mathematics
play

MATHEMATICS 1 CONTENTS Unconstrained optimization Constrained - PowerPoint PPT Presentation

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


  1. Constrained optimization BUSINESS MATHEMATICS 1

  2. CONTENTS Unconstrained optimization Constrained optimization Lagrange method Old exam question Further study 2

  3. UNCONSTRAINED OPTIMIZATION Recall extreme values in two dimensions πœ–π‘” β–ͺ first-order conditions: stationary points occur when πœ–π‘¦ = πœ–π‘” πœ–π‘§ = 0 β–ͺ second-order conditions: extreme value when in addition 2 πœ– 2 𝑔 πœ– 2 𝑔 πœ– 2 𝑔 πœ–π‘§ 2 βˆ’ > 0 πœ–π‘¦ 2 πœ–π‘¦πœ–π‘§ β–ͺ nature of extreme value: πœ– 2 𝑔 β–ͺ minimum when πœ–π‘¦ 2 > 0 πœ– 2 𝑔 β–ͺ maximum when πœ–π‘¦ 2 < 0 3

  4. UNCONSTRAINED OPTIMIZATION Recall Cobb-Douglas production function β–ͺ π‘Ÿ 𝐿, 𝑀 = 𝐡𝐿 𝛽 𝑀 𝛾 Obviously (?) you want to maximize output πœ–πΏ = 𝐡𝛽𝐿 π›½βˆ’1 𝑀 𝛾 = 0 πœ–π‘Ÿ β–ͺ so set πœ–π‘Ÿ πœ–π‘€ = 𝐡𝐿 𝛽 𝛾𝑀 π›Ύβˆ’1 = 0 β–ͺ and This never happens! β–ͺ check! But you still want to maximize output β–ͺ within the constraints of your budget 4

  5. CONSTRAINED OPTIMIZATION Suppose: β–ͺ the price of 1 unit of 𝐿 is 𝑙 β–ͺ the price of 1 unit of 𝑀 is π‘š β–ͺ the available budget is 𝑛 Budget line: 𝑙𝐿 + π‘šπ‘€ = 𝑛 5

  6. CONSTRAINED OPTIMIZATION Iso-production line for π‘Ÿ 1 Iso-production line for π‘Ÿ 2 𝑀 Optimum point on line for optimum Budget line production π‘Ÿ βˆ— 𝑀 βˆ— π‘Ÿ βˆ— π‘Ÿ 2 π‘Ÿ 1 𝐿 βˆ— 𝐿 6

  7. CONSTRAINED OPTIMIZATION Problem formulation: β–ͺ ࡝ maximize π‘Ÿ 𝐿, 𝑀 = 𝐡𝐿 𝛽 𝑀 𝛾 subject to 𝑙𝐿 + π‘šπ‘€ = 𝑛 More in general α‰Š max 𝑔 𝑦, 𝑧 s.t. 𝑕 𝑦, 𝑧 = 𝑑 Constrained optimization problem β–ͺ 𝑔 𝑦, 𝑧 is the objective function β–ͺ 𝑕 𝑦, 𝑧 = 𝑑 is the constraint β–ͺ 𝑦 and 𝑧 are the decision variables 7

  8. EXERCISE 1 Given is the profit function 𝜌 𝑏, 𝑐 = 5𝑏 2 + 12𝑐 2 and the capacity constraint 8𝑏 + 4𝑐 = 20 . Write as α‰Šmin/max subject to . 8

  9. CONSTRAINED OPTIMIZATION Visualisation of 𝑨 = 𝑔 𝑦, 𝑧 β–ͺ in 3D β–ͺ as level curve 10

  10. LAGRANGE METHOD How to solve the constrained optimization problem? α‰Š max 𝑔 𝑦, 𝑧 s.t. 𝑕 𝑦, 𝑧 = 𝑑 Trick: β–ͺ introduce extra variable (Lagrange multiplier) πœ‡ β–ͺ define new function (Lagrangian) β„’ 𝑦, 𝑧, πœ‡ β–ͺ as follows β„’ 𝑦, 𝑧, πœ‡ = 𝑔 𝑦, 𝑧 βˆ’ πœ‡ 𝑕 𝑦, 𝑧 βˆ’ 𝑑 Observe that β„’ is a function of 3 variables, while the objective function 𝑔 is a function of 2 variables 11

  11. LAGRANGE METHOD Motivation: β–ͺ solutions of the constrained optimization are among the stationary points of β„’ β–ͺ in other words: all stationary points of β„’ are candidate solutions of the original constrained maximization problem Lagrange method β–ͺ Joseph-Louis Lagrange (1736-1813) 12

