queen zone

Discussion of chess software programming and technical issues.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
User avatar
xr_a_y
Posts: 663
Joined: Sat Nov 25, 2017 1:28 pm
Location: France

queen zone

Post by xr_a_y » Fri Aug 09, 2019 6:26 pm

I'm working on a kind of queen safety now. I'm trying to use the same pattern as king, using a queen zone and summing threats in that zone, thinking that a "surrounded" queen can be in bad shape.

However, I wonder how this will be redundant with queen mobility ...

I'll give feedback soon.

User avatar
hgm
Posts: 23385
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: queen zone

Post by hgm » Sat Aug 10, 2019 6:46 am

It would essentially be 'safe mobility' for a Queen, an penalizing cases where this gets very low. This would be very rare if the total mobility is not very low also.

It would make more sense to extend evasions of an attack on the Queen just like you extend check evasions.

User avatar
xr_a_y
Posts: 663
Joined: Sat Nov 25, 2017 1:28 pm
Location: France

Re: queen zone

Post by xr_a_y » Sat Aug 10, 2019 8:10 am

hgm wrote:
Sat Aug 10, 2019 6:46 am
It would essentially be 'safe mobility' for a Queen, an penalizing cases where this gets very low. This would be very rare if the total mobility is not very low also.

It would make more sense to extend evasions of an attack on the Queen just like you extend check evasions.
Indeed, last night tests were not a success. I tested two things : tropism as with king and additional danger using safe mobility of the queen.
I'll try your idea next.

User avatar
xr_a_y
Posts: 663
Joined: Sat Nov 25, 2017 1:28 pm
Location: France

Re: queen zone

Post by xr_a_y » Sat Aug 10, 2019 8:28 am

xr_a_y wrote:
Sat Aug 10, 2019 8:10 am
hgm wrote:
Sat Aug 10, 2019 6:46 am
It would essentially be 'safe mobility' for a Queen, an penalizing cases where this gets very low. This would be very rare if the total mobility is not very low also.

It would make more sense to extend evasions of an attack on the Queen just like you extend check evasions.
Indeed, last night tests were not a success. I tested two things : tropism as with king and additional danger using safe mobility of the queen.
I'll try your idea next.
First result is that extending threats on queen outside of PV leads to a huge tree, totally impossible.
Extending only at PV node is in test but threats on queen extension are 10 times more than singular extension so I suspect this won't be good. An additional condition shall be used I guess, maybe a low mobility ...

User avatar
hgm
Posts: 23385
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: queen zone

Post by hgm » Sat Aug 10, 2019 9:53 am

Note that exendn threats is not the same as extending evasions; there will be moves after the threat which are not evasions. For check evasions such moves would be illegal.

Extending too many moves is bound to be problematic, though, as it is likely to lead to infinite recursion in some positions. (Which in orthodox Chess is not possible with checks.)

User avatar
xr_a_y
Posts: 663
Joined: Sat Nov 25, 2017 1:28 pm
Location: France

Re: queen zone

Post by xr_a_y » Sat Aug 10, 2019 11:38 am

hgm wrote:
Sat Aug 10, 2019 9:53 am
Note that exendn threats is not the same as extending evasions; there will be moves after the threat which are not evasions. For check evasions such moves would be illegal.

Extending too many moves is bound to be problematic, though, as it is likely to lead to infinite recursion in some positions. (Which in orthodox Chess is not possible with checks.)
ok i see. It is easy to extend queen move when queen is under attack. I'll check that.

User avatar
hgm
Posts: 23385
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: queen zone

Post by hgm » Sat Aug 10, 2019 1:25 pm

You could also limit this to evasions that are not capture of the attacker (after having tried that first), as some implementations of check extension do to avoid extending pointless checks.

The idea behind this is that you want to avoid a side that stands to lose something can too easily push the trouble over the horizon by attacking the Queen as delaying tactics. But full-ply extensions are always risky, as they open the possibility for infinite recursion when the condition for triggering them can be kept up indefinitely by both sides. For checks in orthodox Chess this is not possible. But for Queen attacks... Imagine two Queens on an empty board. They have a great capacity for making a new attack on each other every move, without repeating. Especially when they start next to each other. These attacks would keep them unprotected, though, so if such suicidal attacks are not extended this would not be a problem.

Post Reply