【数学】概率论与数理统计(四)
文章目录
- @[toc]
- 分布函数
- 分布函数
- 性质
- 离散型随机变量的分布函数
- 连续型随机变量的分布函数
- 示例1
- 问题
- 解答
- 正态随机变量
- 示例
- 问题
- 解答
- 示例2
- 问题
- (1)
- (2)
- 解答
- (1)
- (2)
- 随机变量函数的分布
- 离散型随机变量函数的分布
- 连续型随机变量函数的分布
- 示例1
- 问题
- 解答
- 定理1
- 证明
- 定理2
- 示例2
- 问题
- 解答
- 解法一
- 解法二
- 示例3
- 问题
- 解答
文章目录
- @[toc]
- 分布函数
- 分布函数
- 性质
- 离散型随机变量的分布函数
- 连续型随机变量的分布函数
- 示例1
- 问题
- 解答
- 正态随机变量
- 示例
- 问题
- 解答
- 示例2
- 问题
- (1)
- (2)
- 解答
- (1)
- (2)
- 随机变量函数的分布
- 离散型随机变量函数的分布
- 连续型随机变量函数的分布
- 示例1
- 问题
- 解答
- 定理1
- 证明
- 定理2
- 示例2
- 问题
- 解答
- 解法一
- 解法二
- 示例3
- 问题
- 解答
分布函数
分布函数
- 设 X X X是一个随机变量, x x x是任意实数,称 F ( x ) = P { X ≤ x } F(x) = P\set{X \leq x} F(x)=P{X≤x}为 X X X的分布函数
- P { x 1 < X ≤ x 2 } = P { X ≤ x 2 } − P { X ≤ x 1 } = F ( x 2 ) − F ( x 1 ) P\set{x_{1} < X \leq x_{2}} = P\set{X \leq x_{2}} - P\set{X \leq x_{1}} = F(x_{2}) - F(x_{1}) P{x1<X≤x2}=P{X≤x2}−P{X≤x1}=F(x2)−F(x1)
性质
-
0 ≤ F ( x ) ≤ 1 , x ∈ R 0 \leq F(x) \leq 1 , x \in R 0≤F(x)≤1,x∈R
-
F ( − ∞ ) = lim x → − ∞ F ( x ) = 0 F(- \infty) = \lim\limits_{x \rightarrow - \infty}{F(x)} = 0 F(−∞)=x→−∞limF(x)=0, F ( + ∞ ) = lim x → + ∞ F ( x ) = 1 F(+ \infty) = \lim\limits_{x \rightarrow + \infty}{F(x)} = 1 F(+∞)=x→+∞limF(x)=1
-
F ( x ) F(x) F(x)是单调不减的函数,即 F ( x 1 ) ≤ F ( x 2 ) , x 1 < x 2 F(x_{1}) \leq F(x_{2}) , x_{1} < x_{2} F(x1)≤F(x2),x1<x2,事实上, F ( x 2 ) − F ( x 1 ) = P { x 1 < X ≤ x 2 } ≥ 0 F(x_{2}) - F(x_{1}) = P\set{x_{1} < X \leq x_{2}} \geq 0 F(x2)−F(x1)=P{x1<X≤x2}≥0,故 F ( x 1 ) ≤ F ( x 2 ) F(x_{1}) \leq F(x_{2}) F(x1)≤F(x2)
-
F ( x ) F(x) F(x)右连续,即 F ( x ) = F ( x + 0 ) F(x) = F(x + 0) F(x)=F(x+0)
-
满足上述 4 4 4个性质的函数也一定是某个随机变量的分布函数
离散型随机变量的分布函数
-
设 F ( x ) F(x) F(x)是离散型随机变量 X X X的分布函数,则 F ( x ) = P { X ≤ x } = P { ⋃ x k ≤ x ( X = x k ) } = ∑ x k ≤ x P { X = x k } = ∑ x k ≤ x p k F(x) = P\set{X \leq x} = P\set{\bigcup\limits_{x_{k} \leq x}{(X = x_{k})}} = \sum\limits_{x_{k} \leq x}{P\set{X = x_{k}}} = \sum\limits_{x_{k} \leq x}{p_{k}} F(x)=P{X≤x}=P{xk≤x⋃(X=xk)}=xk≤x∑P{X=xk}=xk≤x∑pk,其中和式是对于所有 x k ≤ x x_{k} \leq x xk≤x的指标 k k k求和
