[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] influence and more

From: Heikki Levanto
Subject: Re: [gnugo-devel] influence and more
Date: Mon, 25 Nov 2002 22:45:18 +0100
User-agent: Mutt/1.4i

On Mon, Nov 25, 2002 at 09:06:54PM +0100, Inge Wallin wrote:
> Never mind the sigmoid.  Any other function with the same nice curve
> might do.  Do you agree that a non-linear curve might be a good idea? 

I have not followed this discussion very well, so I am little uncertain
where you want to use the (non?)linear function. But I am sligthly worried
about the "same nice curve" kind of argument.

I can see two possibilities:

1) We are adding the function values to get a composite value (for example
influence radiating from different sources). In such case we ought to have
some mathematical argument to choose which function to use. This would call
for more discussion than what I have seen so far.


2) The values are never added, just compares to various limits. In that case
the non-linear function does not matter at all, as long as it is monotonic.
Changing the function just means we need different limits for the value.

So, which way is it?


Heikki Levanto  LSD - Levanto Software Development   <address@hidden>

reply via email to

[Prev in Thread] Current Thread [Next in Thread]