Practice: The Matrix Form of a Linear Program

Recognition · Interpretation

A program has A x ≤ b with A x = ( 135 , 90 , 225 ) T and b = ( 120 , 90 , 250 ) T . Is the point feasible?

2 hints available, least help first.

Hint 1: Retrieval cue

How many separate statements does A x ≤ b stand for?

Hint 2: Concept cue

Compare the entries one position at a time.

Construction · Direct application · Interpretation · Explanation

A dairy blends two milks into three products. Per litre of product A: 0.6 L whole milk, 0.4 L skimmed, 2 minutes of line time. Product B: 0.3, 0.7, 3 minutes. Product C: 0.5, 0.5, 1 minute. Available each shift: 900 L whole milk, 700 L skimmed, 2,400 minutes of line time. Contribution per litre is £0.45, £0.30 and £0.38. At least 200 litres of product C must be produced to meet a contract.

(a) Assemble A , b and c , stating your variable and constraint order, and write the program in matrix form.

(b) Read row 3 back as a sentence, and say what column 2 tells you.

(c) State what A x ≤ b asserts about the entries.

(d) Explain how the contract requirement enters, given that the form takes ≤ rows.

Write your answer, then compare it with the worked solution.

3 hints available, least help first.

Hint 1: Retrieval cue

How many rows does A need, and how many columns?

Hint 2: Concept cue

Every row of A x ≤ b must be a ≤ relation. What does a stated minimum become?

Hint 3: Strategy cue

Assemble first, then translate one row and one column back into sentences and check them against the description.

Compare with the worked solution

Comparing does not record a result. Judging your own written answer cannot show that you can do this without help.

(a) Order and assembly. Variables x 1 , x 2 , x 3 are litres of products A, B, C per shift. Constraints ordered: whole milk, skimmed milk, line time, contract.

A = ( 0.6 0.3 0.5 0.4 0.7 0.5 2 3 1 0 0 − 1 ) , b = ( 900 700 2400 − 200 ) , c = ( 0.45 0.30 0.38 ) .

A is 4 × 3 : four constraints, three variables. The program is max c T x subject to A x ≤ b , x ≥ 0 ; as a minimisation, min ( − c ) T x with the value negated on the way out.

(b) Row 3 and column 2. Row 3 is ( 2 , 3 , 1 ) with b 3 = 2400 :

2 x 1 + 3 x 2 + x 3 ≤ 2400 ,

the line-time limit in minutes per shift. Column 2 is ( 0.3 , 0.7 , 3 , 0 ) T : one litre of product B uses 0.3 L whole milk, 0.7 L skimmed, 3 minutes of line time, and appears in the contract row with coefficient zero because the contract concerns product C only.

(c) What the relation asserts. Four scalar inequalities holding simultaneously, one per row. A plan failing any single row is infeasible, whatever the others do and whatever the totals are.

(d) The contract. It is a lower bound, x 3 ≥ 200 , and the form takes ≤ rows, so it is negated on both sides to − x 3 ≤ − 200 : the row ( 0 , 0 , − 1 ) with right-hand side − 200 . Entering it as ( 0 , 0 , 1 ) with 200 would assert a ceiling of 200 litres rather than a floor, which is a different program that happens to parse.

A check worth doing. Take x = ( 500 , 200 , 300 ) T . Then A x = ( 0.6 ( 500 ) + 0.3 ( 200 ) + 0.5 ( 300 ) , 0.4 ( 500 ) + 0.7 ( 200 ) + 0.5 ( 300 ) , 2 ( 500 ) + 3 ( 200 ) + 300 , − 300 ) T = ( 510 , 490 , 1900 , − 300 ) T , and every entry is at most the corresponding entry of b , so the plan is feasible. Contribution is 0.45 ( 500 ) + 0.30 ( 200 ) + 0.38 ( 300 ) = 225 + 60 + 114 = 399 .

A complete answer does each of these:

  • assembles with correct shapes
  • preserves every constraint
  • row reading
  • column reading
  • entrywise inequality

Representation translation · Interpretation · Explanation

A program has

A = ( 2 1 3 1 2 2 4 3 5 ) , b = ( 120 90 250 ) ,

for three products against machine hours, labour hours and material.

(a) State constraint 2 as a sentence, using the row reading.

(b) State what column 1 says about product 1, using the column reading.

(c) Evaluate A x at x = ( 10 , 15 , 20 ) T twice: once by rows, once as a combination of columns. Confirm the results agree.

(d) A colleague asks which products can be produced at all given the resources, and a second colleague asks whether a specific plan is allowed. Say which reading answers each, and why the other does not.

Write your answer, then compare it with the worked solution.

2 hints available, least help first.

Hint 1: Retrieval cue

A row is indexed by a constraint; a column is indexed by a variable.

Hint 2: Concept cue

Group the nine products of A x two ways: three per row, or three per column.