-
p k = P { X = x k } = F ( x k ) − F ( x k − 0 ) p_{k} = P\set{X = x_{k}} = F(x_{k}) - F(x_{k} - 0) pk=P{X=xk}=F(xk)−F(xk−0)
-
若已知 X X X的分布函数 F ( x ) F(x) F(x),则 P { a ≤ X ≤ b } = P { a < X ≤ b } + P { X = a } = F ( b ) − F ( a ) + P { X = a } P\set{a \leq X \leq b} = P\set{a < X \leq b} + P\set{X = a} = F(b) - F(a) + P\set{X = a} P{a≤X≤b}=P{a<X≤b}+P{X=a}=F(b)−F(a)+P{X=a}
连续型随机变量的分布函数
- 若 X X X是连续型随机变量,其概率密度为 f ( x ) f(x) f(x),则 F ( X ) = P { X ≤ x } = ∫ − ∞ x f ( t ) d t F(X) = P\set{X \leq x} = \int_{- \infty}^{x}{f(t) dt} F(X)=P{X≤x}=∫−∞xf(t)dt,即分布函数是概率密度函数的可变上限的定积分
- 在 f ( x ) f(x) f(x)的连续点,有 d F ( x ) d x = f ( x ) \frac{dF(x)}{dx} = f(x) dxdF(x)=f(x),即概率密度函数是分布函数的导数
示例1
问题
- 证明:若 X ∼ N ( μ , σ 2 ) X \sim N(\mu, \sigma^{2}) X∼N(μ,σ2),则 Y = X − μ σ ∼ N ( 0 , 1 ) Y = \frac{X - \mu}{\sigma} \sim N(0, 1) Y=σX−μ∼N(0,1)
解答
-
P { a ≤ Y ≤ b } = P { a ≤ X − μ σ ≤ b } = P { μ + σ a ≤ X ≤ μ + σ b } = ∫ μ + σ a μ + σ b 1 σ 2 π e − ( x − μ ) 2 2 σ 2 d x P\set{a \leq Y \leq b} = P\set{a \leq \frac{X - \mu}{\sigma} \leq b} = P\set{\mu + \sigma a \leq X \leq \mu + \sigma b} = \int_{\mu + \sigma a}^{\mu + \sigma b}{\frac{1}{\sigma \sqrt{2 \pi}} e^{- \frac{(x - \mu)^{2}}{2 \sigma^{2}}} dx} P{a≤Y≤b}=P{a≤σX−μ≤b}=P{μ+σa≤X≤μ+σb}=∫μ+σaμ+σbσ2π1e−2σ2(x−μ)2dx
-
令 t = x − μ σ t = \frac{x - \mu}{\sigma} t=σx−μ,得 P { a ≤ Y ≤ b } = ∫ a b 1 2 π e − t 2 2 d t P\set{a \leq Y \leq b} = \int_{a}^{b}{\frac{1}{\sqrt{2 \pi}} e^{- \frac{t^{2}}{2}} dt} P{a≤Y≤b}=∫ab2π1e−2t2dt,则 Y ∼ N ( 0 , 1 ) Y \sim N(0, 1) Y∼N(0,1)
正态随机变量
-
若 X ∼ N ( μ , σ 2 ) X \sim N(\mu, \sigma^{2}) X∼N(μ,σ2),则 Y = a X + b ∼ N ( a μ + b , ∣ a ∣ 2 σ 2 ) Y = aX + b \sim N(a \mu + b, |a|^{2} \sigma^{2}) Y=aX+b∼N(aμ+b,∣a∣2σ2),正态随机变量的线性函数仍为正态随机变量
-
对于标准正态分布,通常用 φ ( x ) \varphi(x) φ(x)表示概率密度函数,用 Φ ( x ) \Phi(x) Φ(x)表示分布函数
-
