Constrained optimization BUSINESS MATHEMATICS 1
CONTENTS Unconstrained optimization Constrained optimization Lagrange method Old exam question Further study 2
UNCONSTRAINED OPTIMIZATION Recall extreme values in two dimensions ππ βͺ first-order conditions: stationary points occur when ππ¦ = ππ ππ§ = 0 βͺ second-order conditions: extreme value when in addition 2 π 2 π π 2 π π 2 π ππ§ 2 β > 0 ππ¦ 2 ππ¦ππ§ βͺ nature of extreme value: π 2 π βͺ minimum when ππ¦ 2 > 0 π 2 π βͺ maximum when ππ¦ 2 < 0 3
UNCONSTRAINED OPTIMIZATION Recall Cobb-Douglas production function βͺ π πΏ, π = π΅πΏ π½ π πΎ Obviously (?) you want to maximize output ππΏ = π΅π½πΏ π½β1 π πΎ = 0 ππ βͺ so set ππ ππ = π΅πΏ π½ πΎπ πΎβ1 = 0 βͺ and This never happens! βͺ check! But you still want to maximize output βͺ within the constraints of your budget 4
CONSTRAINED OPTIMIZATION Suppose: βͺ the price of 1 unit of πΏ is π βͺ the price of 1 unit of π is π βͺ the available budget is π Budget line: ππΏ + ππ = π 5
CONSTRAINED OPTIMIZATION Iso-production line for π 1 Iso-production line for π 2 π Optimum point on line for optimum Budget line production π β π β π β π 2 π 1 πΏ β πΏ 6
CONSTRAINED OPTIMIZATION Problem formulation: βͺ ΰ΅ maximize π πΏ, π = π΅πΏ π½ π πΎ subject to ππΏ + ππ = π More in general α max π π¦, π§ s.t. π π¦, π§ = π Constrained optimization problem βͺ π π¦, π§ is the objective function βͺ π π¦, π§ = π is the constraint βͺ π¦ and π§ are the decision variables 7
EXERCISE 1 Given is the profit function π π, π = 5π 2 + 12π 2 and the capacity constraint 8π + 4π = 20 . Write as αmin/max subject to . 8
CONSTRAINED OPTIMIZATION Visualisation of π¨ = π π¦, π§ βͺ in 3D βͺ as level curve 10
LAGRANGE METHOD How to solve the constrained optimization problem? α max π π¦, π§ s.t. π π¦, π§ = π Trick: βͺ introduce extra variable (Lagrange multiplier) π βͺ define new function (Lagrangian) β π¦, π§, π βͺ as follows β π¦, π§, π = π π¦, π§ β π π π¦, π§ β π Observe that β is a function of 3 variables, while the objective function π is a function of 2 variables 11
LAGRANGE METHOD Motivation: βͺ solutions of the constrained optimization are among the stationary points of β βͺ in other words: all stationary points of β are candidate solutions of the original constrained maximization problem Lagrange method βͺ Joseph-Louis Lagrange (1736-1813) 12
LAGRANGE METHOD So, α max π π¦, π§ π π¦, π§ = π is equivalent to max β π¦, π§, π s.t. βͺ if we define β π¦, π§, π = π π¦, π§ β π π π¦, π§ β π We have written the constrained optimization problem with 2 decision variables as an unconstrained optimization problem with 3 decision variables βͺ trade-off βͺ unconstrained easier than constrained βͺ 3 decision variables harder than 2 13
EXERCISE 2 Given is the profit function π π, π = 5π 2 + 12π 2 and the capacity constraint 8π + 4π = 20 . Write the Lagrangian. 14
LAGRANGE METHOD Example 1 Constrained problem: π π¦, π§ = π¦ 2 + π§ 2 βͺ ΰ΅ max π π¦, π§ = π¦ 2 + π¦π§ + π§ 2 = 3 s.t. Lagrangian: βͺ β π¦, π§, π = π¦ 2 + π§ 2 β π π¦ 2 + π¦π§ + π§ 2 β 3 First-order conditions for stationary points: πβ βͺ ππ¦ = 0 πβ βͺ ππ§ = 0 πβ βͺ ππ = 0 16
