Lecture 1A: Sums and product (and Chapters 2.8, 2.9 and 2.11) Summation operator Σ (the uppercase of sigma, 𝜎): 𝑛 ∑ 𝑥𝑖 = 𝑥𝑚 + 𝑥𝑚+1 + ⋯ + 𝑥𝑛−1 + 𝑥𝑛 𝑖=𝑚 where Σ is the summation sign; 𝑥𝑖 is its typical elements; 𝑖 the index of summation, i.e. the variable which is being summed; 𝑚 the starting point (or the lower limit of summation); 𝑛 the stopping point (or upper limit of summation). This reads as “the sum from 𝑖 = 𝑚 to 𝑖 = 𝑛, of 𝑥𝑖 ”. Properties of sums: ∑(𝑥𝑖 + 𝑦𝑖 ) = ∑ 𝑥𝑖 + ∑ 𝑦𝑖 Additivity: Homogeneity: ∑ 𝑐𝑥𝑖 = 𝑐 ∑ 𝑥𝑖 𝑛+𝑚 Extending terms: ∑𝑛𝑖=1 𝑥𝑖 + ∑𝑛+𝑚 𝑖=𝑛+1 𝑥𝑖 = ∑𝑖=1 𝑥𝑖 𝑛 ∑ Constant term: 𝑖=1 𝑐 = 𝑛𝑐 ∑𝑛𝑖=1(𝑥𝑖 − 𝑥𝑖−1 ) = 𝑥𝑛 − 𝑥0 Telescopic series: Double summation ∑ ∑ : 𝑛 𝑚 ∑ ∑ 𝑥𝑖𝑗 = ⏟ 𝑥11 + ⋯ + 𝑥1𝑚 + ⏟ 𝑥21 + ⋯ + 𝑥2𝑚 + ⋯ + ⏟ 𝑥𝑛1 + ⋯ + 𝑥𝑛𝑚 𝑖=1 𝑗=1 𝑖=1 𝑖=2 Product operator ∏ : 𝑛 ∏ 𝑥𝑖 = 𝑥𝑚 × 𝑥𝑚+1 × ⋯ × 𝑥𝑛−1 × 𝑥𝑛 𝑖=𝑚 Properties: 𝑛+𝑚 Extending terms: ∏𝑛𝑖=1 𝑥𝑖 × ∑𝑛+𝑚 𝑖=𝑛+1 𝑥𝑖 = ∑𝑖=1 𝑥𝑖 𝑛 𝑛 𝑛 ∑ ∑ Constant terms: 𝑖=1 𝑐𝑥𝑖 = 𝑐 𝑖=1 𝑥𝑖 𝑖=𝑛 Lecture 5A: Indefinite integrals The antiderivative (also called the indefinite integral) is the inverse operation of the derivative. Example: 𝐹(𝑥) = 4𝑥 2 − 2𝑥 + 𝐶 𝑑𝑦 𝑑𝑥 (4𝑥 2 − 2𝑥 + 𝐶) ∫(8𝑥 − 2) 𝑑𝑥 + 𝐶 𝑑𝑦 𝑑𝑥 = 8𝑥 − 2 Definition: We write 𝐹(𝑥) for the antiderivative of 𝑓(𝑥) In formula ∫ 𝑓(𝑥)𝑑𝑥 = 𝐹(𝑥) + 𝑐 where 𝑓(𝑥) is the integrand 𝑥 is the variable of integration 𝐶 is the constant of integration Basic indefinite integrals: 𝑓(𝑥) = 𝑎: ∫ 𝑎𝑑𝑥 = 𝑎𝑥 + 𝐶 o example: 𝑓(𝑥) = 10 ⇒ ∫ 10𝑑𝑥 = 10𝑥 + 𝐶 𝑓(𝑥) = 𝑥 𝑎 : ∫ 𝑥 𝑎 𝑑𝑥 = 1 𝑎+1 𝑥 𝑎+1 + 𝐶 (with 𝑎 ≠ −1) 1 o example: 𝑓(𝑥) = 𝑥 4 ⇒ ∫ 𝑥 4 𝑑𝑥 = 𝑥 5 + 𝐶 5 1 1 𝑥 𝑥 𝑓(𝑥) = : ∫ 𝑑𝑥 = ln(|𝑥|) + 𝐶 o example: 𝑓(𝑥) = 1 10 ⇒∫ 1 10 𝑑𝑥 = ln(|10|) + 𝐶 1 𝑓(𝑥) = 𝑒 𝑎𝑥 : ∫ 𝑒 𝑎𝑥 𝑑𝑥 = 𝑒 𝑎𝑥 + 𝐶 (with 𝑎 ≠ 0) 𝑎 o example: 𝑓(𝑥) = 𝑒 𝑥 ⇒ ∫ 𝑒 𝑥 𝑑𝑥 = 𝑒 𝑥 + 𝐶) 𝑓(𝑥) = 𝑎 𝑥 : ∫ 𝑎 𝑥 𝑑𝑥 = 1 ln(𝑎) 𝑎 𝑥 + 𝐶 (with 𝑎 > 0 𝑎𝑛𝑑 𝑎 ≠ 1) o example: 𝑓(𝑥) = 10𝑥 ⇒ ∫ 10𝑥 𝑑𝑥 = 1 ln 10 10𝑥 + 𝐶) Properties of indefinite integrals 𝑎 ⋅ 𝑓(𝑥): ∫(𝑎𝑓 (𝑥))𝑑𝑥 = 𝑎 ∫ 𝑓(𝑥)𝑑𝑥 𝑓(𝑥) + 𝑎: ∫(𝑓(𝑥) + 𝑎)𝑑𝑥 = ∫ 𝑓(𝑥)𝑑𝑥 + 𝑎𝑥 𝑓(𝑥) + 𝑔(𝑥): ∫(𝑓(𝑥) + 𝑔(𝑥))𝑑𝑥 = ∫ 𝑓(𝑥)𝑑𝑥 + ∫ 𝑔(𝑥)𝑑𝑥 Playing with constants: Suppose that ∫ 𝑓(𝑥)𝑑𝑥 = 𝐹(𝑥) + 𝐶 and ∫ 𝑔(𝑥)𝑑𝑥 = 𝐺(𝑥) + 𝐷 then: ∫(𝑓(𝑥) + 𝑔(𝑥))𝑑𝑥 = ∫ 𝑓(𝑥)𝑑𝑥 + ∫ 𝑔(𝑥)𝑑𝑥 = = 𝐹(𝑥) + 𝐶 + 𝐺(𝑥) + 𝐷 = = 𝐹(𝑥) + 𝐺(𝑥) + 𝐸 Lecture 9B: Numbers and units An integer is any number in the set {… , −3, −2, −1, 0, 1, 2, 3, … } A positive integer is any number in the set {1, 2, 3, … } A negative integer is any number in the set {−1, −2, −3, … } A nonnegative integers is {0} A natural number is any integer greater or equal to than 0. Natural numbers begin at 1 and increment to infinity: {1, 2, 3, 4, 5, … } One of the numbers 0, 1, 2, 3,… Fractions: 2/3, 3 ¼, … Decimal numbers: 1.82, 0.034, … Mind the confusion between decimal point and decimal comma Mind the optional use of a thousand separator (as in 1.234.567,8) So what is 1.234.567? It is ambiguous Leading and trailing zeros: Are often skipped: 00123.45600=123.456 Not always in monetary amounts: $12.50, not $12.5 What is 3. and what is .3? Negative numbers: -3, -4 ½, -0.012, … Mind that -4 ½ < 4 Mind that the minus sign is long (not -4 but −4) Non-computed numbers Scientific notation: Decimal notation 0.001 0.01 0.1 1 10 100 1000 -10 Scientific notation 1 × 10−3 1 × 10−2 1 × 10−1 1 × 100 1 × 101 1 × 102 1 × 103 −1 × 101 Software 1E-3 1E-2 1E-1 1E0 1E1 1E2 1E3 -1E1 Words: Value in scientifi c notation 10^0 10^1 10^2 10^3 10^6 10^9 Value in propositional notation 1 10 100 1,000 1,000,000 Short scale (US, modern UK) Name Logic Long scale (continental, older UK) one ten hundred thousand million one ten hundred thousand million 1,000,000,000 billion 10^12 1,000,000,000,000 trillion 10^15 1,000,000,000,000,000 10^18 1,000,000,000,000,000,000 10^21 1,000,000,000,000,000,000,000 sextillion 10^24 1,000,000,000,000,000,000,000,0 00 septillion Unit: quadrillio