因 φ ( x ) \varphi(x) φ(x)是偶函数,即 φ ( − x ) = φ ( x ) \varphi(- x) = \varphi(x) φ(−x)=φ(x),于是 Φ ( − x ) = 1 − Φ ( x ) \Phi(- x) = 1 - \Phi(x) Φ(−x)=1−Φ(x),事实上 Φ ( − x ) = ∫ − ∞ − x φ ( t ) d t = ∫ x + ∞ φ ( u ) d u ( 令 t = − u ) = ∫ − ∞ + ∞ φ ( u ) d u − ∫ − ∞ x φ ( u ) d u = 1 − Φ ( x ) \Phi(- x) = \int_{- \infty}^{- x}{\varphi(t) dt} = \int_{x}^{+ \infty}{\varphi(u) du} (令 t = - u) = \int_{- \infty}^{+ \infty}{\varphi(u) du} - \int_{- \infty}^{x}{\varphi(u) du} = 1 - \Phi(x) Φ(−x)=∫−∞−xφ(t)dt=∫x+∞φ(u)du(令t=−u)=∫−∞+∞φ(u)du−∫−∞xφ(u)du=1−Φ(x)
-
正态分布 N ( μ , σ 2 ) N(\mu, \sigma^{2}) N(μ,σ2)的分布函数 F ( x ) = Φ ( x − μ σ ) F(x) = \Phi(\frac{x - \mu}{\sigma}) F(x)=Φ(σx−μ)
示例
问题
- 设 X ∼ N ( μ , σ 2 ) X \sim N(\mu, \sigma^{2}) X∼N(μ,σ2),求 P { ∣ X − μ ∣ < σ } P\set{|X - \mu| < \sigma} P{∣X−μ∣<σ}, P { ∣ X − μ ∣ < 2 σ } P\set{|X - \mu| < 2 \sigma} P{∣X−μ∣<2σ}, P { ∣ X − μ ∣ < 3 σ } P\set{|X - \mu| < 3 \sigma} P{∣X−μ∣<3σ}
解答
-
P { ∣ X − μ ∣ < σ } = P { μ − σ < X < μ + σ } = Φ ( μ + σ − μ σ ) − Φ ( μ − σ − μ σ ) = Φ ( 1 ) − Φ ( − 1 ) = 2 Φ ( 1 ) − 1 = 0.6827 P\set{|X - \mu| < \sigma} = P\set{\mu - \sigma < X < \mu + \sigma} = \Phi(\frac{\mu + \sigma - \mu}{\sigma}) - \Phi(\frac{\mu - \sigma - \mu}{\sigma}) = \Phi(1) - \Phi(- 1) = 2 \Phi(1) - 1 = 0.6827 P{∣X−μ∣<σ}=P{μ−σ<X<μ+σ}=Φ(σμ+σ−μ)−Φ(σμ−σ−μ)=Φ(1)−Φ(−1)=2Φ(1)−1=0.6827
-
P { ∣ X − μ ∣ < 2 σ } = 0.9545 P\set{|X - \mu| < 2 \sigma} = 0.9545 P{∣X−μ∣<2σ}=0.9545
-
P { ∣ X − μ ∣ < 3 σ } = 0.9973 P\set{|X - \mu| < 3 \sigma} = 0.9973 P{∣X−μ∣<3σ}=0.9973
-
X X X几乎全部落在 ( μ − 3 σ , μ + 3 σ ) (\mu - 3 \sigma, \mu + 3 \sigma) (μ−3σ,μ+3σ)区间内,称为三倍标准差原则( 3 σ 3 \sigma 3σ准则)
示例2
问题
- 设 F 1 ( x ) F_{1}(x) F1(x)和 F 2 ( x ) F_{2}(x) F2(x)都是分布函数, a > 0 a > 0 a>0, b > 0 b > 0 b>0均为常数,且 a + b = 1 a + b = 1 a+b=1
(1)
- 求证 F ( x ) = a F 1 ( x ) + b F 2 ( x ) F(x) = a F_{1}(x) + b F_{2}(x) F(x)=aF1(x)+bF2(x)也是一个分布函数
(2)
- 由此讨论分布函数是否只有离散型和连续型两种类型
解答
(1)
- 因 F 1 ( x ) F_{1}(x) F1(x), F 2 ( x ) F_{2}(x) F2(x)均为分布函数,则由分布函数的性质,得当 x 1 < x 2 x_{1} < x_{2} x1<x2时,有 F 1 ( x 1 ) ≤ F 1 ( x 2 ) F_{1}(x_{1}) \leq F_{1}(x_{2}) F1(x1)≤F1(x2), F 2 ( x 1 ) ≤ F 2 ( x 2 ) F_{2}(x_{1}) \leq F_{2}(x_{2}) F2(x1)≤F2(x2)