Compare with the worked solution

Comparing does not record a result. Judging your own written answer cannot show that you can do this without help.

(a) Row reading. Row 2 is ( 1 , 2 , 2 ) with b 2 = 90 : the plan uses x 1 + 2 x 2 + 2 x 3 labour hours, and at most 90 are available.

(b) Column reading. Column 1 is ( 2 , 1 , 4 ) T : one unit of product 1 consumes 2 machine hours, 1 labour hour and 4 kg of material. It is that product's resource profile across all three constraints.

(c) Both evaluations. By rows:

A x = ( 2 ( 10 ) + 1 ( 15 ) + 3 ( 20 ) 1 ( 10 ) + 2 ( 15 ) + 2 ( 20 ) 4 ( 10 ) + 3 ( 15 ) + 5 ( 20 ) ) = ( 95 80 185 ) .

By columns:

10 ( 2 1 4 ) + 15 ( 1 2 3 ) + 20 ( 3 2 5 ) = ( 95 80 185 ) .

The same vector, because the two are groupings of one set of nine multiplications.

(d) Which reading answers which question. The second colleague asks about a specific plan, which is a row question: form A x and compare entrywise with b . The first asks what combinations the resources permit, which is a column question: the reachable totals are the combinations x 1 A 1 + x 2 A 2 + x 3 A 3 with x ≥ 0 , and the constraint is that this combination stay under b .

The other reading does not answer each question because of what it hides. A row mixes all three products into one number, so it cannot isolate what one product contributes. A single column says nothing about whether a plan is allowed, because feasibility depends on every constraint at once and a column is one activity across all of them.

A complete answer does each of these:

  • row reading
  • column reading

Error diagnosis · Evaluation · Explanation

A description says: product 1 uses 2 machine hours and 4 kg; product 2 uses 1 machine hour and 3 kg; product 3 uses 3 machine hours and 5 kg. At most 120 machine hours and 250 kg are available, and total output must be at least 20 units.

A modeller submits

A = ( 2 1 3 4 3 5 1 1 1 ) , b = ( 120 250 20 ) ,

with A x ≤ b , and reports that x = ( 0 , 0 , 45 ) T is feasible because the entries of A x total less than the entries of b .

Identify every fault, and say what each one does to the program.

Write your answer, then compare it with the worked solution.

2 hints available, least help first.

Hint 1: Retrieval cue

Translate each row back into the sentence it came from. Do all three match?

Hint 2: Concept cue

Compute A x at the reported point and compare it with b one entry at a time.

Compare with the worked solution

Comparing does not record a result. Judging your own written answer cannot show that you can do this without help.

Fault 1: the third constraint has the wrong direction. The description sets a floor, x 1 + x 2 + x 3 ≥ 20 . Entered as row ( 1 , 1 , 1 ) with b 3 = 20 under A x ≤ b , it asserts a ceiling of 20 units. The feasible set is now bounded where the description bounded it below, and plans meeting the requirement are excluded. Correct it by negating both sides: row ( − 1 , − 1 , − 1 ) with right-hand side − 20 .

Fault 2: the feasibility claim uses the wrong relation. A x ≤ b is three scalar inequalities holding at once, not a comparison of totals. At x = ( 0 , 0 , 45 ) T ,

A x = ( 3 ( 45 ) 5 ( 45 ) 45 ) = ( 135 225 45 ) .

Row 1 gives 135 > 120 , so the point is infeasible on the machine-hours constraint. The totals happen to compare the other way, 405 against 390 , but that comparison decides nothing either way.

What is correct. The first two rows and their right-hand sides match the description, and the column order is consistent across rows: column 3 is ( 3 , 5 , 1 ) T , which is product 3's profile as stated.

The general lesson. Both faults produce a well-formed object. Nothing in the arithmetic reports them; translating each row back into the sentence it came from reports the first, and reading the relation entrywise reports the second.

A complete answer does each of these:

  • preserves every constraint
  • entrywise inequality
Practice data

Your practice record is stored in this browser only. Clearing it removes every answer and every scheduled review, and cannot be undone.

Results update as you type. Use the up and down arrow keys to move between results, Enter to open one, and Escape to close.

Type to search.

Settings

Appearance

Interface density

Your record

Your progress is stored in this browser and nowhere else: an identifier, the answers you have given, the mastery states and review schedule derived from them, and the lesson you last opened. Clearing it makes you a new learner on this device. It cannot be undone, and it will not affect your appearance or density settings.

Focus timer

Focus--minutes remaining

Phase

Kept in this browser only, and used to label the session in your own history.

Today

Nothing recorded yet. Finish a focus session and it will appear here.

Settings

Focus sessions between long breaks.

Sessions you are aiming for in a day.

Notifications

Your history

Sessions are stored in this browser and nowhere else. They are not evidence and never reach your mastery record.