = -1
Δy, y4 0 0 0 1 1 c̄4 =1
5 6
- - - λ=5 ⎛ ⎞
1 1 −1 1 1 0 0
t=1 N B N N B (b) A = ⎝ 5 0 0 1 0 ⎠,
y(1) 0 5 0 0 1 1 0 1 0 0 1
Δy, y1 1 -1 0 0 1 c̄2 =1
Δy, y3 0 -1 1 0 1 c̄3 =1
b = (2, 10, 4) (c) The columns will be a basis
Δy, y4 0 0 0 1 1 c̄4 =1 if they are a maximal set of linearly
independent vectors, or equivalently, if the
Optimality with a positive artificial variable submatrix they form has nonzero
sum =6 indicates infeasibility. determinant. This is true for {y1 , y2 , y5 } with
5-24. Bases cannot repeat if the step size matrix determinant =-5, for {y3 , y4 , y5 } with
λ > 0 at every step produces strict determinant =+1, and for {y1 , y4 , y5 } with
improvement in the objective function value. determinant = -1. Sets {y2 , y3 , y5 } and
Thus a bound can be obtained by determining {y1 , y2 , y4 , y5 } are linearly dependent because
the number of distinct combinations of the column of one member can be expressed
variables that could make up a basis. (a) as a weighted sum of the others. Set {y3 , y5 }
With 4 variables and 2 constraints the bound has too few members. (d) We obtain basic
is (4!) / (2!2!) = 6 (b) With 4 variables and 2 solutions by fixing nonbasic variables =0 and

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 11 11

solving Ay = b for the basics. For {y1 , y2 , y5 } {y1 , y2 , y5 } with determinant =+1, for
this gives y = (2, 4, 0, 0, 0) which is feasible {y2 , y3 , y5 } with determinant =1, {y2 , y3 , y5 }
because basic components are nonnegative. with determinant =-1, and {y2 , y4 , y5 } with
For {y3 , y4 , y5 } the result is y = (0, 0, 2, 10, 4) determinant =+2. Sets {y2 , y3 , y4 } and
which is also feasible. For {y1 , y4 , y5 } the {y1 , y3 , y5 } are linearly dependent. Set
basic solution is y = (−2, 0, 0, 20, 4) which is {y1 , y5 } has too few members. (d) Sets
infeasible because its first component is {y1 , y2 , y4 }, {y1 , y3 , y4 } and {y1 , y4 , y5 } all
negative. (e) Corresponding points on the produce basic solution y(0) = (6, 0, 0, 2, 0). (e)
plot are the first two components of each Choosing {y1 , y2 , y4 }, Δy for y3 is
basic solution, i.e. y = (2, 4), y = (0, 0), and (0,-1/2,1,1/2,0)). The component for y2 is
y = (−2, 0). Basic feasible solutions are negative, but y2 = 0 the current solution,
extreme points, but basic infeasible solutions which will make step λ = 0, and leave y3 = 0.
just represent cases where enough constraints
are active to define a point. (f ) {y1 , y2 , y3 }, 5-28.
{y1 , y2 , y4 }, and {y1 , y2 , y5 } all compute (a)
x2
solution y = (2, 4) in the plot. (g) These
solutions are degenerate because some basic 10

components =0. From basis {y1 , y2 , y3 },

introducing y4 gives direction 8

Δy = (−1/5, 0, −1/5, 1, 0), which will make x (2)= x* = (3,6)

λ = 0 because decreasing y3 already = 0. 6
alternative
optima

4
5-27.
(a)
2

x1
4 6 8 10
x (0) = x(1) = (0,0)

⎛ ⎞
1 1 1 0 0
(b) A = ⎝ −2 1 0 1 0 ⎠,
1 −2 0 0 1
b = (9, 0, 0), c = (1, 1, 0, 0, 0)

⎛ ⎞
1 2 1 0 0
(b) A = ⎝ 0 1 0 1 0 ⎠, b = (6, 2, 6)
1 0 0 0 1
(c) The columns will be a basis if they are a
maximal set of linearly independent vectors,
or equivalently, if the submatrix they form
has nonzero determinant. This is true for sets