- F ( x 1 ) = a F 1 ( x 1 ) + b F 2 ( x 1 ) ≤ a F 1 ( x 2 ) + b F 2 ( x 2 ) = F ( x 2 ) F(x_{1}) = a F_{1}(x_{1}) + b F_{2}(x_{1}) \leq a F_{1}(x_{2}) + b F_{2}(x_{2}) = F(x_{2}) F(x1)=aF1(x1)+bF2(x1)≤aF1(x2)+bF2(x2)=F(x2)
- lim x → − ∞ F ( x ) = lim x → − ∞ [ a F 1 ( x ) + b F 2 ( x ) ] = 0 \lim\limits_{x \rightarrow - \infty}{F(x)} = \lim\limits_{x \rightarrow - \infty}{[a F_{1}(x) + b F_{2}(x)]} = 0 x→−∞limF(x)=x→−∞lim[aF1(x)+bF2(x)]=0
- lim x → + ∞ F ( x ) = lim x → + ∞ [ a F 1 ( x ) + b F 2 ( x ) ] = a + b = 1 \lim\limits_{x \rightarrow + \infty}{F(x)} = \lim\limits_{x \rightarrow + \infty}{[a F_{1}(x) + b F_{2}(x)]} = a + b = 1 x→+∞limF(x)=x→+∞lim[aF1(x)+bF2(x)]=a+b=1
- F ( x + 0 ) = a F 1 ( x + 0 ) + b F 2 ( x + 0 ) = a F 1 ( x ) + b F 2 ( x ) = F ( x ) F(x + 0) = a F_{1}(x + 0) + b F_{2}(x + 0) = a F_{1}(x) + b F_{2}(x) = F(x) F(x+0)=aF1(x+0)+bF2(x+0)=aF1(x)+bF2(x)=F(x)
- 因此 F ( x ) F(x) F(x)也是分布函数
(2)
- 取 a = b = 1 2 a = b = \frac{1}{2} a=b=21,并令 F 1 ( x ) = { 0 , x < 0 1 , x ≥ 0 F_{1}(x) = \begin{cases} 0 , & x < 0 \\ 1 , & x \geq 0 \end{cases} F1(x)={0,1,x<0x≥0, F 2 ( x ) = { 0 , x < 0 x , 0 ≤ x ≤ 1 1 , x > 1 F_{2}(x) = \begin{cases} 0 , & x < 0 \\ x , & 0 \leq x \leq 1 \\ 1 , & x > 1 \end{cases} F2(x)=⎩ ⎨ ⎧0,x,1,x<00≤x≤1x>1,则 F ( x ) = { 0 , x < 0 1 + x 2 , 0 ≤ x ≤ 1 1 , x > 1 F(x) = \begin{cases} 0 , & x < 0 \\ \frac{1 + x}{2} , & 0 \leq x \leq 1 \\ 1 , & x > 1 \end{cases} F(x)=⎩ ⎨ ⎧0,21+x,1,x<00≤x≤1x>1
- 显然,此分布函数 F ( x ) F(x) F(x)既不是阶梯函数也不是连续函数,于是 F ( x ) F(x) F(x)所对应的随机变量既不是离散型也不是连续型,这就是说随机变量并非只有离散型和连续型两大类型
随机变量函数的分布
离散型随机变量函数的分布
- 设随机变量 X X X的分布律 P { X = x k } = p k , k = 1 , 2 , ⋯ P\set{X = x_{k}} = p_{k} , k = 1, 2, \cdots P{X=xk}=pk,k=1,2,⋯,则当 Y = g ( X ) Y = g(X) Y=g(X)的所有取值为 y j ( j = 1 , 2 , ⋯ ) y_{j} (j = 1, 2, \cdots) yj(j=1,2,⋯)时,随机变量 Y Y Y的分布律为 P { Y = y j } = q j , j = 1 , 2 , ⋯ P\set{Y = y_{j}} = q_{j} , j = 1, 2, \cdots P{Y=yj}=qj,j=1,2,⋯,其中 q j q_{j} qj是所有满足 g ( x i ) = y j g(x_{i}) = y_{j} g(xi)=yj的 x i x_{i} xi对应的 X X X的概率 P { X = x i } = p i P\set{X = x_{i}} = p_{i} P{X=xi}=pi的和,即 P { Y = y j } = ∑ g ( x i ) = y j P { X = x i } P\set{Y = y_{j}} = \sum\limits_{g(x_{i}) = y_{j}}{P\set{X = x_{i}}} P{Y=yj}=g(xi)=yj∑P{X=xi}
