complex decisions algorithms

Expand Messages
• Hello, I have noticed that on page 619 of the international edition (figure 17.3) the best action for position 3,1 is to move North. However the figure 17.2
Message 1 of 2 , Jun 9, 2006
Hello,
I have noticed that on page 619 of the international edition (figure 17.3) the
best action for position 3,1 is to move North. However the figure 17.2 indicates
that the optimal policy sets West for that position.

Using my beta applet at
www.artificialidea.com/my_programs/complex_decisions/index.html
I get the same results (West using modified-policy-iterator and North using
value-iterator)

I wonder what I have missed.

Regards,
Ivan F. Villanueva B.
--
Ivan F. Villanueva B.
A.I. library: http://www.artificialidea.com
<<< The European Patent Litigation Agreement (EPLA) >>>
<<< will bring Software patents by the backdoor >>>
<<< http://www.no-lobbyists-as-such.com/florian-mueller-blog/epla/ >>>
<<< http://wiki.ffii.de/EplaEn >>>
• ... I ve found my mistake: the best action in figure 17.3 for 3,1 is *not* to move North, even if the value of the position 3,2 is higher than the one in 2,1.
Message 2 of 2 , Jun 15, 2006
On Sat, Jun 10, 2006 12:51:03AM +0200, Ivan F. Villanueva B. wrote:
> I have noticed that on page 619 of the international edition (figure 17.3) the
> best action for position 3,1 is to move North. However the figure 17.2 indicates
> that the optimal policy sets West for that position.
>
> Using my beta applet at
> www.artificialidea.com/my_programs/complex_decisions/index.html
> I get the same results (West using modified-policy-iterator and North using
> value-iterator)
>
> I wonder what I have missed.

I've found my mistake: the best action in figure 17.3 for 3,1 is *not* to move
North, even if the value of the position 3,2 is higher than the one in 2,1. The
best move in 3,1 is, nevertheless, West. The reason is that trying moving North
could lead the agent to position 4,1 with probability 0.1. But moving West makes
impossible to land in 4,1, and possible to go to 3,2 too.

BTW, if anyone have looked at the applet in the link above, I will appreciate