MATH39011 Mathematical Programming I Example Sheet 1 1.

By graphical means minimize z (x1 ; x2 ) = subject to x1 + x2 10 41 8 0

x1 + 2x2 5x1 + 3x2 2x1 3x2 x1 ; x2

Identify all the corners (vertices) of the feasible region and the optimal vertex. What is the minimum value of the LP? By adding slack variables, write down the constraints in standard form Ax = b Identify the basic variables xB at the optimum and the optimal basis B . Show that xB = B
1

2. Let x be any basic feasible solution (BFS) of the system Ax = b x 0

corresponding to some basis B: Show that the LP minimization problem with cost z = cT x where c is dened by 0 if Aj 2 B cj = 1 if Aj 2 =B has the unique minimum solution at x : [Hint: Observe that z (x) 0 for any feasible x and that z (x ) = 0: To show uniqueness, suppose that y is also optimal and deduce that y = x :] 3. Write the following LP problem in standard form: min/max z = 2x1 + x2 subject to x1 + x2 + 2x3 x1 + 4x2 x3 x1 ; x2 ; x3 x3 6 4 0

and solve by evaluating the objective function at all basic feasible solutions. What bounds on x1 ; x2 and x3 are implied by constraint 1? If the rst constraint is replaced by x1 + x2 2x3 6 show that x3 is no longer bounded and deduce that enumeration of vertices no longer results in a minimum point.

4. [The renery problem] An oil renery can buy two types of oil: light crude oil and heavy crude oil. The costs per barrel of each type of crude are $11 (L) and $9 (H) respectively. The following quantities (mixes) of gasoline, kerosene and jet fuel are produced per barrel of each type of oil. GASOLINE 0.4 0.32 KEROSENE 0.2 0.4 JET FUEL 0.35 0.2

Light crude oil Heavy crude oil

[Note that 5% and 8% of the crude are lost respectively during the rening process, though this doesn t aect the formulation.] The renery has contracted to deliver 1 million barrels of gasoline, 400,000 barrels of kerosene and 250,000 barrels of jet fuel. Find the number of barrels of each crude oil that should be rened in order to full the order at minimum cost. 5. Consider the following problem in which x2 is a free variable minimize subject to z = x1 3x2 + 4x3 x1 2x2 + x3 = 5 2x1 3x2 + x3 = 6 x1 0; x3 0

Show by substituting for x2 from the rst constraint, that we can obtain a 2-variable LP problem in standard form which can be solved graphically. (An alternative approach is to substitute x2 = u2 v2 where u2 ; v2 0:) 6. Solve the following "continuous knapsack" problem by inspection, and justify your solution in terms of the basic solutions of the problem in standard form. Maximize subject to 5x1 6x2 + 3x3 5x4 + 12x5 90 0

x1 + 3x2 + 5x3 + 6x4 + 3x5 x1 ; x2 ; x3 ; x4 x5

7. Show that the hyperplane X = x 2 Rn : pT x = k has no extreme points. [Hint: Consider adding y to x 2 Rn where pT y = 0:] (a) Show that X = f(x1 ; x2 ) : x1 + x2 1g R2 has no extreme points. [Hint: Consider the eect of adding ("; ") to any x 2 X ]. x1 = u1 x2 = u2 v1 v2

(b) Consider the transformation to new non-negative variables u1 ; v1 ; u2 ; v2 :

and show that the new feasible region U does have extreme points e.g. (u1 ; v1 ; u2 ; v2 ) = (1; 0; 0; 0) :