连续型随机变量函数的分布
示例1
问题
- 设随机变量 X X X的概率密度为 f X ( x ) f_{X}(x) fX(x),求 Y = a X + b Y = aX + b Y=aX+b( a a a, b b b为常数,且 a ≠ 0 a \neq 0 a=0)的概率密度
解答
- 设 X X X和 Y Y Y的分布函数分别为 F X ( x ) F_{X}(x) FX(x)和 F Y ( y ) F_{Y}(y) FY(y), Y Y Y的概率密度为 f Y ( y ) f_{Y}(y) fY(y)
- 由分布函数定义,得 F Y ( y ) = P { Y ≤ y } = P { a X + b ≤ y } F_{Y}(y) = P\set{Y \leq y} = P\set{aX + b \leq y} FY(y)=P{Y≤y}=P{aX+b≤y}
- 当 a > 0 a > 0 a>0时,有 F Y ( y ) = P { X ≤ y − b a } = F X ( y − b a ) = ∫ − ∞ y − b a f X ( x ) d x F_{Y}(y) = P\set{X \leq \frac{y - b}{a}} = F_{X}(\frac{y - b}{a}) = \int_{- \infty}^{\frac{y - b}{a}}{f_{X}(x) dx} FY(y)=P{X≤ay−b}=FX(ay−b)=∫−∞ay−bfX(x)dx
- 当 a < 0 a < 0 a<0时,有 F Y ( y ) = P { X ≥ y − b a } = ∫ y − b a + ∞ f X ( x ) d x = − ∫ − ∞ y f X ( t − b a ) ⋅ 1 a d t , ( 令 x = t − b a ) F_{Y}(y) = P\set{X \geq \frac{y - b}{a}} = \int_{\frac{y - b}{a}}^{+ \infty}{f_{X}(x) dx} = - \int_{- \infty}^{y}{f_{X}(\frac{t - b}{a}) \cdot \frac{1}{a} dt} , (令 x = \frac{t - b}{a}) FY(y)=P{X≥ay−b}=∫ay−b+∞fX(x)dx=−∫−∞yfX(at−b)⋅a1dt,(令x=at−b)
- 对 F Y ( y ) F_{Y}(y) FY(y)求导,得 Y Y Y的概率密度
f Y ( y ) = { 1 a f X ( y − b a ) , a > 0 − 1 a f X ( y − b a ) , a < 0 f_{Y}(y) = \begin{cases} \cfrac{1}{a} f_{X} (\cfrac{y - b}{a}) , & a > 0 \\ - \cfrac{1}{a} f_{X} (\cfrac{y - b}{a}) , & a < 0 \end{cases} fY(y)=⎩ ⎨ ⎧a1fX(ay−b),−a1fX(ay−b),a>0a<0
-
即 f Y ( y ) = 1 ∣ a ∣ f X ( y − b a ) ( a ≠ 0 , y ∈ R ) f_{Y}(y) = \frac{1}{|a|} f_{X}(\frac{y - b}{a}) (a \neq 0 , y \in R) fY(y)=∣a∣1fX(ay−b)(a=0,y∈R)
-
y = f ( x ) y = f(x) y=f(x)的反函数为 x = h ( y ) = y − b a x = h(y) = \frac{y - b}{a} x=h(y)=ay−b,且 ∣ h ′ ( y ) ∣ = 1 ∣ a ∣ |h^{'}(y)| = \frac{1}{|a|} ∣h′(y)∣=∣a∣1,故 f Y ( y ) = f X ( h ( y ) ) ⋅ ∣ h ′ ( y ) ∣ f_{Y}(y) = f_{X}(h(y)) \cdot |h^{'}(y)| fY(y)=fX(h(y))⋅∣h′(y)∣
定理1
