Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Continuous Random Variables Chapter 5 Important Concepts Continuous Random Variables Introductory Example Random Angles • Distribution Functions. For the Uniform Spinner, • Density Functions. Ω = (−π, π], • The Median and Other Quantiles. P (a, b] = • The Mean and Variance b−a 2π for −π ≤ a < b ≤ π. Let • Special Distributions X(ω) = ω Uniform for −π < ω ≤ π. Then X is random variable. Exponential Note: P [X = x] = 0 for all x ∈ IR. Normal Two Important Relations Distributions Functions If X ∼ F , then Def. Given a model (Ω, P ) and a RV P [a < X ≤ b] = F (b) − F (a) X : Ω → IR, for all −∞ < a < b < ∞, since the distribution function of X is defined by F (x) = P [X ≤ x] (∗) for −∞ < x < ∞. = F (a) + P [a < X ≤ b] Also, Example. For the random angle F (x) = F (b) = P [X ≤ b] = P [X ≤ a] + P [a < X ≤ b] 1 x + 2 2π for −π < x ≤ π, since P [X ≤ x] = P [−π < X ≤ x] x+π = P ((π, x]) = . 2π Notation: Write X ∼ F if X is a RV with DF F ; that is, if (*) holds. P [X > b] = 1 − P [X ≤ b] = 1 − F (b). Example. In the random angle example, F (x) = 1 x + , 2 2π 2 1 1 1 1 1 F ( π) − F ( π) = ( + ) − ( + ) 3 3 2 3 2 6 1 = . 6 Some Definitions Def. A function F : IR → IR is non-decreasing if F (x) ≤ F (y) whenver x ≤ y. Characteristic Properties One Sided Limits F (x+) = lim F (y), lim F (y). y→x,y>x and F (x−) = y→x,y<x Theorem. A function F : IR → IR is the distribution function of some RV iff F (x) ≤ F (y) whenever x ≤ y, (1) F (x) = lim F (y) each all x, (2) y↓x lim F (x) = 0, (3a) lim F (x) = 1. (3b) x→−∞ x→∞ Proof. Later. Note: Henceforth DF means a function satisfying (1), (2), and (3). The Discrete Case Densities If X is discrete with range X and PMF f , then F (x) = P [X ≤ x] X = f (y). y∈X ,y≤x If also Def. A function f : IR → IR is a density if f (x) ≥ 0, for all x, Z ∞ f (x)dx = 1. −∞ X ⊆ {· · · − 1, 0, 1, 2, · · · }, then F (n) = X f (k) k≤n If X ∼ F and f is a density then X and F have density f if Z x f (y)dy (4) F (x) = −∞ for all −∞ < x < ∞. and f (n) = F (n) − F (n − 1). Theorem. If f is any density, then (4) defines a DF. Proof. Exercise. Corollary. If f is any density, then there is a RV X with density f . Example Uniform Distributions Consequences Of F (x) = Z x f (y)dy (4) −∞ If X ∼ F with density f , then P [a < X ≤ b] = F (b) − F (a) = Z b f (x)dx a for −∞ ≤ a < b ≤ ∞. If −∞ < α < β < ∞, then 1/(β − α) if α < x ≤ β f (x) = 0 if otherwise is a density, since f (x) ≥ 0 for all x and Z ∞ Z β dx f (x)dx = β −α α ∞ 1 (β − α) = 1. = β−α Then If (4) holds, then if x ≤ α 0 F (x) = (x − α)/(β − α) if α < x ≤ β 1 if x > β d F (x) f (x) = F 0 (x) = dx at continuity points of f . Example: Random Angles: α = −π and β = π. Standard α = 0 and β = 1. The Standard Normal Density Fact Example Exponential Distributions If 0 < λ < ∞, then 