Cfish 8 121516 for the macOS

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
hgm
Posts: 23772
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Cfish 8 121516 for the macOS

Post by hgm » Fri Dec 16, 2016 2:45 pm

These features (e.g.delaying response) are not available through the protocol, because they are GUI tasks, which should not concern the engine. Btw, there actually are UCI standard option UCI_LimitStrength /UCI_Elo that are supposed to fulfil these functions.

I have no idea what you mean by "randomizing nodes between x and y", or why it would be useful to do that.

MikeB
Posts: 3541
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Cfish 8 121516 for the macOS

Post by MikeB » Fri Dec 16, 2016 2:56 pm

hgm wrote:These features (e.g.delaying response) are not available through the protocol, because they are GUI tasks, which should not concern the engine. Btw, there actually are UCI standard option UCI_LimitStrength /UCI_Elo that are supposed to fulfil these functions.

I have no idea what you mean by "randomizing nodes between x and y", or why it would be useful to do that.
Will look into LimitStrength.


Randomizing nodes between x and y - limits strength with more variability in play. Appreciate your response - I will try a different apporach using limit strength - any addional info on limitstrength- or is a just a pretty straightforward spin type option ?

kinderchocolate
Posts: 411
Joined: Mon Nov 01, 2010 5:55 am
Full name: Ted Wong
Contact:

Re: Cfish 8 121516 for the macOS

Post by kinderchocolate » Fri Dec 16, 2016 3:05 pm

Michael,

The Stockfish code has skill levels implemented, it's done by randomly selecting a move from multi-pv. The code is not available in Cfish because Ronald didn't port it. Search "#if 0" in search.c, and you will see it. Maybe you can help him?

User avatar
hgm
Posts: 23772
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Cfish 8 121516 for the macOS

Post by hgm » Fri Dec 16, 2016 3:38 pm

UCI specs say this:

Code: Select all

		*  = UCI_LimitStrength, type check, should be false by default,
			The engine is able to limit its strength to a specific Elo number,
		   This should always be implemented together with "UCI_Elo".
		*  = UCI_Elo, type spin
			The engine can limit its strength in Elo within this interval.
			If UCI_LimitStrength is set to false, this value should be ignored.
			If UCI_LimitStrength is set to true, the engine should play with this specific strength.
		   This should always be implemented together with "UCI_LimitStrength".

MikeB
Posts: 3541
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Cfish 8 121516 for the macOS

Post by MikeB » Fri Dec 16, 2016 3:41 pm

kinderchocolate wrote:Michael,

The Stockfish code has skill levels implemented, it's done by randomly selecting a move from multi-pv. The code is not available in Cfish because Ronald didn't port it. Search "#if 0" in search.c, and you will see it. Maybe you can help him?
Thanks Ted - apprieciate your input. Will stick with the variable nodes apporach but using it with the limitstrength option. That is very clever coding , but that is limited in levels and I'm not fond of using sub optimal move due to scoring value randomness as opposed to simply not seeing the optimal move due to a node/search limitation. I have played with both and I prefer the latter- attribute it to personal preference , not knocking the work that was done there.

Rowen
Posts: 14
Joined: Tue Nov 15, 2016 12:19 pm
Location: Cheshire, England

Re: Cfish 8 121516 for the macOS

Post by Rowen » Fri Dec 16, 2016 9:35 pm

Hi guys

What is needed is a handicapping system that plays like a human, I gather this is difficult, we have open source stockfish which is superb and produces a super strong engine, but it would be good if we had some sort of open source program to produce a human type engine, with human styles and believable handicapping systems,different personalities etc.

As far as handicapping is concerned engines need to play sub optimal moves ( which they do), but not miss easy moves that even a beginner wouldnt miss, (which they often dont).

Thanks

MikeB
Posts: 3541
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Cfish 8 121516 for the macOS

Post by MikeB » Sat Dec 17, 2016 5:03 am

hgm wrote:UCI specs say this:

Code: Select all

		*  = UCI_LimitStrength, type check, should be false by default,
			The engine is able to limit its strength to a specific Elo number,
		   This should always be implemented together with "UCI_Elo".
		*  = UCI_Elo, type spin
			The engine can limit its strength in Elo within this interval.
			If UCI_LimitStrength is set to false, this value should be ignored.
			If UCI_LimitStrength is set to true, the engine should play with this specific strength.
		   This should always be implemented together with "UCI_LimitStrength".
Thank you HGM.

Post Reply