- 设 X X X为连续型随机变量, f X ( x ) f_{X}(x) fX(x)为 X X X的概率密度,若 y = g ( x ) y = g(x) y=g(x)为严格单调的连续函数,且反函数 x = h ( y ) x = h(y) x=h(y)有连续导数,则 Y = g ( X ) Y = g(X) Y=g(X)为连续型随机变量,且概率密度为 f Y ( y ) = f X [ h ( y ) ] ⋅ ∣ h ′ ( y ) ∣ f_{Y}(y) = f_{X}[h(y)] \cdot |h^{'}(y)| fY(y)=fX[h(y)]⋅∣h′(y)∣
证明
-
设 y = g ( x ) y = g(x) y=g(x)为严格单调增加的连续函数,定义域为 ( a , b ) ⊂ ( − ∞ , + ∞ ) (a, b) \subset (- \infty, + \infty) (a,b)⊂(−∞,+∞),值域为 ( α , β ) ⊂ ( − ∞ , + ∞ ) (\alpha, \beta) \subset (- \infty, + \infty) (α,β)⊂(−∞,+∞),则其反函数 x = h ( y ) x = h(y) x=h(y)在 ( α , β ) (\alpha, \beta) (α,β)上也为严格单调增加的连续函数( h ′ ( y ) > 0 h^{'}(y) > 0 h′(y)>0)
-
F Y ( y ) = P { Y ≤ y } = P { g ( X ) ≤ y } = P { X ≤ h ( y ) } = ∫ − ∞ h ( y ) f X ( x ) d x = ∫ a h ( y ) f X ( x ) d x F_{Y}(y) = P\set{Y \leq y} = P\set{g(X) \leq y} = P\set{X \leq h(y)} = \int_{- \infty}^{h(y)}{f_{X}(x) dx} = \int_{a}^{h(y)}{f_{X}(x) dx} FY(y)=P{Y≤y}=P{g(X)≤y}=P{X≤h(y)}=∫−∞h(y)fX(x)dx=∫ah(y)fX(x)dx
-
f Y ( y ) = F Y ′ ( y ) = f X [ h ( y ) ] ⋅ h ′ ( y ) , h ′ ( y ) > 0 f_{Y}(y) = F_{Y}^{'}{(y)} = f_{X}[h(y)] \cdot h^{'}(y) , h^{'}(y) > 0 fY(y)=FY′(y)=fX[h(y)]⋅h′(y),h′(y)>0
-
类似地,当 y = g ( x ) y = g(x) y=g(x)为严格单调减少时,有 f Y ( y ) = − f X [ h ( y ) ] ⋅ h ′ ( y ) , h ′ ( y ) < 0 f_{Y}(y) = - f_{X}[h(y)] \cdot h^{'}(y) , h^{'}(y) < 0 fY(y)=−fX[h(y)]⋅h′(y),h′(y)<0
定理2
- 设 Y = g ( X ) Y = g(X) Y=g(X), X X X为连续型随机变量, f X ( x ) f_{X}(x) fX(x)为 X X X的概率密度,若 y = g ( x ) y = g(x) y=g(x)为连续函数,它在不相重叠的区间 I 1 I_{1} I1, I 2 I_{2} I2, ⋯ \cdots ⋯上逐段严格单调,对应的反函数分别为 h 1 ( y ) h_{1}(y) h1(y), h 2 ( y ) h_{2}(y) h2(y), ⋯ \cdots ⋯,而且 h 1 ′ ( y ) h_{1}^{'}{(y)} h1′(y), h 2 ′ ( y ) h_{2}^{'}{(y)} h2′(y), ⋯ \cdots ⋯均连续,则 Y = g ( X ) Y = g(X) Y=g(X)为连续型随机变量,且其概率密度为 f Y ( y ) = ∑ i f X [ h i ( y ) ] ⋅ ∣ h i ′ ( y ) ∣ f_{Y}(y) = \sum\limits_{i}{f_{X}[h_{i}(y)] \cdot |h_{i}^{'}{(y)}|} fY(y)=i∑fX[hi(y)]⋅∣hi′(y)∣
- 约定:对使反函数 h ( y ) h(y) h(y)或其导数 h ′ ( y ) h^{'}(y) h′(y)无意义的 y y y, f Y ( y ) = 0 f_{Y}(y) = 0 fY(y)=0
示例2
问题
- 设质点 M M M随机地落在以原点为圆心以 R R R为半径的圆周上,且对弧长是均匀分布的,求质点 M M M的横坐标 X X X的概率密度
