Russell Whitaker wrote:
> >From: "Ross A. Finlayson" <raf@tiki-lounge.com>
> >
> >Russell Whitaker wrote:
> >
> > > >From: "Ross A. Finlayson" <raf@tiki-lounge.com>
> > > [SNIP]
> > > >
> > > >About the simulation, if you ask a question, and the simulation has an
> > > >answer,
> > > >and it's different than the expected result, then you can tell it why
> >and
> > > >it
> > > >could explain how it arrived at its conclusion.
> > > >
> > >
> > > What if the "expected result" is wrong, but the "simulated
> > > result" is the once actually consonant with reality? Where
> > > do you get the data for the "actual result"?
> > >
> > > This is why we do flight tests, and the profession of
> > > test pilot is not going away anytime soon, no matter how
> > > useful the simulations are.
> > >
> > > Russell
> > > _________________________________________________________________
> > > Get your FREE download of MSN Explorer at http://explorer.msn.com
> >
> >The flight test data goes back into the simulation. If you asked a bank
> >simulation if a plane would fly, it would probably not have an idea, if you
> >told it the expected result is that it would, then that's what it would
> >think
> >unless there was contradictory knowledge.
> >
>
> Ross,
>
> I think you missed my point. Please re-read with emphasis
> on my reaction to your phrase "different than the expected result".
> Keyword: "expected"
>
> Correct me if I'm wrong, but it seemed to me
> that you were advocating this cycle:
> 1.) run a simulation of a physical system
> 2.) if the simulation produces results *contrary to expectation*,
> "educate" the simulation program to produce _expected_
> results
> 3.) goto #1
>
> You then respond with what seems a partial modification -
> feed real world results back into the simulation... OK, but
> what does the follow-on utterance about domain specificity have to do
> with anything? We weren't talking about that.
>
> Russell
>
> _________________________________________________________________
> Get your FREE download of MSN Explorer at http://explorer.msn.com
Well, you'd expect the simulation, or expert rule set, or hybrid simulation, to
have a pretty good idea of the domain area you mention. Anything that it knows
would have to have been programmed into it, or it must have been programmed with
the ability to learn from information and given information. For example, if
there is a simulation of a wing surface flying the air, then the simulation
might know and do the equations to determine air pressure and velocity around
the wings. If the test trainer has sensors on the wings, and the sensor data
from the real world doesn't expect the actual data from the flight, where it si
not the expected result, then there is some reason why the expected result data
was different than reality. Perhaps some other element of the plane body
affected the lift and drag characteristics, or a wind shear affected the test
craft.
A flight simulator can't tell you very much about the weather besides how it
affects the plane. A weather simulator just has the weather, but if a bunch of
nuclear bombs explode, the weather would change.
Ross
-- Ross Andrew Finlayson Finlayson Consulting Ross at Tiki-Lounge: http://www.tiki-lounge.com/~raf/ "The best mathematician in the world is Maplev in Ontario." - Pertti L.
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:56:24 MDT