n quintillion 1,000×1000^ 1 1,000×1000^ 2 1,000×1000^ 3 1,000×1000^ 4 1,000×1000^ 5 1,000×1000^ 6 1,000×1000^ 7 Name thousand million billion Alternativ e name 1,000,000^1 milliard billiard thousand billion trillion thousand trillion quadrillio n Logic 1,000×1,000,000^ 1 1,000,000^2 1,000×1,000,000^ 2 1,000,000^3 trilliard 1,000×1,000,000^ 3 1,000,000^4 Lecture 10A: Constrained optimization Problem formulation max 𝑓(𝑥, 𝑦) { s. t. 𝑔(𝑥, 𝑦) = 𝑐 Constrained optimization problem 𝑓(𝑥, 𝑦) is the objective function 𝑔(𝑥, 𝑦) = 𝑐 is the constraint 𝑥 and 𝑦 are the decision variables Lagrange method Introduce extra variable (Lagrange multiplier) 𝜆 Define new function Lagrangain) ℒ(𝑥, 𝑦, 𝜆) as follows ℒ(𝑥, 𝑦, 𝜆) = 𝑓(𝑥, 𝑦) − 𝜆(𝑔(𝑥, 𝑦) − 𝑐) Find stationary points of the Lagrangian Motivation: Solutions of the constrained optimization are among the stationary points of ℒ. In other words: all stationary points of ℒ are candidate solutions of the original constrained maximization problem max So, { s. t. 𝑓(𝑥, 𝑦) is in some sense equivalent to max ℒ(𝑥, 𝑦, 𝜆) 𝑔(𝑥, 𝑦) = 𝑐 Where we have defined ℒ(𝑥, 𝑦, 𝜆) = 𝑓(𝑥, 𝑦) − 𝜆(𝑔(𝑥, 𝑦) − 𝑐) We have written the constraint optimization with 2 decision variables as an unconstrained optimization problem with 3 decision variables 1. Rewrite constrained optimization problem in 2D as an unconstrained optimization problem in 3D (from max 𝑓(𝑥, 𝑦) { to max ℒ(𝑥, 𝑦, 𝜆)) s. t. 𝑔(𝑥, 𝑦) = 𝑐 𝜕ℒ =0 2. Find 𝑥 and 𝑦 such that 𝜕𝑥 𝜕ℒ 𝜕𝑦 𝜕ℒ =0 {𝜕𝜆 = 0 Constrained optimization: Suppose: The price of 1 unit of 𝐾 is 𝑘 The price of 1 unit of 𝐿 is 𝑙 The available budget is 𝑚 Budget line: 𝑘𝐾 + 𝑙𝐿 = 𝑚 Iso-production line for 𝑞1 Iso-production line for 𝑞2 𝐿 Optimum point on line for optimum production 𝑞 ∗ Budget line 𝐿∗ 𝑞∗ 𝑞2 𝑞1 𝐾∗ 𝐾 Problem formulation +local min local max + local min + Example Given is the profit function 𝜋(𝑎, 𝑏) = 5𝑎2 + 12𝑏 2 and the capacity constraint 8𝑎 + 4𝑏 = 20 min/max Write as { subject to 2 2 Sol {max 5𝑎 + 12𝑏 s. t. 8𝑎 + 4𝑏 = 20 +local max max 𝑓(𝑥, 𝑦) { s. t. 