|
\[\prod _{k=1}^n \left(1+\exp\left(\frac{i t}{2^k}\right)\right)=\sum _{k=0}^{2^n-1} \exp \left(\frac{i k t}{2^n}\right)\]如何证明呢?
来源: Characteristic function (probability theory)
$d_k$为独立 Bernoulli$(\frac12)$ 分布的随机变量, 当$n→∞$时$\prod_{k=1}^nd_k/2^k$趋于$[0,1]$上的均匀分布.
Wolfram reference–CharacteristicFunction 下方展开“Applications”
Verify that the sum $\prod_{k=1}^nd_k/2^k$ where $d_k$ are independent identically distributed BernoulliDistribution[1/2] variates tends in distribution to UniformDistribution[] for large $n$:
- Subscript[cf, k] =
- CharacteristicFunction[
- TransformedDistribution[Subscript[d, k]/2^k,
- Subscript[d, k] \[Distributed] BernoulliDistribution[1/2]], t]
Copy the Code Use a combinatorial equality for product $\prod_{k=1}^n\text{cf}_k$\[\prod _{k=1}^n \left(\frac{1}{2}+\frac{1}{2} \exp\left(\frac{i t}{2^k}\right)\right)=\frac{\sum _{k=0}^{2^n-1} \exp \left(\frac{i k t}{2^n}\right)}{2^n}\] |
|