  12. LAGRANGE METHOD So, α‰Š max 𝑔 𝑦, 𝑧 𝑕 𝑦, 𝑧 = 𝑑 is equivalent to max β„’ 𝑦, 𝑧, πœ‡ s.t. β–ͺ if we define β„’ 𝑦, 𝑧, πœ‡ = 𝑔 𝑦, 𝑧 βˆ’ πœ‡ 𝑕 𝑦, 𝑧 βˆ’ 𝑑 We have written the constrained optimization problem with 2 decision variables as an unconstrained optimization problem with 3 decision variables β–ͺ trade-off β–ͺ unconstrained easier than constrained β–ͺ 3 decision variables harder than 2 13

  13. EXERCISE 2 Given is the profit function 𝜌 𝑏, 𝑐 = 5𝑏 2 + 12𝑐 2 and the capacity constraint 8𝑏 + 4𝑐 = 20 . Write the Lagrangian. 14

  14. LAGRANGE METHOD Example 1 Constrained problem: 𝑔 𝑦, 𝑧 = 𝑦 2 + 𝑧 2 β–ͺ ࡝ max 𝑕 𝑦, 𝑧 = 𝑦 2 + 𝑦𝑧 + 𝑧 2 = 3 s.t. Lagrangian: β–ͺ β„’ 𝑦, 𝑧, πœ‡ = 𝑦 2 + 𝑧 2 βˆ’ πœ‡ 𝑦 2 + 𝑦𝑧 + 𝑧 2 βˆ’ 3 First-order conditions for stationary points: πœ–β„’ β–ͺ πœ–π‘¦ = 0 πœ–β„’ β–ͺ πœ–π‘§ = 0 πœ–β„’ β–ͺ πœ–πœ‡ = 0 16

  15. LAGRANGE METHOD Example 1 (continue) πœ–β„’ β–ͺ πœ–π‘¦ = 2𝑦 βˆ’ πœ‡ 2𝑦 + 𝑧 = 0 πœ–β„’ β–ͺ πœ–π‘§ = 2𝑧 βˆ’ πœ‡ 𝑦 + 2𝑧 = 0 πœ–β„’ πœ–πœ‡ = βˆ’π‘¦ 2 βˆ’ 𝑦𝑧 βˆ’ 𝑧 2 + 3 = 0 β–ͺ So: 2𝑦 𝑦+2𝑧 β‡’ 2𝑦 2 + 4𝑦𝑧 = 4𝑦𝑧 + 2𝑧 2 β‡’ 𝑦 2 = 𝑧 2 2𝑧 β–ͺ 2𝑦+𝑧 = β–ͺ βˆ’π‘¦ 2 Β± 𝑦 2 βˆ’ 𝑦 2 + 3 = 0 β‡’ 3𝑦 2 = 3 ∨ 𝑦 2 = 3 β–ͺ 𝑦 = βˆ’1 ∨ 𝑦 = 1 ∨ 𝑦 = 3 ∨ 𝑦 = βˆ’ 3 β–ͺ 𝑧 follows, and so does πœ‡ 17

  16. LAGRANGE METHOD Example 1 (continued) Four stationary points 𝑦, 𝑧, πœ‡ of Lagrangian: 3 , βˆ’1, βˆ’1, 2 3 , 3, βˆ’ 3, 2 and βˆ’ 3, β–ͺ 1,1, 2 3, 2 β–ͺ check! Four other points 𝑦, 𝑧, πœ‡ do not satisfy all equations: 1, βˆ’1,2 , βˆ’1,1,2 , 3 and βˆ’ 3, βˆ’ 3, 2 β–ͺ 3, 2 3, 3 β–ͺ check! So, 𝑦, 𝑧 = 1,1 , βˆ’1, βˆ’1 , 3, βˆ’ 3 and 3 are candidate solutions to the original βˆ’ 3, constrained problem 18

  17. LAGRANGE METHOD Example 1 (continued) Determine nature of these points (min/max/saddle) and then determine the maximum β–ͺ 𝑔 1,1 = 1 2 + 1 2 = 2 β–ͺ 𝑔 βˆ’1, βˆ’1 = βˆ’1 2 + βˆ’1 2 = 2 2 + βˆ’ 3 2 = 6 β–ͺ 𝑔 3, βˆ’ 3 = 3 2 + 2 = 6 β–ͺ 𝑔 βˆ’ 3, 3 = βˆ’ 3 3 So: 3, βˆ’ 3 and βˆ’ 3, 3 are maximum points β–ͺ β–ͺ and (1,1) and βˆ’1, βˆ’1 are mimimum points 19

