Two-stage Simplex and big-M methodsEdexcel A-Level Further Maths: Revision notes
Section 1
Why artificial variables cause trouble
For a constraint the slack variable gives a ready-made starting solution. A or constraint does not: for the equation at gives , which is not allowed. So we add an artificial variable : , which starts with . But has no meaning in the real problem. A genuine solution must have . The big-M and two-stage methods are two ways of forcing artificial variables out of the solution. Summary: gives a slack variable; gives a surplus and an artificial variable; gives an artificial variable.
Adding a surplus variable and nothing else: the origin is then infeasible so the Simplex algorithm cannot start.
Section 2
The big-M method
Use one objective that penalises artificial variables. For a maximising problem subtract , where is a very large positive number: Since is basic, its column must be a unit column, so eliminate from the row: subtract the row. For and : Choose the pivot column by comparing the coefficients of first (the most negative for large ): here . Ratios: and , so the row is the pivot row.
Forgetting to eliminate from the row before the first iteration.
Keep as a symbol, and compare the coefficients of first when choosing the pivot column.
Section 3
Completing the big-M example
After the first pivot ( row, column) the row is and the row is . Now has left the basis and appears only in the column. The most negative entry is (the column). Only the row has a positive entry (, ratio 18), so it is the pivot row. The result is the row , with no negatives. So , , (check: , ). If the final tableau still has , the problem has no feasible solution.
Once is non-basic you can ignore its column; in the column no longer affects the pivot choice.
Section 4
The two-stage method
Stage 1 ignores profit and just finds a feasible start. Define (the sum of the artificial variables) and minimise it, i.e. maximise . Eliminate to get the row. For and : . Carry the row (written as usual) through stage 1 so it stays correct.
- If the minimum of is 0: a basic feasible solution exists, so go to stage 2.
- If the minimum of is positive: no feasible solution. Stage 2: delete the row and the artificial column(s), and continue Simplex on the original row until no negatives remain.
Dropping the row in stage 1: it must be updated by the same row operations.
In stage 1 the objective row to pivot on is the row, not the row.
Section 5
Two-stage worked example and minimising
Maximise subject to and . Stage 1: pivot on the 3 in the row ( column; ratios 10 and 2), giving and . The row becomes . Stage 2: enters (ratios 6 and 12, so the row leaves), then enters (the row leaves). Final row : , , . Minimising: maximise . For minimise with and , use (big-M) or then (two-stage). The result is , and .
Big-M keeps one tableau with a symbolic ; two-stage uses only numbers, which is easier to calculate by hand.
That's the notes covered.
Carry on to the next subtopic.
Exam questions on Two-stage Simplex and big-M methods
- A maximising problem has the constraint , to which a surplus variable and an artificial variable are added. The objective is and the big-M method is used, where is a very large positive number.Explain the purpose of the term , and what it means if in the final optimal tableau.2 marks
- A maximising problem has the constraints and , with . It is solved using the two-stage Simplex method, with a slack variable in the first constraint and a surplus variable and an artificial variable in the second.Explain, without using the Simplex algorithm, why the stage 1 result is correct.2 marks
- A company makes tonnes of product X and tonnes of product Y. It wishes to maximise the profit subject to and , with and . The big-M method is used, with slack variable , surplus variable and artificial variable . Tableau columns are in the order , , , , , then the value.Write the constraints as equations and write down the objective function for the big-M method.3 marks
Written by the Exaim team, led by Shaun Daswani (Head of Upper Secondary, Improve ME Institute; MSc Financial Mathematics, Imperial College London; BSc, UCL) and Jason Daswani (operational lead, Improve ME Institute; LSE).