D2 Linear Programming - Simplex Algorithm
D2 Linear Programming - Simplex Algorithm
com
(a) Taking the most negative number in the profit row to indicate the pivot column, perform
one complete iteration of the simplex algorithm, to obtain tableau T. State the row
operations that you use.
Basic Row
x y z r s Value
variable operations
Basic Row
x y z r s Value
variable operations
Basic Row
x y z r s Value
variable operations
(5)
(c) State whether tableau T is optimal. Give a reason for your answer.
(1)
(Total 7 marks)
2. The tableau below is the initial tableau for a linear programming problem in x, y and z. The
objective is to maximise the profit, P.
(a) Write down the profit equation represented in the initial tableau.
P
(1)
(b) Taking the most negative number in the profit row to indicate the pivot column at each
stage, solve this linear programming problem. Make your method clear by stating the row
operations you use.
(9)
(c) State the final value of the objective function and of each variable.
(3)
(Total 13 marks)
3. While solving a maximising linear programming problem, the following tableau was obtained.
4. A two person zero-sum game is represented by the following pay-off matrix for player A.
(a) Formulate the game as a linear programming problem for player A, writing the constraints
as equalities and clearly defining your variables.
(5)
(b) Explain why it is necessary to use the simplex algorithm to solve this game theory
problem.
(1)
(c) Write down an initial simplex tableau making your variables clear.
(2)
(d) Perform two complete iterations of the simplex algorithm, indicating your pivots and
stating the row operations that you use.
(8)
(Total 16 marks)
1. (a)
b.v x y z r s Value Raw ops
1 1
z 0 1 0 20 R1 ÷ 4 M1 A1
2 4
1
s 0 4 0 − 1 120 R2 – 2R1 M1 A1 ft
2
P 8 –8 0 5 0 400 R3 + 20R1 A1 ft 5
(c) Not optimal since there is a negative number in the profit row B1ft 1
[7]
2. (a) P − x − 2 y − 6z =0 B1 1
Note
1B1: cao
(b)
b.v x y z r s t Value
r 0 1 2 1 0 0 24
s 2 1 4 0 1 0 28
1
t –1 2 3 0 0 1 22
P –1 –2 –6 0 0 0 0
z 1
2
1
4
1 0 1
4
0 7 R2 ÷ 4 M1 A1ft
t – 52 – 14 0 0 – 34 1 1 R3 – 3R2 A1
P 2 – 12 0 0 3
2
0 42 R4 + 6R2 5
z 1 0 1 – 12 1
2 0 2 R2 – 14 R1 M1 A1
1
t –3 0 0 2 –1 1 6 R3 + 14 R1
P 1 0 0 1 1 0 52 R4 + 12 R1 4
Notes
1M1: correct pivot located, attempt to divide row
1A1: pivot row correct including change of b.v.
2M1: (ft) Correct row operations used at least once or stated correctly.
1A1ft: Looking at non zero-and-one columns, one column ft correct
2A1: cao.
3M1: (ft) Correct pivot identified – negative pivot gets M0 M0
1A1: ft pivot row correct including change of bv – but don’t penalise
b.v. twice.
4M1: (ft) Correct row operations used at least once or stated correctly.
1A1: cao
Misread Alternative 1
Increasing x first,
b.v. x y z r s t Value row ops
r 0 1 2 1 0 0 24 R1 no change
1 1
x 1 2 2 0 2 0 14 R2 ÷ 2
1
t 0 1 5 1 2 1 36 R3 + R2
P 0 − 32 –4 0 1
2 0 14 R4 + R 2
then y next,
b.v. x y z r s t Value row ops
y 0 1 2 1 0 0 24 R1 ÷ 1
x 1 0 1 − 12 1
2 0 2 R2 – 1
2
R1
1
t 0 0 3 –1 2 1 12 R3 – R1
3 1 3
P 0 0 –1 2 2 1 50 R4 + 2
R1
then z.
b.v. x y z r s t Value row ops
y –2 1 0 2 –1 0 20 R1 – 2R2
z 1 0 1 − 12 1
2 0 2 R2 ÷ 2
1
t –3 0 0 2 –1 1 6 R3 – 3R2
P 0 0 0 1 1 1 52 R4 + R2
Misread Alternative 2
Increasing x first,
b.v. x y z r s t Value row ops
r 0 1 2 1 0 0 24 R1 no change
1 1
x 1 2 2 0 2 0 14 R2 ÷ 2
1
t 0 1 5 0 2 1 36 R3 + R2
P 0 − 32 –4 0 1
2 0 14 R4 + R 2
Increasing z next,
b.v. x y z r s t Value row ops
r –1 1
2 0 1 − 12 0 10 R1 – 2R2
1 1 1
z 2 4 1 0 4 0 7 R2 ÷ 2
t − 52 − 14 0 0 3
4 1 1 R3 – 5R2
P 2 − 12 0 0 3
2 0 42 R4 + 4R2
then increasing y.
