Re: ceerano and the seed

[Vernon Hamberg <vhamberg@xxxxxxxxx>]

Peter

Would it make sense to treat a user-supplied seed the same way, since it is
limited to a 4-byte signed integer?

BTW, what does the expression "multiplicative congruential" mean? The first
word is simple enough. Are all forms of this kind of generator essentially
the same, just different constants? I've seen things on the net about
2^31-1 and 2^61-1 (or something like that)

I seem to recall problems with skewed distributions when generating
integers between 2 values. Does this "bucketizing" diminish the quality of
the distribution? The random number generator actually returns values
between 0 and 1.

Thanks

At 12:48 AM 9/17/02 -0400, you wrote:
> Vernon,
>
> Yes, I am talking about the theory of generating pseudorandom numbers.
>
> I did a lot of work with these when I was doing Processing Plant Simulations
> and Statistical Process Control a long time ago with General Foods.
>
> So, hopefully CEERAN0 uses GMT in the form HHMMSSTHT and then multiplies by
> 8 and adds 1 to get a large odd seed (since odd numbers have a chance of
> being prime).
>
> Peter.
>
> ----- Original Message -----
> From: "Vernon Hamberg" <vhamberg@attbi.com>
> To: <midrange-l@midrange.com>
> Sent: Tuesday, September 17, 2002 12:05 AM
> Subject: Re: ceerano and the seed
>
>
> > Peter
> >
> > I'm not sure of the context. The usage instructions for CEERAN0 say that
> > the seed should be a non-negative integer between 0 and 2,147,483,646.
> Zero
> > makes it generate its own seed from the GMT at the time the function is
> > called. And it returns a value that can be used as the next seed. Of
> > course, the new seed is generated by the original process, hence, as Leif
> > says, it might not be that good an idea to use it. That was why I
> suggested
> > using 0, as that forces a somewhat unpredictable value for the seed each
> time.
> >
> > Are you speaking of the theory of generating pseudorandom numbers, that
> > uses a couple large relatively prime numbers? Am I anywhere close? I've
> > never studied the theory that closely, I just want a function I can use
> > that I can trust to generate even distributions.
> >
> > The rand() function in SQL can take a seed or not. Without a seed (a
> > smallint or integer) it must generate its own seed, as it starts with a
> > different sequence each time. If you use a seed, such as rand(100), the
> > sequence of random floating point numbers is the same each time, as is
> > expected. CEERAN0 has the same behavior
> >
> > Vern