Re: [Bug-gnubg] Possible small UI simplification

[ya6hd4]

>
In ... case, I absolutely have to use the
non-deterministic choice and here's why : ..."

To me, and using the word absolutely, this is a wrong question .
And even nobody has absolutely to play backgammon.

Before changing things, whatever that may be, f.e. our behavoir , livestyle
etc. etc. we first has to find what the effect will be.
As you stated that the " its documentation is rather obscure"
I read that as "not very clear or not clear at all."
the first thing to do is to find the effect of non-determinitic and
deteministic noise and the  difference between them.
A long time ago a did  chat on fibs with inim and he told me
that  noone has examined the effect of the noise during gameplay
over a long period.
Why not doing that first just to find if non-deterministic  is indeed
not usefull.

Lucas

N.B. Before michael posted the issue of deterministic I was intended
to do it on local disk just because I like to know the diffrence
I bot Worlclass and another with non-determonistic
I bot worldclass againt deteministic
Using a very large dicefile a few 1.000.000 rolls
(So both have the same rolls )

-----Oorspronkelijk bericht----- 
From: Philippe Michel
Sent: Saturday, August 22, 2015 12:12 AM
To: address@hidden
Cc: address@hidden
Subject: Re: [Bug-gnubg] Possible small UI simplification

On Wed, 19 Aug 2015, address@hidden wrote:

> Is there any reasonof any importance, beside the not important  and VERY
> small UI simplification, to remove the non-deterministic noise ?

It seems useless to me (but I may have missed something, that's why I
asked) and its documentation is rather obscure. Can you read what follows
and complete the sentence "In ... case, I absolutely have to use the
non-deterministic choice and here's why : ..." ?


If noise is added to the evaluations GNU Backgammon will take a Box Müller
transform of a point in the unit circle and add to each possible outcome
probability. This means that the addition is random, but distributed so
that it's more likely to have an noise addition close to zero than a noise
addition close to the limit. The noise addition is limited to the number
you put into the noise field in the dialog box.

If you check the box Deterministic noise, the noise added to each
evaluation will be based on a sum of the bytes in the hash of the board
position, which (by the central limit theorem) should have a normal
distribution. In that way you will always have that same noise amount to a
position, since the noise added to the evaluation is only depending on the
position itself.