| |
 |
|
|
Science Forum Index » Statistics - Math Forum » The product of Gaussian process and Poisson process...
Page 1 of 1
|
| Author |
Message |
| luolg05... |
 Posted: Wed Jun 25, 2008 5:14 am |
|
|
|
Guest
|
| How're its statistics properties? Has anyone done it before? |
|
|
| Back to top |
|
| luolg05... |
| Posted: Wed Jun 25, 2008 6:18 am |
|
|
|
Guest
|
On 6月26日, 上午12时09分, "David Jones" <dajx... at (no spam) ceh.ac.uk> wrote:
Quote: luolg05 wrote:
How're its statistics properties? Has anyone done it before?
You might mean two things, which are different:
(i) Gaussian times Poisson process, meaning you get isolated "events" each having an associated value from the Gaussian process. This would be one type of "labelled Poisson process", on which quite a bit of theory has been done.
(i) Gaussian times Poisson count process, meaning that you get a continuous time series as output. I don't recall seeing this, but the theory of moments etc. would be easy, at least if indpendence of the two processes is assumed. Not sure if anything can realistically be modelled this way.
David Jones
Thank you very much.
I mean the second one.And the indpendence of the two processes is
assumed. |
|
|
| Back to top |
|
| luolg05... |
| Posted: Wed Jun 25, 2008 6:35 am |
|
|
|
Guest
|
On 6月26日, 上午12时18分, luolg05 <move... at (no spam) gmail.com> wrote:
Quote: On 6月26日, 上午12时09分, "David Jones" <dajx... at (no spam) ceh.ac.uk> wrote:
luolg05 wrote:
How're its statistics properties? Has anyone done it before?
You might mean two things, which are different:
(i) Gaussian times Poisson process, meaning you get isolated "events" each having an associated value from the Gaussian process. This would be one type of "labelled Poisson process", on which quite a bit of theory has been done.
(i) Gaussian times Poisson count process, meaning that you get a continuous time series as output. I don't recall seeing this, but the theory of moments etc. would be easy, at least if indpendence of the two processes is assumed. Not sure if anything can realistically be modelled this way.
David Jones
Thank you very much.
I mean the second one.And the indpendence of the two processes is
assumed.
And the Gaussian process is Wide smooth. |
|
|
| Back to top |
|
| David Jones... |
| Posted: Wed Jun 25, 2008 11:09 am |
|
|
|
Guest
|
luolg05 wrote:
Quote: How're its statistics properties? Has anyone done it before?
You might mean two things, which are different:
(i) Gaussian times Poisson process, meaning you get isolated "events" each having an associated value from the Gaussian process. This would be one type of "labelled Poisson process", on which quite a bit of theory has been done.
(i) Gaussian times Poisson count process, meaning that you get a continuous time series as output. I don't recall seeing this, but the theory of moments etc. would be easy, at least if indpendence of the two processes is assumed. Not sure if anything can realistically be modelled this way.
David Jones |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Fri Dec 05, 2008 10:35 am
|
|