[Next Message in Time] | [Previous Message in Time] | [Next Message in Topic] | [Previous Message in Topic]

Message ID: 14745
Date: Thu Feb 10 18:07:45 GMT 2000
Author: hho@bang-olufsen.dk
Subject: RE: Battle bard vs singing bard (was Does CHR affect Largos?)


>Message: 23
> Date: Thu, 10 Feb 2000 09:12:59 -0800
> From: "Perhaps a Princess..." <sarah@...>
>Subject: RE: Battle bard vs singing bard (was Does CHR affect Largos?)
>
>There is lots of debate among programmers about how random the
>random number generators used in computer programming really are,
>and the merits and flaws of various methods of generating a random
>number. If you play Nethack, you will notice the same types of
>problems with their RNG (random number generator) There is no
>perfect way of doing this. Just lots of options. Amazingly
>enough, generating a true random number is not as easy as it
>sounds.

Ohh I like this subject :o)
Actually most random generators will give fairly random numbers (depending on
what application youre using it for offcause) if you just use em right. The one
in Standard c (as far as I remember) is based and seeded with a 16bit value..
The trick, if you dont want to destroy the whiteness (and uniform distribution)
of this function is to NOT seed it ever again. just let it run its course and
you will get 2^16 different values which is infact very much random before the
pattern repeats it self.
Now the kind of random numbers you would want in a game should be a uniform
distribution (Every number has the same proberbility to show up) and it is
rather simple with sufficient output from the rand function to determine wheter
this holds true or not.
I would guess that the generator used in EQ itself is infact very uniform
distributed, but when you take a signal with _any_ ditribution and puts it
through a linear system/function. The output will change towards a gausian
distribution. With a gausian distribution most numbers will be most likely to
turn up around the center (if rand's range is 0 to 1 most numbers will be around
0.5). I dont think the latter is the case, but I dont know much about games or
the functions they might use for calculating succes/failure. I just know that
the standard random generators would give atleast 2^16 uniformly distributed
numbers before it repeats it self. Since 2^16 numbers are enough for most of
what I do I havent spent time looking into what happens if you go beyond that. I
expect that you will start seeing a system in the numbers.
Humm well if people are interested in Rand functions you should go here:

http://www.ulib.org/webRoot/Books/Numerical_Recipes/bookcpdf.html

Enough rambling from me today.. Time to go home

Musiker Silverdrum, bard of the 30th song, Bertox




>
>Sarah Heacock sarah@...
>Meara, level 13 Enchanter, Master Jeweler, owner Meara's Boutique of
>Mystiques, Phoenix, Solusek Ro /
>Faerain, level 18 Soldier of Tunare, Phoenix, Solusek Ro /
>Faesong, level 8(!) HalfElf Bard of Solusek Ro, Phoenix, Solusek Ro
>(currently hunting giant wasps in Misty Thicket)
>
>PS if you see my corpse in Misty Thicket today, keep it company? I had to go
>back to bed last night. I'll make the trip from Qeynos to pick it up
>tomorrow night.