g*******s
发帖数: 59 | 1 one fair dice, how to use it to generate uniform random variable? | c******r
发帖数: 300 | 2 U[0,1] = \sum\limits_{i=1}B_i/2^i
where B_i are i.i.d Bernoulli. This is not "efficient" though, a slightly
better one is
U[0,1] = \sum\limits_{i=1}D_i/6^i
where the D_i is the result of rolling the dice.
anyway, you need countably infinite discrete random variables to generate
some continuous distribution (considering the entropy)
【在 g*******s 的大作中提到】
: one fair dice, how to use it to generate uniform random variable?
| m****o
发帖数: 114 | 3 i think he is talking about finite and discrete, not continuous
【在 c******r 的大作中提到】
: U[0,1] = \sum\limits_{i=1}B_i/2^i
: where B_i are i.i.d Bernoulli. This is not "efficient" though, a slightly
: better one is
: U[0,1] = \sum\limits_{i=1}D_i/6^i
: where the D_i is the result of rolling the dice.
: anyway, you need countably infinite discrete random variables to generate
: some continuous distribution (considering the entropy)
| e****d
发帖数: 333 | 4 I agree. several simulation books explain the algorithm for this problem. i
don't remember exactly. long time ago.
应该是 BYRON 那本书后一道习题。原题记得是生成0-9999的随机数。很久以前了。
【在 m****o 的大作中提到】
: i think he is talking about finite and discrete, not continuous
|
|