𝑔(𝑥, 𝑦) = 𝑐 Constrained optimization problem 𝑓(𝑥, 𝑦) is the objective function 𝑔(𝑥, 𝑦) = 𝑐 is the constraint 𝑥 and 𝑦 are the decision variables Visualization of 𝑧 = 𝑓(𝑥, 𝑦) In 3D As level curve Lagrange method Introduce extra variable (Lagrange multiplier) 𝜆 Define new function Lagrangain) ℒ(𝑥, 𝑦, 𝜆) as follows ℒ(𝑥, 𝑦, 𝜆) = 𝑓(𝑥, 𝑦) − 𝜆(𝑔(𝑥, 𝑦) − 𝑐) Find stationary points of the Lagrangian Example Given is the profit function 𝜋(𝑎, 𝑏) = 5𝑎2 + 12𝑏 2 and the capacity constraint 8𝑎 + 4𝑏 = 20 Write the Lagrangian Sol ℒ(𝑎, 𝑏, 𝜆) = 5𝑎2 + 12𝑏 2 − 𝜆(8𝑎 + 4𝑏 − 20) Motivation: Solutions of the constrained optimization are among the stationary points of ℒ. In other words: all stationary points of ℒ are candidate solutions of the original constrained maximization problem max 𝑓(𝑥, 𝑦) So, { is in some sense equivalent to s. t. 𝑔(𝑥, 𝑦) = 𝑐 max ℒ(𝑥, 𝑦, 𝜆) Where we have defined ℒ(𝑥, 𝑦, 𝜆) = 𝑓(𝑥, 𝑦) − 𝜆(𝑔(𝑥, 𝑦) − 𝑐) We have written the constraint optimization with 2 decision variables as an unconstrained optimization problem with 3 decision variables Full procedure 1. Rewrite constrained optimization problem in 2D as an unconstrained optimization problem in 3D (from max 𝑓(𝑥, 𝑦) { to max ℒ(𝑥, 𝑦, 𝜆)) s. t. 𝑔(𝑥, 𝑦) = 𝑐 𝜕ℒ =0 2. Find 𝑥 and 𝑦 such that 𝜕𝑥 𝜕ℒ 𝜕𝑦 𝜕ℒ =0 {𝜕𝜆 = 0 3. Check if the stationary points are indeed a maximum Note that you should not use ( ( 𝜕2 ℒ 𝜕𝑥𝜕𝑦 2 ) for 𝜕2 ℒ 𝜕𝑥 2 )( 𝜕2 ℒ 𝜕𝑦 2 )− Example 1 Constrained problem: max 𝑓(𝑥, 𝑦) = 𝑥 2 + 𝑦 2 { s. t. 𝑔(𝑥, 𝑦) = 𝑥 2 + 𝑥𝑦 + 𝑦 2 = 3 Lagrangian: ℒ(𝑥, 𝑦, 𝜆) = 𝑥 2 + 𝑦 2 − 𝜆(𝑥 2 + 𝑥𝑦 + 𝑦 2 − 3) First-order conditions for stationary points: 𝜕ℒ = 2𝑥 − 𝜆(2𝑥 + 𝑦) = 0 𝜕𝑥 𝜕ℒ = 2𝑦 − 𝜆(𝑥 + 2𝑦) = 0 𝜕𝑦 𝜕ℒ 2 2 { 𝜕𝜆 = −𝑥 − 𝑥𝑦 − 𝑦 + 3 = 0 (1) (2) (3) the constrained case Phase 1: Solve (1) and (2) for 𝜆 and