LAGRANGE METHOD Example 1 (continue) πβ βͺ ππ¦ = 2π¦ β π 2π¦ + π§ = 0 πβ βͺ ππ§ = 2π§ β π π¦ + 2π§ = 0 πβ ππ = βπ¦ 2 β π¦π§ β π§ 2 + 3 = 0 βͺ So: 2π¦ π¦+2π§ β 2π¦ 2 + 4π¦π§ = 4π¦π§ + 2π§ 2 β π¦ 2 = π§ 2 2π§ βͺ 2π¦+π§ = βͺ βπ¦ 2 Β± π¦ 2 β π¦ 2 + 3 = 0 β 3π¦ 2 = 3 β¨ π¦ 2 = 3 βͺ π¦ = β1 β¨ π¦ = 1 β¨ π¦ = 3 β¨ π¦ = β 3 βͺ π§ follows, and so does π 17
LAGRANGE METHOD Example 1 (continued) Four stationary points π¦, π§, π of Lagrangian: 3 , β1, β1, 2 3 , 3, β 3, 2 and β 3, βͺ 1,1, 2 3, 2 βͺ check! Four other points π¦, π§, π do not satisfy all equations: 1, β1,2 , β1,1,2 , 3 and β 3, β 3, 2 βͺ 3, 2 3, 3 βͺ check! So, π¦, π§ = 1,1 , β1, β1 , 3, β 3 and 3 are candidate solutions to the original β 3, constrained problem 18
LAGRANGE METHOD Example 1 (continued) Determine nature of these points (min/max/saddle) and then determine the maximum βͺ π 1,1 = 1 2 + 1 2 = 2 βͺ π β1, β1 = β1 2 + β1 2 = 2 2 + β 3 2 = 6 βͺ π 3, β 3 = 3 2 + 2 = 6 βͺ π β 3, 3 = β 3 3 So: 3, β 3 and β 3, 3 are maximum points βͺ βͺ and (1,1) and β1, β1 are mimimum points 19
LAGRANGE METHOD Example 2 Constrained problem: βͺ ΰ΅ maximize π πΏ, π = π΅πΏ 0.4 π 0.7 (π΅ > 0) subject to 25πΏ + 10π = π (π > 0) Lagrangian βͺ β πΏ, π, π = π΅πΏ 0.4 π 0.7 β π 25πΏ + 10π β π First-order conditions for stationary points πβ ππΏ = 0.4π΅πΏ β0.6 π 0.7 β 25π = 0 βͺ πβ ππ = 0.7π΅πΏ 0.4 π β0.3 β 10π = 0 βͺ πβ βͺ ππ = β 25πΏ + 10π β π = 0 20
LAGRANGE METHOD Example 2 (continued) 0.4π΅πΏ β0.6 π 0.7 0.7π΅πΏ 0.4 π β0.3 βͺ π = = 25 10 βͺ multiply both sides with πΏ 0.6 (to get rid of πΏ β0.6 ) and with π 0.3 (to get rid of π β0.3 ) 0.4π΅π 0.7π΅πΏ 10Γ0.4 βͺ = β 25πΏ = π 25 10 0.7 10Γ0.4 0.7 π 0.4 π 10 and πΏ = βͺ π + 10π = π β π = 0.7 1.1 1.1 25 βͺ π = β― (not intetesting) 0.4 π 0.7 π βͺ Candidate maximum point: πΏ, π = 25 , 1.1 1.1 25 21
LAGRANGE METHOD Example 2 (continued) Determine nature of the stationary point of Lagrangian: π βͺ take πΏ = 0 β π = 10 and see that π = 0 π βͺ likewise take π = 0 β πΏ = 25 and see that π = 0 0.4 π π 0.7 π π βͺ so for 0 < 25 and 0 < 10 , also π > 0 25 < 10 < 1.1 1.1 0.4 π 0.7 π βͺ therefore πΏ, π = 10 is a maximum 25 , 1.1 1.1 Here, we look at two β neighbouring β points of the stationary point: one βto the leftβ and one βto the rightβ. We find that both have a lower function value, so the stationary point is a maximum. 22
LAGRANGE METHOD Different point on the same budget Optimal Cobb- line with π < π β Douglas line π β = π΅πΏ 0.4 π 0.7 π Optimum point on Budget line line for optimum 25πΏ + 10π = π production π β π β = 0.7 π Different point on 1.1 10 the same budget line with π < π β πΏ β = 0.4 π πΏ 1.1 25 23
LAGRANGE METHOD Full procedure: βͺ rewrite constrained optimization problem in 2D as an unconstrained optimization problem in 3D (from α max π π¦, π§ s. t. π π¦, π§ = π to max β π¦, π§, π ) Do not use 2 π 2 β π 2 β π 2 β ππ§ 2 β πβ ππ¦ 2 ππ¦ = 0 ππ¦ππ§ for this! πβ βͺ find π¦ and π§ such that ππ§ = 0 πβ ππ = 0 βͺ check if the stationary points are indeed a maximum 24
OLD EXAM QUESTION 22 October 2014, Q2a 25
OLD EXAM QUESTION 27 March 2015, Q2bβ 26
FURTHER STUDY Sydsæter et al. 5/E 14.1-14.2 Tutorial exercises week 5 Lagrange method Lagrange for Cobb-Douglas Intuitive idea of Lagrange 27
Recommend
More recommend