Brain Chess

This weekend I had the pleasure to escort my boys to the City Championship in Chess and they did exceptionally well in their age class U8 and U10. Julian managed to win the U8 and Jonas missed the tournament victory only by half a Buchholz point in the U10.

Congratulations. You are making progress much faster than me with my engine.

Julian Petzke (U8) playing Board 1 in the final round

Size matters not. Look at me. Judge me by my size, do you? Hmm? Hmm. And well you should not. - Yoda

Jonas Petzke coming in at 2nd place in U10

Jonas and Julian Petzke enjoying their success

Chess Engine Parameter Tuning

Currently I have only limited time for my project and I also want to take a bit of a break from programming. So after my release of iCE 3 I decided to go back to an area where I'm not so involved. I thought it is time to do a little evaluation parameter tuning again. Here the computer does the stuff on its own.

My idea was now to not tune the whole set but to keep the current values for material. Those are dominant terms (if the pawn value is badly wrong the eval is bad even if it has a good weight for the rook on a half open file). So if I keep the material values the other minor terms get more pressure and maybe they converge to a better set.

As the only tuning method that worked for me so far is the population based incremental learning (a GA type of optimization) I used this one. I used a population size of 256 and let the tuner run for almost 1500 generations.

After every 100 generations I took the current state of the tuning and let it run a small match (3000 games) against the base version.



After 100 generations the set was already only 50 ELO worse than the base version. This is pretty fast considering the fact that except from material all weights started from random values and converged to a set that gets about 3000 ELO on the CCRL scale. Over time the tuner was able to improve the set a bit further.

In the final iteration the final set was at -6 ELO (16000 games). After I optimized the build for the new set (the base version was also optimized) the new set was at +3 ELO.

In the tuning run the tuner played 750.000 games at very fast time controls.

So it confirms the method works but it is not able to improve the weights any further significantly. However it is better than the Texel method for iCE. After minimizing the evaluation error using this method the engine lost 24 ELO.

I think I will try some other methods in the future and hopefully can squeeze out a bit more than 3 ELO.

District Chess Championship U8

Julian Petzke (2nd from left) among the runner ups (2nd places) in the District Chess Championship 2016

Julian, my younger son, just played a good tounament and with 6 out of 7 points he finished 2nd. Only 1 buchholtz point was the distance to the 1st place. Anyway. Well done!

Time control was 75 minutes for 40 moves and 15 minutes for the rest of the game. This is an eternity for such young kids. And probably halve the games were already finished after 30 minutes. Julian was among the ones that took their time which was probably a good part of his success.

Now he is qualified for the U8 Chess championship of Saxony.

G 6 - International Gsei Web Tournament III ed.

iCE played this January as guest engine in the G6 - International Gsei Web Tournament III ed. It had a good start and only lost in the semi final against the later winner Chiron, a commercial Italian engine with 2:4. In the game for 3rd place iCE scored an incredible 6:0 victory against Senpai, the new engine by Fabien Letouzey.

 

iCE wins division 2 in Grahams Amateur Tournament

Five years ago early iCE started in the lowest divison 7 of this touranment. Now it is finally entering the top league. It was a long journey.

iCE 3 ELO gain

iCE ELO gain (absolute values relate to the private ELO list of Lars Hallerström)

The first testing results are now available. I rely envy those guys for their hardware. Looks like iCE 3 gained something in the area of about 140 ELO. This number matches the early results from the CEGT tests so is pretty reliable, at least for TC of up to 5 minutes per game.

Below is the attached cross table. Thank you, Lars, for running the test!

