Caroline,
Thanks for the email to ornet.
Just a couple of comments.
(1). I think you're unlikely to have used a full inverse transformation method
to generate Normal deviates. There is a partial inversion, developed originally
by Box & Muller and revised (I think) by George Marsaglia. It's fully detailed
in my book Computer Simulation in Management Science. As far as negative values
are concerned, yes that's always a risk with a Normal distribution (that's why
it's always nonsense when people say things like 'heights of adults are
Normally distributed'). The answer is to reject negative values and to
re-sample when they occur - which is what most commercial simulation software
probably does.
(2). On LogNormal: be careful that you use the correct parameters, since you
need to make an algebraic transformation between the Normal and LogNormal
parameters. (Again see my book, if you're desperate). For much more detail on
sampling from LogNormal, any book on simulation by George S. Fishman will have
full details.
Regards
Mike Pidd
> -----Original Message-----
> From: Caroline Collins [SMTP:[log in to unmask]]
> Sent: 15 April 1999 21:27
> To: ornet
> Subject: Discrete Event Simulation - Log-Normal Distribution
>
> I have written a DES model to simulate canal traffic passing through a
> flight of 6 locks. The system consists of two-way traffic(customers), with
> bi-directional locks (servers).
>
> In the model random numbers for the service procedure are generated using
> the inverse transformation method, to create 'a day in the life' of the lock
> system. I inititally modelled the lock procedure(service) using the Normal
> distribution, as it fits this distribution. However on occasions, due to
> high variance, there is a small chance that negative values can be produced.
> This is clearly not acceptable as it cannot take minus 0.5 minutes for a
> boat to enter a lock. In view of constraint I have now chosen Log-Normal
> because it is closely related to the Normal distribution, but is also is a
> positive and positively skewed distribution.
>
> Has anyone else used a Log-Normal distribution to describe length of service
> in a simulation model. Can anyone offer more information on this
> distribution as there appears to be very little written about it, compared
> to the Normal.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|