零相关与独立
Uncorrelated vs. Independent Normal Variables
题目详情
令 、 为两个标准正态随机变量。若它们的相关系数为 0,即 ,是否能推出 与 独立?
英文原题
Let and be two standard normal random variables. If their correlation is zero, , does this imply that and are independent?
解析
一般不能。
反例:令 ,再取独立的随机符号 ,且 ,令
则 仍服从 ,且
所以相关为 0。但 与 显然不独立(例如 几乎处处成立)。
补充:若进一步假设 是联合正态(bivariate normal),则“零相关”才等价于独立。
英文解析
Generally, no.
Counterexample: Let , and let be an independent random sign with . Define
Then also follows , and
so the correlation is zero. However, and are clearly not independent (for example, holds almost surely).
Supplement: If we further assume that is jointly normal (bivariate normal), then "zero correlation" is equivalent to independence.