Actually, I don't see it that way at all. The decisions and the
information they're based on are cumulative. As you're dealt each new
card, you add that information to the previous cards, and to the
accumulating data regarding the past and present behavior of your
opponents, your stake, the pot, whatever. What's relevant is the
percentage of the total amount of data that changes at each cycle.
It's basically one new card being added to the total sum of
information at each turn, and that's likely a small percentage.
Furthermore, depending on how the program is designed, some of the
"rules" may in fact be encoded as data themselves, meaning that the
rate of data of mutation is even smaller.
On Nov 25, 2008, at 7:37 PM, Ken Banks wrote:
Thank you for your considered responses. I'm wondering if my system
will be able to capitalise on the advantages of the Rete algorithm.
Quoting from 'Jess In Action':
" in the worst case, where every fact changes on every cycle and there
is no network sharing between rules, then the performance for later
cycles is the same as well. This means Jess won't be very efficient
if you populate the working memory, run the pattern matcher for just
one cycle, and then reset working memory and repopulate it again from
scratch"
The basic mechanics of the rule engine will:
1. Receive hole cards. Make a decision.
2. Receive flop. Make a decision.
3. Receive turn. Make a decision.
4. Receive river. Make a decision.
5. Repeat.
My understanding of the situation is that the performance will always
be the same as the "first cycle" performance. If this is true is this
an inherent limitation given the problems domain in which I am
working. Or is there a way to build the rules which capitalise upon
Rete? If I always get "first cycle" performance than what is a "back
of the envelope" estimate on how this would compare to plain if/else
statements in java?
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users
[EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify [EMAIL PROTECTED]
.
--------------------------------------------------------------------
---------------------------------------------------------
Ernest Friedman-Hill
Informatics & Decision Sciences, Sandia National Laboratories
PO Box 969, MS 9012, Livermore, CA 94550
http://www.jessrules.com
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify [EMAIL PROTECTED]
--------------------------------------------------------------------