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 12 12

(c) (c)

x1 x2 x3 x4 x5 x1 x2 x3 x4
max c 1 1 0 0 0 b max c 1 0 0 0 b
A 1 1 1 0 0 9 A 6 3 1 0 18
-2 1 0 1 0 0 12 -3 0 1 0
1 -2 0 0 1 0 t=0 N N B B
t=0 N N B B B x(0) 0 0 18 0 0
x(0) 0 0 9 0 0 0 Δx, x1 1 0 -6 -12 c̄1 = 1
Δx, x1 1 0 -1 2 -1 c̄1 =1 Δx, x2 0 1 -3 3 c̄2 = 0
Δx, x2 0 1 -1
9 -1
0 2 c̄2 = 1 - -
18 0
- - - λ=0 6 12 λ=0
1 1
t=1 B N B N
t=1 N B B N B x(1) 0 0 18 0 0
x(1) 0 0 9 0 0 0 Δx, x2 0.25 1 -4.50 0 c̄2 = 0.25
Δx, x1 1 2 -3 0 3 c̄1 = 3 Δx, x4 -0.08 0 0.50 1 c̄4 =-0.08
Δx, x4 0 -1 1 1 -2 c̄4 =-1 18
- -
- - λ=3 - λ=4
9 t=2 B B 4.50
- -
3 N N
t=2 B B N N B x(2) 1 4 0 0 1

x(2) 3 6 0 0 9 9 Δx, x3 -0.06 -0.22 1 0 c̄3 =-0.06

Δx, x3 -0.33 -0.67 1 0 -1 c̄3 =-1 Δx, x4 -0.06 0.11 0 1 c̄4 =-0.06
Δx, x4 0.33 -0.33 0 1 -1 c̄4 =0
(e) Bases, and thus simplex directions
(e) Bases, and thus simplex directions changed even though the solution did not.
changed even though the solution did not. 5-30. (a) The basis matrix is B =
5-29. −1 4 0 .500
(a) with B−1 = (b)
2 0 .250 .125
x2

6 v = (10, 0)B−1 = (0, 5) (c) c̄2 = c2 − v ·a(2)

= 2 - (0,5) · (1,6) = -28 which is not
5
improving for a maximize problem. c̄4 =
c4 − v ·a(4) = 0 - (0,5) · (21,-2) = 10 which is
improving for a maximize problem. (d) E =
4 x (2) = x* = (1,4)

1 .2 1 .050 .525
= EB 1
=
0 .2 −
, B−
.050 .025
3 5-31. (a) The basis matrix is B =
3 1 1/3 1/3
with B−1 = (b)
2 0 −1 0 −1
v = (0, 1)B−1 = (0, −1) (c) c̄1 = c1 − v ·a(1)
1 = 8 - (0,-1) · (13,-4) = 4 which is not

x(0) = x(1) = (0,0) c2 − v ·a(2) = -5 - (0,-1) · (2,1) = -4 which is

1 2 3 4 5 improving for a minimize problem.
)A =
(b 126 −33 01 10 , b = (18, 0) ,
c = (1, 0, 0, 0)

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 13 13

1 0
(d) E = , B− 1 = EB −1 = (b)
1 1
x1 x2 x3 x4 x5
1/3 1/3 max c 5 -10 0 0 0 b
1/3 −2/3 A 1 -1 2 0 4 2
5-32. 1 1 0 2 1 8
t=0 N N 1st 2nd N
x1 x2 x3 x4 x5 x(0) 0 0 1 4 0 0
min c 5 4 3 2 16 b 1/2 0 0
A 2 0 1 0 6 8 B−1 = v=
0
0 1/2
0 1 1 2 3 12 c̄ 5 -10 0 0 0
t=0 1st 2nd N N N Δx, x1 1 0 -.5 -.5 0
x(0) 4 12 0 0 0 68 2 8
0.5 0 2.5
B−1 = , v= t=1 1st N N 2nd N
3 0 10
0 1 4 x(1) 2 0 0

c̄ 0 0 -3.5 -6 -11 E= 2 0 , B−1 = 1 0

Δx, x5 -3 -3 0 0 1 −1 1 −1/2 1/2

4 12 v = (5, 0)
- - - c̄ 5 -10 0 0 0
3 3

t=1 N 2nd N N 1st

x(1) 0 9 0 0 4/3 57.33 Now x(1) is optimal because all nonbasic
c̄j ≤ 0 in a max problem.