Tournament M5 Cross Table at Round 40

Pos	NAME	Rtg	Fed	Pts	Results	Pct
1	Ice 3.0 (64)	2980		607.5	+451 =313 -236	60,75%
2	Rybka 4.1 SSE42 (64-4cores)	3169		30.0	+24 =12 -4	75,00%
3	Critter 1.6a (64-4cores)	3199		29.0	+20 =18 -2	72,50%
4	Hannibal 1.5 (64-4cores)	3086		24.5	+15 =19 -6	61,25%
5	Texel 1.05 (64-4cores)	3066		24.0	+17 =14 -9	60,00%
6	Chiron 2 (64-4cores)	3109		23.0	+18 =10 -12	57,50%
7	Spike 1.4 (4cores)			21.5	+12 =19 -9	53,75%
8	Senpai 1.0 sse42 (64-4cores)	3052		20.5	+13 =15 -12	51,25%
9	Gaviota 1.0 (64-4cores)	2951		19.5	+9 =21 -10	48,75%
10	Ex 7.88b (64-4cores)	2866		18.5	+9 =19 -12	46,25%
11	Arasan 18.1 (64-4cores)	2914		17.5	+8 =19 -13	43,75%
12	SmarThink 1.80 (64)	2988		17.0	+10 =14 -16	42,50%
13	BobCat 7.1 (64-4cores)	2800		16.5	+9 =15 -16	41,25%
	Spark 1.0 (64-4cores)	2981		16.5	+10 =13 -17	41,25%
14	Cheng 4.37 rc0 (64-4cores)	2887		15.0	+10 =10 -20	37,50%
	Crafty 25.0 JA (64-4cores)	2892		15.0	+8 =14 -18	37,50%
15	Scorpio 2.7.6 JA (64-4cores)	2887		14.0	+8 =12 -20	35,00%
16	Sjeng WC2008 (64-4cores)	2912		13.5	+8 =11 -21	33,75%
17	Tornado 7 (64-4cores)	2944		10.5	+6 =9 -25	26,25%
18	Deuterium 14.01 (64)	2946		8.0	+5 =6 -29	20,00%
19	Pedone 1.3 (64-4cores)	2699		7.5	+5 =5 -30	18,75%
20	Rodent 1.7 (64)	2883		7.0	+3 =8 -29	17,50%
21	Octo 5190 (64-4cores)	2865		6.5	+3 =7 -30	16,25%
	Ktulu 9			6.5	+2 =9 -29	16,25%
22	Fruit 2.3.1	2937		6.0	+2 =8 -30	15,00%
23	Bug2 1.9 (64-4cores)	2829		5.0	+2 =6 -32	12,50%
Tournament start		05 Jan 2016 00:00
Latest update		10 Jan 2016 17:45
Site/ Country		Skrutten644
Level		300

iCE 3 gets released

I'm finally ready to release version 3 of my little chess engine project. It is still only single core as I haven't found the motivation to implement SMP. I'm still on the way to make the engine smarter not faster. So SMP is maybe something for the next release then.

What did change in v3

UCI Options Cleanup The UCI options of iCE are restructured and renamed. Details can be found here: iCE 3 UCI Options

Multi PV support iCE is now supporting Multi PV mode.

Bugfix The ponder protocol implementation is fixed. In Shredder GUI it did not work.

Evaluation: A bit of weight tuning was perfromed. However most changes failed or had only minor impact.

Search Changes: iCE 2 added History Heuristics, Late Move Pruning, Razoring and Counter Move Heuristics. For iCE 3 I optimized the conditions a bit further to make smarter pruning decisions.

Books: I finally touched my book code again. It was necessary as the books contained some errors coming from changes to the internal move format. I also improved the book building framework to use perfromance data of iCE in certain lines.

Time Management: I changed the algorithm for time allocation to not use all available time on the last move before TC.

What do you get

I make 3 versions of iCE available

• ice3-x64-modern - This is the release you probably want to use. It makes use of some HW instructions found on recent CPU.
• ice3-x64 - This is the fall back release if your HW doesn't support the instructions used in the modern compile
• ice3-x32 - As the name suggests, if you still on a 32 bit platform like Windows XP use this one

The versions along with new books can be found at my download site

http://www.fam-petzke.de/cp_download_en.shtml

How to use

iCE is only a chess engine so it integrates into a graphical interface and communicates with it via a protocol. All interfaces that support the UCI (Universal Chess Interface) are usable. Among the available interfaces are

• Shredder Classic GUI
• Fritz GUI
• Arena

The above interface are tested in at least some version, but others should work as well. Pick your favorite choice.

iCE 3 playing in the Shredder GUI