在概率论中,柯尔莫哥洛夫二级数定理是关于随机变量序列的无穷求和收敛性的定理。该定理以苏联数学家安德雷·柯尔莫哥洛夫命名,可以用于证明强大数定律。
设 X n {\displaystyle X_{n}} 为独立的随机变量,如果 ∑ n = 1 ∞ Var ( X n ) {\displaystyle \sum _{n=1}^{\infty }\operatorname {Var} (X_{n})} 有限,那么 ∑ n = 1 ∞ ( X n − E [ X n ] ) {\displaystyle \sum _{n=1}^{\infty }(X_{n}-\operatorname {E} [X_{n}])} 几乎必然收敛。
不妨假设 X n {\displaystyle X_{n}} 的期望值均为0。设 S n = ∑ i = 1 n X i {\displaystyle S_{n}=\sum _{i=1}^{n}X_{i}} ,下面我们证明 lim sup n → ∞ S n − lim inf n → ∞ S n = 0 {\displaystyle \limsup _{n\to \infty }S_{n}-\liminf _{n\to \infty }S_{n}=0} 几乎必然成立。从而 S n {\displaystyle S_{n}} 几乎必然收敛。 对于任意正整数m ,
lim sup n → ∞ S n − lim inf n → ∞ S n = lim sup n → ∞ ( S n − S m ) − lim inf N → ∞ ( S N − S m ) ≤ 2 max k > m | ∑ i = m + 1 k X i | {\displaystyle \limsup _{n\to \infty }S_{n}-\liminf _{n\to \infty }S_{n}=\limsup _{n\to \infty }\left(S_{n}-S_{m}\right)-\liminf _{N\to \infty }\left(S_{N}-S_{m}\right)\leq 2\max _{k>m}\left|\sum _{i=m+1}^{k}X_{i}\right|}
因此,对于任意 ε > 0 {\displaystyle \varepsilon >0} 和正整数m,都有 P ( lim sup n → ∞ S n − lim inf n → ∞ S n ≥ ε ) = P ( max k > m | ∑ i = m + 1 k X i | ≥ ε 2 ) {\displaystyle \mathbb {P} \left(\limsup _{n\to \infty }S_{n}-\liminf _{n\to \infty }S_{n}\geq \varepsilon \right)=\mathbb {P} \left(\max _{k>m}\left|\sum _{i=m+1}^{k}X_{i}\right|\geq {\frac {\varepsilon }{2}}\right)}
由柯尔莫哥洛夫不等式,
P ( max k > m | ∑ i = m + 1 k X i | ≥ ε 2 ) ≤ 4 ε 2 ∑ i = m + 1 ∞ Var ( X n ) {\displaystyle \mathbb {P} \left(\max _{k>m}\left|\sum _{i=m+1}^{k}X_{i}\right|\geq {\frac {\varepsilon }{2}}\right)\leq {\frac {4}{\varepsilon ^{2}}}\sum _{i=m+1}^{\infty }\operatorname {Var} (X_{n})}
由方差之和有限的假设,当 m → ∞ {\displaystyle m\to \infty } 时,上式右边趋于0。这样就证明了 lim sup n → ∞ S n − lim inf n → ∞ S n = 0 a.s. {\displaystyle \limsup _{n\to \infty }S_{n}-\liminf _{n\to \infty }S_{n}=0{\text{ a.s.}}}