B−1 and v as above. (b) See above. (c) x1 x2 x3 x4 x5

6 0
From above, new B = . min c 2 4 6 10 7 b
3 1 A 1 0 0 1 0 6
0 1 1 -1 2 9
.33 0

E = −1 1 producing new t=0 1st 2nd N N N

x(0) 6 9 0 0 0 48

B−1 = and v = (2/3, 4) B−1 =

−1/2 1 v=
0 1 4
5-33. (a) c̄ 0 0 2 12 -1
Δx, x5 0 -2 0 0 1
min c 0 1 0 1 1 b 4.5
x(1) 6 0 0 0 4.5 43.5
A -2 1 0 2 0 7 t=1 1st N N N1 2nd
1 0 0
0 0 0 1 1 5 E= , B−1 =
0 1 1 -1 0 3 0 1/2 0 1/2

t=0 1st N 2nd 3rd N v = (2, 7/2)

c̄ 0 1/2 5/2 23/2 0

x(0) 3/2 0 8 5 0 5

−1/2 1 0 0
B−1 = 0 1 1 ,v= 1 Now x(1) is optimal because all nonbasic
0 1 0 0

c̄ 0 1 0 0 0 c̄j ≥ 0 in a min problem.

5-34. (a) Nonbasic x5 is upper-bounded, so
x(0) is optimal because all nonbasic c̄j ≥ 0 in we adopt the negative of the direction. It has
a min problem. c̄5 = (3, −4, 1, −4, 10) · (−1, 1, 0, 0, −1) < 0
which is not improving for a maximize
problem. For the step size, λ+ =
min{(5 − 2)/1} = 3, λ− = min{2/1, 5/1} = 2,
giving λ = 2.

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 14 14

5-35. 5-36.
(a) (a)
z 2
z
8 2

6 z (2) =
z(0) = (6,5) (1,3)
3 z (3) = z* = (2.5,2.5)
z (1)= (0,3)
4
2

2 1
z (2) = z* = (3,0) z (0) = (0,0)
z(1) = (6,0)

2 4 8
1 3 1 0
(b) A = , b = (10, 5),
1 1 −1 0 1 1 0 1
(b) A = , b = (3, 8), c = (6, 8, 0, 0)
3 2 0 −1

c = (5, 6, 0, 0) (c) See Table 5.2.

(c) See Table 5.1.

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 15 15

z1 z2 z3 z4
min c 5 6 0 0 b
A 1 1 -1 0 3
3 2 0 -1 8
0 0 0 0
u 6 5 - -
t= 0 U U 1st 2nd
z(0) 6 5 8 20 60
−1 −1 0 0
B = v=
0 −1 0

c̄ -5 -6 0 0
Δz, z2 0 -1 -1 -2
5 8 20
- λ=5
1 1 2

t= 1 U L 1st 2nd
z(1) 6 0 3 10 30
−1 0 v= 0
B−1 =
0 −1 0

c̄ -5 6 0 0
Δz, z1 -1 0 -1 -3
6 3 10
- λ=3
1 1 3

t= 2 1st L L 2nd
z(2) 3 0 0 1 3
1 0 v = 5
B−1 =
3 −1 0

c̄ 0 1 5 0

Table 5.1: Lower-Upper Bounded Simplex Computation for Exercise 5-35

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 16 16

z1 z2 z3 z4
max c 6 8 0 0 b
A 1 3 1 0 10
1 1 0 1 5
0 0 0 0
u 4 3 - -
t= 0 L L 1st 2nd
z(0) 0 0 10 5 0
1 0 v = 0
B−1 = 0
0 1

