| knowledge-database (beta) |
 |
Current group: comp.simulation
Simulation-based Parameter Estimation, any available tool?
|
|
| | From: | Neon | | Subject: | Simulation-based Parameter Estimation, any available tool? | | Date: | Thu, 06 Jan 2005 06:38:14 -0600 |
|
|
 | Hi, Dear All,
Not sure whether this problem could be named as "simulation based
parameter estimation", so I will describe this as below in detail:
For a complex stochasitc system, you can observe its output process
from its output, with one sample trace.
For this system, you can describe it with some complicated model in
which there are some parameters you are interested in. Simulation can
be developed with this model.
We can not get the close form of its output process, either mean value
or other order statistics. How could we estimate the parameters within
that model???
Without close form, maximum likelihood and lease square can not be used
for parameter estimation. So could we estimate the parameter in a
simulated approach?
How do you develop parameter estimation for complex system with some
output observations, wich can only be simulated other than modeled with
close form formula?
Thank you.
Neon
|
|
| | From: | Nicolaas Vroom | | Subject: | Re: Simulation-based Parameter Estimation, any available tool? | | Date: | Thu, 06 Jan 2005 06:55:59 -0600 |
|
|
| "Neon" schreef in bericht
news:1105009741.432630.96830@f14g2000cwb.googlegroups.com...
> Hi, Dear All,
>
> Not sure whether this problem could be named as "simulation based
> parameter estimation", so I will describe this as below in detail:
Where are the details ?
Do you really expect to get some valuable feedback with the level
of "detail" you have supplied ?
> For a complex stochasitc system, you can observe its output process
> from its output, with one sample trace.
>
> For this system, you can describe it with some complicated model in
> which there are some parameters you are interested in. Simulation can
> be developed with this model.
>
> We can not get the close form of its output process, either mean value
> or other order statistics. How could we estimate the parameters within
> that model???
Try the following:
http://www.google.com/search?hl=en&lr=&ie=ISO-8859-1&q=Parameter+estimation+
process+control
> Without close form, maximum likelihood and lease square can not be used
> for parameter estimation. So could we estimate the parameter in a
> simulated approach?
Try the following:
http://www.google.com/search?hl=en&lr=&ie=ISO-8859-1&q=Parameter+estimation+
simulation
> How do you develop parameter estimation for complex system with some
> output observations, wich can only be simulated other than modeled with
> close form formula?
>
> Thank you.
>
> Neon
>
Good Luck
Nicolaas Vroom
|
|
|
| | |
|
 |