equate 2𝑥 2𝑥 + 𝑦 2𝑦 2𝑦 − 𝜆(𝑥 + 2𝑦) ⇒ 2𝑦 = 𝜆(𝑥 + 2𝑦) ⇒ 𝜆 = 2𝑦 + 𝑥 2𝑥 2𝑦 ⋅ = 2𝑥(2𝑦 + 𝑥) = 2𝑦(2𝑥 + 4) 2𝑥 + 𝑦 2𝑦 + 𝑥 𝑥 2 = 𝑦2 𝑥 = ±𝑦 2𝑥 − 𝜆(2𝑥 + 𝑦) ⇒ 2𝑥 = 𝜆(2𝑥 + 𝑦) ⇒ 𝜆 = Phase 2 (fill in constraints) −𝑥 2 − 𝑥𝑦 − 𝑦 2 + 3 = 0 Let 𝑥 = 𝑦 in (3) −𝑥 2 − 𝑥 2 − 𝑥 2 + 3 = 0 −3𝑥 2 = −3 𝑥2 = 1 𝑥 = ±1 (1,1), (−1,1) Let 𝑥 = −1 ⇔ −𝑥 = 𝑦 −𝑥 2 − 𝑥(−𝑥)— 𝑥2 + 3 = 0 −𝑥 2 = −3 𝑥2 = 3 𝑥 = ±√3 (√3, −√3), (−√3, √3) Phase 3 Candidates: (1,1), (−1,1) (√3, −√3), (−√3, √3) Determine the nature of these points (min/max/saddle) and then determine the maximum 𝑓(1,1) = 12 + 12 = 2 𝑓(−1, −1) = (−1)2 + (−1)2 = 2 2 2 𝑓(√3, −√3) = (√3) + (−√3) = 6 2 2 √−√3, √3 = −(√3) + (√3) = 6 So: (√3, −√3) and (−√3, √3) are maximum points And (1,1) and (−1, −1) are minimum points Example 2: Constrained problem: max 𝑞(𝐾, 𝐿) = 𝐴𝐾 0.4 𝐿0.7 (𝐴 > 0) { s. t. 25𝐾 + 10𝐿 = 𝑚 (𝑚 > 0) Lagrangian: ℒ(𝐾, 𝐿, 𝜆) = 𝐴𝐾 0.4 𝐿0.7 − 𝜆(25𝐾 + 10𝐿 − 𝑚) First-order conditions for stationary points: 𝜕ℒ = 0.4𝐴𝐾 −0.6 𝐿0.7 − 25𝜆 𝜕𝐾 𝜕ℒ = 0.7𝐴𝐾 0.4 𝐿0.3 − 10𝜆 𝜕𝐿 𝜕ℒ { 𝜕𝜆 = −(25𝐾 + 10𝐿 − 𝑚) 0.4𝐴𝐾 −0.6 𝐿0.7 (1) (2) (3) 0.7𝐴𝐾 0.4𝐿−0.3 (𝜆) = = 25 10 Multiply both sides with 𝑘 0.6 (to get rid of 𝐾 −0.6 ) and with 𝐿0.3 (to get rid of 𝐿−0.3 ) 0.4𝐴𝐿 0.7𝐴𝐾 10 × 0.4 = ⇒ 25𝐾 = 𝐿 25 10 0.7 10×0.4 0.7 𝑚 0.4 𝑚 𝐿 + 10𝐿 = 𝑚 ⇒ and 𝐾 = 0,7 1.1 10 1.1 25 𝜆 = ⋯ (not interesting) Candidate maximum point: (𝐾, 𝐿) = ( 0.4 𝑚 0.7 𝑚 , 1.1 25 1.1 10 ) Determine nature of the stationary point: 𝑚 Take 𝐾 = 0 ⇒ 𝐿 = and see that 𝑞 = 0 10 Likewise take 𝐿 = 0 ⇒ 𝑘 = So for 0 < 0.4 𝑚 1.1 25 < 𝑚 25 𝑚 25 and 0 < Therefore (𝐾, 𝐿) = (( and see that 𝑞 = 0 0,7 𝑚 1.1 10 0.4 𝑚 0.7 𝑚 , 1.1 25 1.1 10 < 𝑚 10 , also 𝑞 > 0 ) is a maximum Different point on the same budget line with 𝑞 < 𝑞∗ 𝐿 𝐿 Optimal CobbDouglas line 𝑞∗ = 𝐴𝐾 0.4𝐿0.7 Optimum point on line for optimum production 𝑞∗ 0 .7 𝑚 Budget line 25 𝐾 + 10 𝐿 = 𝑚 Different point on the same budget line with 𝑞 < 𝑞∗ 0 .4 𝑚 𝐾∗ = 1 . 𝐾