c̄ 6 8 0 0
Δz, z2 0 1 -3 -1
3 10 5
- λ=3
1 3 1

t= 1 L U 1st 2nd
z(1) 0 3 1 2 24
1 0 v = 0
B−1 = 0
0 1

c̄ 6 -8 0 0
Δz, z1 1 0 -1 -1
4 1 2
- λ=1
1 1 1

t= 2 1st U L 2nd
z(2) 1 3 0 1 30
1 0 6
B−1 = v =
−1 1 0
c̄ 0 10 -6 0
Δz, z2 3 -1 0 -2
3 3 1
- λ = 0.50
3 1 2

t= 3 1st 2nd L L
z(3) 2.50 2.50 0 0 35
−0.50 1.50 v = 1
B−1 =
−0.50 0.50 5

c̄ 0 0 -1 -5

Table 5.2: Lower-Upper Bounded Simplex Computations for Exercise 5-36

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 17 17

5-37. (a) See Table 5.3 (c)

(b)
x1 x2 x3 x4 x5 x1 x2 x3 x4 x5
max c 5 -10 0 0 0 min c 2 4 6 10 7
A 1 -1 2 0 4 2 A 1 0 0 1 0 6
1 1 0 2 1 8 0 1 1 -1 2 9
0 0 0 0 0 0 0 0 0 0
u 2 2 2 2 2 u 4 4 4 4 4
t=0 U U L 1st 2nd t=0 U U U 1st 2nd
x(0) 2 2 0 7/4 1/2 -10

−1/8 1 0
B−1 =
1/2
, v = (0, 0) B−1 = , v = (27/2, 7/2)
1/4 0 1/2 1/2

c̄ 5 -10 0 0 0 c̄ -23/2 1/2 5/2 0 0

Δx, x2 0 -1 0 5/8 -1/4 Δx, x2 0 -1 0 0 1/2
2 2/5 2 4 5

x(1) 4 4 0 2 7/2 137/2

x(1) 2 8/5 0 2 2/5 -6
1 0
8/5 0 −1/5 4/5 B−1 = , v = (27/2, 7/2)
E=
2/5 1 , B−1 = 1/2 1/2
1/5 1/5

v = (2, −8) c̄ -23/2 1/2 5/2 0 0

c̄ 11 0 -4 16 0 Δx, x2 0 -1 0 0 1/2
4 1
No c̄j now indicates an opportunity to t=2 U 2nd L 1st U
improve, so x(1) is optimal. x(2) 4 3 0 2 4 68
1 0 1 0
E=
0 2 , B−1 =
1 1

v = (14, 4)
c̄ -12 0 2 0 -1

No c̄j now indicates an opportunity to improve,

so x(2) is optimal.

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
 18 18

x1 x2 x3 x4 x5
min c 0 1 0 1 1
A -2 1 0 2 0 7
0 0 0 1 1 5
0 1 1 -1 0 3
0 0 0 0 0
u 3 3 3 3 3
t=0 1st U 2nd U 3rd
x(0) 1 ⎛ 3 3 ⎞3 2 8
−1/2 0 0
B−1 = ⎝ 0 0 1 ⎠, v = (0, 1, 0)
0 1 0
c̄ 0 1 0 0 0
Δx, x2 -1/2 -1 1 0 0
2 3 0
t= 1 1st 2nd U U 3rd
1 3⎞ 3 ⎛3 2 8 ⎞
x(1) ⎛ 1 1/2 0 −1/2 0 1/2

E =⎝ 0 1 0 ⎠, B−1 = ⎝ 0 0 1 ⎠
0 0 1 0 1 0
v = (0, 1, 1)

c̄ 0 0 -1 1 0
Δx, x4 -3/2 -1 0 -1 1
2/3 3 3 1
t= 2 L 2nd U 1st 3rd
x(2)⎛ )0 7/3⎞ 3 7/3
⎛ 8/3 22/3 ⎞
−2/3 0 0 1/3 0 −1/3

E = ⎝ 2/3 1 0 ⎠, B−1 = ⎝ 1/3 0 2/3 ⎠

−2/3 0 1 −1/3 1 1/3

v = (1/3, 1, 2/3)
c̄ 2/3 0 -2/3 0 0

No c̄j now indicates an opportunity to improve, so x(2) is optimal.

Table 5.3: Lower-Upper Bounded Simplex Computations for Exercise 5-37a

© 2017 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently

exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.