j*p
发帖数: 115 | 1 feynman kac是说那个pay off function的期望可以用pde的解算。
我的理解是 pde的解其实就是fundamental solution和初值的卷积.
但是那样的话，为啥两个积分，一个是卷积，一个不是。（懒得想了，大侠们指点一小
下吧） | x********o
发帖数: 519 | 2 none of them are convolution
【在 j*p 的大作中提到】
: feynman kac是说那个pay off function的期望可以用pde的解算。
: 我的理解是 pde的解其实就是fundamental solution和初值的卷积.
: 但是那样的话，为啥两个积分，一个是卷积，一个不是。（懒得想了，大侠们指点一小
: 下吧）
| j*p
发帖数: 115 | 3 你不搞pde吧。。。
我昨天推了一下，因为正态分布是偶函数，所以卷积和联合是一样的
【在 x********o 的大作中提到】
: none of them are convolution
| x********o
发帖数: 519 | 4 whether I am a pde person is not important.
only for special cases you will have convolution, for example, if the
fundamental solution is the heat kernel.
for most other cases, they are not convolution. it is simple an integral.
take the risk neutral pricing formula for example, under the BS framework,
the return is log-normal, which is
not normal, so you do not have convolution.
you have convolution if and only if the kernel satisfies G(t,x,y)=G(t,x-y)
【在 j*p 的大作中提到】
: 你不搞pde吧。。。
: 我昨天推了一下，因为正态分布是偶函数，所以卷积和联合是一样的
| j*p
发帖数: 115 | 5 I was not blaming you... sorry if it sounded like that.
I said that because I thought anyone who studied pdes would say it was a
convolution... even in the case you used.
Anyways, it not very important to me now. :) We can discuss about it
sometime later if you are still interested.
(Any DaNiu wants to comment a little bit?)
【在 x********o 的大作中提到】
: whether I am a pde person is not important.
: only for special cases you will have convolution, for example, if the
: fundamental solution is the heat kernel.
: for most other cases, they are not convolution. it is simple an integral.
: take the risk neutral pricing formula for example, under the BS framework,
: the return is log-normal, which is
: not normal, so you do not have convolution.
: you have convolution if and only if the kernel satisfies G(t,x,y)=G(t,x-y)
| x********o
发帖数: 519 | 6 no offense to you neither.
I was just answering your question.
【在 j*p 的大作中提到】
: I was not blaming you... sorry if it sounded like that.
: I said that because I thought anyone who studied pdes would say it was a
: convolution... even in the case you used.
: Anyways, it not very important to me now. :) We can discuss about it
: sometime later if you are still interested.
: (Any DaNiu wants to comment a little bit?)
|
|