  18. LAGRANGE METHOD Example 2 Constrained problem: β–ͺ ࡝ maximize π‘Ÿ 𝐿, 𝑀 = 𝐡𝐿 0.4 𝑀 0.7 (𝐡 > 0) subject to 25𝐿 + 10𝑀 = 𝑛 (𝑛 > 0) Lagrangian β–ͺ β„’ 𝐿, 𝑀, πœ‡ = 𝐡𝐿 0.4 𝑀 0.7 βˆ’ πœ‡ 25𝐿 + 10𝑀 βˆ’ 𝑛 First-order conditions for stationary points πœ–β„’ πœ–πΏ = 0.4𝐡𝐿 βˆ’0.6 𝑀 0.7 βˆ’ 25πœ‡ = 0 β–ͺ πœ–β„’ πœ–π‘€ = 0.7𝐡𝐿 0.4 𝑀 βˆ’0.3 βˆ’ 10πœ‡ = 0 β–ͺ πœ–β„’ β–ͺ πœ–πœ‡ = βˆ’ 25𝐿 + 10𝑀 βˆ’ 𝑛 = 0 20

  19. LAGRANGE METHOD Example 2 (continued) 0.4𝐡𝐿 βˆ’0.6 𝑀 0.7 0.7𝐡𝐿 0.4 𝑀 βˆ’0.3 β–ͺ πœ‡ = = 25 10 β–ͺ multiply both sides with 𝐿 0.6 (to get rid of 𝐿 βˆ’0.6 ) and with 𝑀 0.3 (to get rid of 𝑀 βˆ’0.3 ) 0.4𝐡𝑀 0.7𝐡𝐿 10Γ—0.4 β–ͺ = β‡’ 25𝐿 = 𝑀 25 10 0.7 10Γ—0.4 0.7 𝑛 0.4 𝑛 10 and 𝐿 = β–ͺ 𝑀 + 10𝑀 = 𝑛 β‡’ 𝑀 = 0.7 1.1 1.1 25 β–ͺ πœ‡ = β‹― (not intetesting) 0.4 𝑛 0.7 𝑛 β–ͺ Candidate maximum point: 𝐿, 𝑀 = 25 , 1.1 1.1 25 21

  20. LAGRANGE METHOD Example 2 (continued) Determine nature of the stationary point of Lagrangian: 𝑛 β–ͺ take 𝐿 = 0 β‡’ 𝑀 = 10 and see that π‘Ÿ = 0 𝑛 β–ͺ likewise take 𝑀 = 0 β‡’ 𝐿 = 25 and see that π‘Ÿ = 0 0.4 𝑛 𝑛 0.7 𝑛 𝑛 β–ͺ so for 0 < 25 and 0 < 10 , also π‘Ÿ > 0 25 < 10 < 1.1 1.1 0.4 𝑛 0.7 𝑛 β–ͺ therefore 𝐿, 𝑀 = 10 is a maximum 25 , 1.1 1.1 Here, we look at two ” neighbouring ” points of the stationary point: one β€œto the left” and one β€œto the right”. We find that both have a lower function value, so the stationary point is a maximum. 22

  21. LAGRANGE METHOD Different point on the same budget Optimal Cobb- line with π‘Ÿ < π‘Ÿ βˆ— Douglas line π‘Ÿ βˆ— = 𝐡𝐿 0.4 𝑀 0.7 𝑀 Optimum point on Budget line line for optimum 25𝐿 + 10𝑀 = 𝑛 production π‘Ÿ βˆ— 𝑀 βˆ— = 0.7 𝑛 Different point on 1.1 10 the same budget line with π‘Ÿ < π‘Ÿ βˆ— 𝐿 βˆ— = 0.4 𝑛 𝐿 1.1 25 23

  22. LAGRANGE METHOD Full procedure: β–ͺ rewrite constrained optimization problem in 2D as an unconstrained optimization problem in 3D (from α‰Š max 𝑔 𝑦, 𝑧 s. t. 𝑕 𝑦, 𝑧 = 𝑑 to max β„’ 𝑦, 𝑧, πœ‡ ) Do not use 2 πœ– 2 β„’ πœ– 2 β„’ πœ– 2 β„’ πœ–π‘§ 2 βˆ’ πœ–β„’ πœ–π‘¦ 2 πœ–π‘¦ = 0 πœ–π‘¦πœ–π‘§ for this! πœ–β„’ β–ͺ find 𝑦 and 𝑧 such that πœ–π‘§ = 0 πœ–β„’ πœ–πœ‡ = 0 β–ͺ check if the stationary points are indeed a maximum 24

  23. OLD EXAM QUESTION 22 October 2014, Q2a 25

  24. OLD EXAM QUESTION 27 March 2015, Q2b’ 26

  25. 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