Review Material, Exam 1, Ops Research

The exam will cover material from Chapter 3 (we skipped 3.6, 3.7), up through section 4.8. Section 4.9 may
be used to help with LINDO, and we skipped 4.10. We may cover 4.11 on Monday, in which case that material
may be on the exam (it is a very short section).

Background Material: Linear Algebra

Though these may not be asked explicitly, you should be able to do the following (and may be part of solutions):
1. Be able to solve Ax = b for different types of matrices (if A is square and invertible, and if it is tall and
full rank, tall and not full rank, wide and full rank, wide and not full rank).
2. Be able to discuss the solution to Ax = b in terms of the row space, the column space and the null space
of A.
3. Be able to compute the determinant and inverse of a matrix A
4. Note that our book uses “ERO” for elementary row operation.
5. Define what it means for a set of vectors to be linearly independent.
6. Be able to solve for a basis for the row space, column space and null space for a given matrix A.

Definitions:
Linear program, standard form of a linear program, objective function, constraints (binding and non-binding),
extreme point, isoprofit line, feasible region, unrestricted variable (URS), BFS, adjacent BFS, slack/surplus (or
excess) variables, basic solution, BFS, BV, NBV, direction of unboundedness, Convex set, convex combination.
The word “degenerate” is used in two distinct cases: See the top of p. 134 for what it means for a linear
program to be degenerate. In class, we said that a BFS is degenerate if one of the basic variables (for the
optimal solution) is zero.

Skills
1. Translate unrestricted variables so that all variables are non-negative.
2. Be able to show that a given set is convex, use convexity in other arguments (See review questions below
for examples).
3. Be able to set up and solve an LP graphically. Given the objective function, be able to state the direction
in which the function increases (or decreases) the fastest.
4. Be able to set up an LP generally, and of specific types: A diet problem, a work scheduling problem, a
production process model, blending, and multiperiod problems (like the sailboat example).
5. Be able to draw a diagram for the production process models (like we did in class for “Brute and
Chanele”, and in the homework for the dairy farm).
6. Be able to translate a LP into standard form give “less than or equal to” constraints, “greater than or
equal to” constraints, and change the variables so that they are all non-negative.
7. Be able to write the simplex tableau from the linear program (for both a maximization and a minimization
problem).
8. Be able to solve a linear program using the Simplex Method.
9. Given a tableau, be able to tell if it is a terminal tableau, and interpret what the solution is (unique,
multiple, unbounded).
10. (For something that might be take home) Be able to use either LINDO or a spreadsheet program to
solve an LP.

1
Theorems: See the handout from class, the theorems from Chapter 4.

Review Questions
1. What are the four assumptions we make when we write a linear program? Give a short explana-
tion/illustration of each.
2. What are the four possible outcomes when solving a linear program? Hint: The first is that there is a
unique solution to the LP.
3. The following are to be sure you understand the process of constructing a linear program:
(a) Draw a production process diagram and set up the LP for Exercise 6, p. 98 (Sect. 3.9).
(b) Exercise 2, 31 Chapter 3 review (Be sure you can solve an LP graphically)
(c) Exercise 6, 18 Chapter 3 review (A ton is 2000 lbs)
(d) Exercise 22, Chapter 3 review. Hint: Consider using a triple index on your variables.
(e) Exercise 47, 53 in Chapter 3 review.
4. Convert the following LP to one in standard form. Write the result in matrix-vector form, giving x,
c, A, b (from our formulation).
min z = 3x − 4y + 2z
st 2x − 4y ≥ 4
x + z ≥ −5
y+z ≤1
x+y+z =3
with x ≥ 0, y is URS, z ≥ 0.
5. Consider again the “Wyndoor” company example we looked at in class:

min z = 3x1 + 5x2

st x1 ≤ 4
2x2 ≤ 12
3x1 + 2x2 ≤ 18

with x1 , x2 both non-negative.

(a) Rewrite so that it is in standard form.
(b) Let s1 , s2 , s3 be the extra variables introduced in the last answer. Is the following a basic solution?
Is it a basic feasible solution?

x1 = 0, x2 = 6, s1 = 4, s2 = 0, s3 = 6

Which variables are BV, and which are NBV?