解答
- 设 Z Z Z为 x x x轴与 O M OM OM的夹角,则 X = R cos Z X = R \cos{Z} X=RcosZ,据题意知, Z ∼ U [ − π , π ] Z \sim U[- \pi, \pi] Z∼U[−π,π],其概率密度为
f Z ( z ) = { 1 2 π , − π ≤ z ≤ π 0 , 其他 f_{Z}(z) = \begin{cases} \cfrac{1}{2 \pi} , & - \pi \leq z \leq \pi \\ 0 , & 其他 \end{cases} fZ(z)=⎩ ⎨ ⎧2π1,0,−π≤z≤π其他
解法一
- F X ( x ) = P { X ≤ x } = P { R cos z ≤ x } = P { cos z ≤ x R } , ∀ x ∈ ( − ∞ , + ∞ ) F_{X}(x) = P\set{X \leq x} = P\set{R \cos{z} \leq x} = P\set{\cos{z} \leq \frac{x}{R}} , \forall x \in (- \infty, + \infty) FX(x)=P{X≤x}=P{Rcosz≤x}=P{cosz≤Rx},∀x∈(−∞,+∞)
- 当 x ≤ − R x \leq - R x≤−R时, F X ( x ) = 0 F_{X}(x) = 0 FX(x)=0, f X ( x ) = 0 f_{X}(x) = 0 fX(x)=0
- 当 x ≥ R x \geq R x≥R时, F X ( x ) = 1 F_{X}(x) = 1 FX(x)=1, f X ( x ) = 0 f_{X}(x) = 0 fX(x)=0
- 当 − R < x < R - R < x < R −R<x<R时
F X ( x ) = P { R cos z ≤ x } = P { cos z ≤ x R } = P { − π ≤ z ≤ − arccos x R } + P { arccos x R ≤ z ≤ π } = F Z ( − arccos x R ) − F Z ( − π ) + F Z ( π ) − F Z ( arccos x R ) \begin{aligned} F_{X}(x) &= P\set{R \cos{z} \leq x} = P\set{\cos{z} \leq \frac{x}{R}} \\&= P\set{- \pi \leq z \leq - \arccos{\frac{x}{R}}} + P\set{\arccos{\frac{x}{R}} \leq z \leq \pi} \\ &= F_{Z}(- \arccos{\frac{x}{R}}) - F_{Z}(- \pi) + F_{Z}(\pi) - F_{Z}(\arccos{\frac{x}{R}}) \end{aligned} FX(x)=P{Rcosz≤x}=P{cosz≤Rx}=P{−π≤z≤−arccosRx}+P{arccosRx≤z≤π}=FZ(−arccosRx)−FZ(−π)+FZ(π)−FZ(arccosRx)
- 于是, f X ( x ) = F X ′ ( x ) = f Z ( − arccos x R ) ⋅ 1 1 − ( x R ) 2 ⋅ 1 R + f Z ( arccos x R ) ⋅ 1 1 − ( x R ) 2 ⋅ 1 R = 1 π R 2 − x 2 f_{X}(x) = F_{X}^{'}{(x)} = f_{Z}(- \arccos{\frac{x}{R}}) \cdot \frac{1}{\sqrt{1 - (\frac{x}{R})^{2}}} \cdot \frac{1}{R} + f_{Z}(\arccos{\frac{x}{R}}) \cdot \frac{1}{\sqrt{1 - (\frac{x}{R})^{2}}} \cdot \frac{1}{R} = \frac{1}{\pi \sqrt{R^{2} - x^{2}}} fX(x)=FX′(x)=fZ(−arccosRx)⋅1−(Rx)21⋅R1+fZ(arccosRx)⋅1−(Rx)21⋅R1=πR2−x21
f X ( x ) = { 1 π R 2 − x 2 , ∣ x ∣ < R 0 , ∣ x ∣ ≥ R f_{X}(x) = \begin{cases} \frac{1}{\pi \sqrt{R^{2} - x^{2}}} , & |x| < R \\ 0 , & |x| \geq R \end{cases} fX(x)={πR2−x21,0,∣x∣<R∣x∣≥R