b.v. x y z r s t Value row ops
1
y –2 1 0 2 –1 0 20 R1 ÷ 2
z 1 0 1 − 12 1
2 0 2 R2 – 1
4
R1
1 1
t –3 0 0 2 –1 1 6 R3 – 4
R1
1
P 1 0 0 1 1 0 52 R4 + 2
R1
Misread Alternative 3
Increasing y first,
b.v. x y z r s t Value row ops
y 0 1 2 1 0 0 24 R1 ÷ 1
s 2 0 2 –1 1 0 4 R2 – R1
t –1 0 2 − 12 0 1 10 R3 − 12 R1
P –1 0 –2 2 0 0 48 R4 + 2R1
Increasing x next,
b.v. x y z r s t Value row ops
y 0 1 2 1 0 0 24 R1 no change
x 1 0 1 − 12 1
2 0 2 R2 ÷ 2
1
t 0 0 3 –1 2 1 12 R3 – 3R2
3 1
P 0 0 –1 2 2 0 50 R4 + R 2
then increasing z.
b.v. x y z r s t Value row ops
y –2 1 0 2 –1 0 20 R1 – 2R2
z 1 0 1 − 12 1
2 0 2 R2 ÷ 1
1
t –3 0 0 2 –1 1 6 R3 + R2
P 1 0 0 1 1 0 52 R4 + R2
Misread Alternative 4
Increasing y first,
b.v. x y z r s t Value row ops
y 0 1 2 1 0 0 24 R1 ÷ 1
s 2 0 2 –1 1 0 4 R2 – R1
t –1 0 2 − 12 0 1 10 R3 − 12 R1
P –1 0 –2 2 0 0 48 R4 + 2R1
z 1 0 1 − 12 1
2 0 2 R2 ÷ 2
1
t –3 0 0 2 –1 1 6 R3 – 2R2
P 1 0 0 1 1 0 52 R4 + 2R2
3. (a) x = 0, y = 0, z = 2 B2 1 0 2
Note
1B1: Any 2 out of 3 values correct
2B1: All 3 values correct.
5
(b) P – 2 x – 4 y + r = 10 M1 A1 2
4
Note
1M1: One equal sign, modulus of coefficients
correct. All the right ingredients.
1A1: cao – condone terms of zero coefficient
[4]
(c) e.g.
bv V P1 P2 P3 r s t u value
r 1 –5 –3 –6 1 0 0 0 0 M1
s 1 –7 –8 –4 0 1 0 0 0 A1
t 1 –2 –4 –9 0 0 1 0 0 2
u 0 1 1 1 0 0 0 1 1
P –1 0 0 0 0 0 0 0 0
V 1 –5 –3 –6 1 0 0 0 0 R1 / 1 M1 A1
s 0 –2 –5 2 –1 1 0 0 0 R2 – R1 A1
t 0 –3 –1 –3 –1 0 1 0 0 R3 – R1 B1ft
u 0 1 1 1 0 0 0 1 1 R4 stet 4
P 0 –5 –3 –6 1 0 0 0 0 R 5 + R1
P1 P2 P
bv V r s t u value Row ops
3
2. This generated a good spread of marks and differentiated well. The majority of candidates
correctly stated the objective function, although some had the signs reversed. Most were able to
select the first pivot correctly and went on to produce a good first tableau, with the occasional
arithmetic error. The choice of the second pivot caused problems, with many candidates
choosing – 14 . A negative number may never be selected as a pivot. Many candidates realised
they had made an error as negative numbers started to appear for the basic variables and
stopped, but others gave negative values for their final answer. Candidates should be reminded
that they should state the values of all 7 variables at the end; many only gave the four non-zero
values.
3. This question caused problems for most candidates. Very few were able to read off the correct
values, or to write down the correct objective function, expressions with two equal signs being
the most common error here.
4. Many candidates struggled to answer this question and those that did make a decent attempt
encountered difficulties and made a succession of errors. For those candidates who made a
reasonable attempt at the question, by far the most popular approach was to divide the
probabilities by the value of the game. Unfortunately this then created additional difficulties for
the candidate, as for player A it was then necessary to minimise, which a large number of
candidates failed to state. Candidates also either failed to turn their inequalities into equations or
added slack variables when, in this instance, they should have been subtracted. Candidates also
failed to define their probabilities etc. Most candidates who tried to answer the question in this
way then used the equations for player A in their simplex tableau, not realizing that they needed
to change these to player B’s perspective to allow them to maximise. Other candidates who
started in the same manner, incorrectly set up their equations (inequalities) from B’s
perspective, but were then able to use these in their tableau. A minority of candidates adopted
one of the other approaches and these candidates were generally more successful. Many
candidates failed to mention that simplex was necessary because it was a 3 x 3 problem and it
could not be reduced by dominance arguments. Those candidates who reached a correct initial
tableau were generally able to manipulate this correctly, although minor errors, either arithmetic
or omitting the change of base variable, did occur. Some candidates failed to state the row
operations that they had used.