(c) Find the basic feasible solution obtained by taking s1 , s3 as the non-basic variables.
6. Consider Figure 1, with points A(1, 1), B(1, 4) and C(6, 3), D(4, 2) and E(4, 3).

• Write the point E as a convex combination of points A, B and C.

• Can E be written as a convex combination of A, B and D? If so, construct it.
• Can A be written as a linear combination of A, B and D? If so, construct it.

7. Draw the feasible set corresponding to the following inequalities:

x1 + x2 ≤ 6, x1 − x2 ≤ 2 x1 ≤ 3, x2 ≤ 6

with x1 , x2 non-negative.
(a) Find the set of extreme points.

2
 Figure 1: Figure for the convex combinations, Exercise 6.

(b) Write the vector [1, 1]T as a convex combination of the extreme points.
(c) True or False: The extreme points of the region can be found by making exactly two of the
constraints binding, then solve.
(d) If the objective function is to maximize 2x1 + x2 , then (a) how might I change that into a mini-
mization problem, and (b) solve it.
8. Consider the unbounded feasible region defined by
x1 − 2x2 ≤ 4, −x1 + x2 ≤ 3
with x1 , x2 non-negative. Consider the vector p = [5, 2].
(a) Show that p is in the feasible region.
(b) Set up the system you would solve in order to write p in the form given in Theorem 2 above (provide
a specific vector d).
9. Finish the definition: Two basic feasible solutions are said to be adjacent if:
10. Let d be a direction of unboundedness. Using the definition, prove that this means that rd is also a
direction of unboundedness, for any constant r ≥ 0.
11. If C is a convex set, then d 6= 0 is a direction of unboundedness for C iff x + d ∈ C for all x ∈ C (Use
the definition of unboundedness).
12. For an LP in standard form (see above), prove that the vector d is a direction of unboundedness iff
Ad = 0 and d ≥ 0.
13. Show that the set of optimal solutions to an LP (assume in standard form) is convex.
14. Let a feasible region be defined by the system of inequalities below:
−x1 + 2x2 ≤6
−x1 + x2 ≤2
x2 ≥1
x1 , x2 ≥ 0
The point (4, 3) is in the feasible region. Find vectors d and b1 , · · · bk and constants σi so that the
Representation Theorem is satisfied (NOTE: Your vector x from that theorem is more than two dimen-
sional).
15. Let a feasible region be defined by the system of inequalities below:
−x1 + x2 ≤2
x1 − x2 ≤1
x1 + x2 ≤5
x1 , x2 ≥ 0
The point (2, 2) is in the feasible region. Find vectors d and b1 , · · · bk and constants σi so that the
Representation Theorem is satisfied (NOTE: Your vector x from that theorem is more than two dimen-
sional).

3
 n
X
16. Suppose that λ1 ≤ λ2 ≤ · · · λn , and let σ1 , · · · , σn be non-negative constants so that σi = 1. Show
i=1
that
λ1 ≤ σ1 λ1 + σ2 λ2 + · · · σn λn ≤ λn

17. Show that, if x is in the convex hull of vectors b1 , · · · bk , then for any constant vector c,

cT x ≤ max cT bi

i

18. True or False, and explain: The Simplex Method will always choose a basic feasible solution that is
adjacent to the current BFS.
19. Given the current tableau (with variables labeled above the respective columns), answer the questions
below.
x1 x2 s1 s2 rhs
0 −1 0 2 24
0 1/3 1 −1/3 1
1 2/3 0 1/3 4

(a) Is the tableau optimal (and did your answer depend on whether we are maximizing or minimizing)?
For the remaining questions, you may assume we are maximizing.
(b) Give the current BFS.
(c) Directly from the tableau, can I increase x2 from 0 to 1 and remain feasible? Can I increase it to
4?
(d) If x2 is increased from 0 to 1, compute the new value of z, x1 , s1 (assuming s2 stays zero).
(e) Write the objective function and all variables in terms of the non-basic (or free) variables, and then
put them in vector form.