解法二
- 因为 x = R cos z x = R \cos{z} x=Rcosz在 ( − π , 0 ) (- \pi, 0) (−π,0)和 ( 0 , π ) (0, \pi) (0,π)上分别为严格单调的连续函数,相应的反函数分别为 z 1 = h 1 ( x ) = − arccos x R z_{1} = h_{1}(x) = - \arccos{\frac{x}{R}} z1=h1(x)=−arccosRx, z 2 = h 2 ( x ) = arccos x R , ∣ x ∣ < R z_{2} = h_{2}(x) = \arccos{\frac{x}{R}} , |x| < R z2=h2(x)=arccosRx,∣x∣<R
- f X ( x ) = f Z ( − arccos x R ) ∣ ( − arccos x R ) ′ ∣ + f Z ( arccos x R ) ∣ ( arccos x R ) ′ ∣ = 1 π R 2 − x 2 , ∣ x ∣ < R f_{X}(x) = f_{Z}(- \arccos{\frac{x}{R}}) |(- \arccos{\frac{x}{R}})^{'}| + f_{Z}(\arccos{\frac{x}{R}}) |(\arccos{\frac{x}{R}})^{'}| = \frac{1}{\pi \sqrt{R^{2} - x^{2}}} , |x| < R fX(x)=fZ(−arccosRx)∣(−arccosRx)′∣+fZ(arccosRx)∣(arccosRx)′∣=πR2−x21,∣x∣<R
- 当 ∣ x ∣ ≥ R |x| \geq R ∣x∣≥R时, f X ( x ) = 0 f_{X}(x) = 0 fX(x)=0
示例3
问题
- 已知随机变量 X X X的分布函数 F ( x ) F(x) F(x)是严格单调增加的连续函数,证明 F ( X ) F(X) F(X)服从 [ 0 , 1 ] [0, 1] [0,1]上的均匀分布
解答
-
由于 F ( x ) F(x) F(x)是严格单调增加的连续函数, 0 ≤ F ( x ) ≤ 1 0 \leq F(x) \leq 1 0≤F(x)≤1,因而当 0 ≤ y ≤ 1 0 \leq y \leq 1 0≤y≤1时, y = F ( x ) y = F(x) y=F(x)有反函数 x = F − 1 ( y ) x = F^{- 1}(y) x=F−1(y)
-
当 y < 0 y < 0 y<0时, F Y ( y ) = P { Y ≤ y } = P { F ( X ) ≤ y } = 0 F_{Y}(y) = P\set{Y \leq y} = P\set{F(X) \leq y} = 0 FY(y)=P{Y≤y}=P{F(X)≤y}=0
-
当 0 ≤ y < 1 0 \leq y < 1 0≤y<1时, F Y ( y ) = P { F ( X ) ≤ y } = P { X ≤ F − 1 ( y ) } = F ( F − 1 ( y ) ) = y F_{Y}(y) = P\set{F(X) \leq y} = P\set{X \leq F^{- 1}(y)} = F(F^{- 1}(y)) = y FY(y)=P{F(X)≤y}=P{X≤F−1(y)}=F(F−1(y))=y
-
当 y ≥ 1 y \geq 1 y≥1时, F Y ( y ) = P { F ( X ) ≤ y } = 1 F_{Y}(y) = P\set{F(X) \leq y} = 1 FY(y)=P{F(X)≤y}=1
F Y ( y ) = { 0 , y < 0 y , 0 ≤ y < 1 1 , y ≥ 1 F_{Y}(y) = \begin{cases} 0 , & y < 0 \\ y , & 0 \leq y < 1 \\ 1 , & y \geq 1 \end{cases} FY(y)=⎩ ⎨ ⎧0,y,1,y<00≤y<1y≥1
f Y ( y ) = { 1 , 0 ≤ y ≤ 1 0 , 其他 f_{Y}(y) = \begin{cases} 1 , & 0 \leq y \leq 1 \\ 0 , & 其他 \end{cases} fY(y)={1,0,0≤y≤1其他
-
即 Y ∼ U [ 0 , 1 ] Y \sim U[0, 1] Y∼U[0,1]
-
该命题广泛用于计算机仿真模拟中