m*****n
发帖数: 3575 | 1 是不是很好做？
对一个单纯Poisson跳
只要在每个dt
上加一个
1{ U < 入dt }
U ～ Uniform[0,1]
就可以模拟了吧
需要做Risk Neutral
就再减去
入dt
诸位看看，我说的对不对，呵呵 | Q***5
发帖数: 994 | 2 Why bother with such a discrete simulation? You can simply get it more
accurately and efficiently by the following :
The pdf of the first jump is \lambda e^{-\lambda t}, to sample the first
jump, you only need to solve cdf = U, where U is uniformly sampled from [0,
1].
which leads to
t = -\frac{ln(1-U)}{\lambda},
Repeat this process of the second, third,... jumps.
【在 m*****n 的大作中提到】
: 是不是很好做？
: 对一个单纯Poisson跳
: 只要在每个dt
: 上加一个
: 1{ U < 入dt }
: U ～ Uniform[0,1]
: 就可以模拟了吧
: 需要做Risk Neutral
: 就再减去
: 入dt
|
|
