作者bmka (偶素米虫)
站内Statistics
标题Re: [问题] 请问有这种问题的蒙地卡罗模拟方法吗
时间Sat Mar 7 20:29:37 2009
※ 引述《clickhere (clickhere)》之铭言：
: ※ 引述《Jordan23 (ShihChung Chuang)》之铭言：
: : 以上的想法没什麽太大的问题,
: : 但是效率可能非常差.
: : Brownian bridge的方法确实可以让模拟有效率,
: : 但前提是你可以计算出 S(i)|S(j),S(k) j<i<k的分布.
: The idea is correct. I don't know what Brownian Bridge is.
: An efficient way may like the following.
: You can try log(S(t)) = log(S(t-1)) + N(0, 0.2).
: This may suggest to try an autoregression to get the distribution
: of log(S(t)) , and simulation directly from them independently.
Following the comment by clickhere, you may derive
the joint distribution of (log S(0), log S(1), log S(2) )
and the conditional distribution of ( log S(0), log S(1) )
given log S(2). Then try to figure out the distribution of
the order statistics of ( log S(0), log S(1) ) conditional on log S(2).
If you can do this, you should be able to generalize the
result to the setting described in your message.
※ 编辑: bmka 来自: 71.126.187.93 (03/07 20:34)