0 F (x) = 1 − e−λx if x ≤ 0 if else has density f (x) = Z ∞ 1 2 e− 2 z dz = √ 2π. −∞ Let 1 2 1 ϕ(z) = √ e− 2 z 2π Then ϕ is a density, called the standard normal density. The standard normal distribution function is Z z Φ(z) = ϕ(y)dy. −∞ 0 λe −λx if − ∞ < x < 0 if 0 ≤ x < ∞. Note: Tabled Note: Φ(−z) = Notes a). Derivative doesn’t exist when x = 0. = b). Doesn’t matter. −z Z ϕ(y)dy −∞ Z ∞ z ϕ(x)dx = 1 − Φ(z) and Φ(0) = 1 . 2 3 2 1 x 0 Standard Normal Density The Normal Distribution With Parameters Let −2 −1 1 f (x) = ϕ σ −3 Z x f (y)dy = Φ −∞ 0.0 0.1 0.2 0.3 0.4 Density F (m) = th If F (t) = F (x) = p then 0 1 − e−λt F (m) = P [X ≤ x] = p. x−µ σ if ∞ < t < 0 if 0 ≤ t < ∞ 1 2 iff 1 − e−λm = th is called a p -quantile or 100p percentile of X or F . In terms of X, the condition is Example 1 2 is called a median of X or F . More generally, if 0 < p < 1, then any x for which These are the normal distribution with parameters −∞ < µ < ∞ and σ > 0. The Median and Other Quantiles If X ∼ F , then and m for which x−µ σ and F (x) = Figure 1: The Standard Normal Density iff e−λm = 1 2 1 2 iff eλm = 2 iff 1 log(2), λ where log denotes natural logarthim. m= Example: The half life of a radio active substance. The Mean and Variance Def. If X has density f , then the mean of X and/or f is defined by Z ∞ xf (x)dx, µ= −∞ Variance And Standard Deviation provided that the integral converges absolutely. Example: Uniform Distributions. If 1/(β − α) if α < x ≤ β f (x) = 0 if otherwise The variance of X and/or f is Z ∞ (x − µ)2 f (x)dx. σ2 = −∞ Also, then σ = Z ∞ −∞ β xdx α β−α 1 x2 β = | 2 β − α x=α 1 β 2 − α2 = 2 β−α α+β , = 2 µ= Z 2 x2 f (x) − µ2 The standard deviation is √ σ = σ2 . since β 2 − α2 = (β + α)(β − α). An Example A Uniform Distribution Normal Distribution Function If 1 if 0 ≤ x ≤ 1 f (x) = 0 if otherwise then Z µ= Z 1 Z 1 1 xdx = 0 1 21 1 = , x | 2 x=0 2 and σ = are both µ, since 1 1 x dx = x3 |1x=0 = 3 3 2 0 2 The Median and Mean The mean and median of a normal distribution function x−µ F (x) = Φ σ 1 , 2 and 1 1 1 . x dx − µ = − = 3 4 12 2 0 F (µ) = Φ(0) = 2 Z ∞ xf (x)dx = µ + −∞ Z ∞ (x − µ)f (x)dx −∞ Z ∞ =µ+σ More generally, if zϕ(z)dz = µ. −∞ X ∼ Unif(α, β), then σ2 = (β − α)2 . 12 The Variance Similarly, the variance is σ 2 . The Gamma Function Let Γ(α) = for α > 0. Then Z Γ(1) = Z ∞ The Mean and Variance Of Exponential Distributions xα−1 e−x dx 0 If ∞ e −x dx = 0 −e−x |∞ x=0 = 1, and Γ(α + 1) = αΓ(α) for α > 0. For, integrating by parts, Z ∞ Γ(α) = xα e−x dx 0 Z ∞ = −xα e−x |∞ + αxα−1 e−x dx x=0 0 = αΓ(α). So Γ(n) = (n − 1)!. The Mean and Variance Of Exponential Distributions Continued Similarly, Z ∞ x2 λe−λx dx 0 Z ∞2 1 x2 e−λx dx λ2 0 1 = 2 Γ(3) λ 2 = 2 λ = and σ2 = 2 λ2 − µ2 = 1 . λ2 f (x) = then 0 if x < 0 λe−λx µ= Z if 0 ≤ x < ∞ ∞ xλe−λx dx 0 Z 1 ∞ −λx xe dx λ 0 1 = Γ(2) λ 1 = λ =