Bikjump V3.0 is now available for download

[Bart Weststrate]

Thematic in many of the games is the "poisoned pawn". Bikjumps queen likes that pawns.

[pgn][Event "?"]
[Site "?"]
[Date "2026.04.10"]
[Round "152"]
[White "Bikjump"]
[Black "SM_2182b"]
[Result "0-1"]
[ECO "A30"]
[GameDuration "00:01:52"]
[GameEndTime "2026-04-10T10:11:15.801 W. Europe Daylight Time"]
[GameStartTime "2026-04-10T10:09:22.931 W. Europe Daylight Time"]
[Opening "English"]
[PlyCount "65"]
[Termination "adjudication"]
[TimeControl "60/60+1"]
[Variation "Symmetrical variation"]

1.c4
{book}
1...c5
{book}
2.Nf3
{book}
2...Nf6
{book}
3.g3
{book}
3...d5
{book}
4.cxd5
{book}
4...Nxd5
{book}
5.Qb3
{+0.13/13 2.1s}
5...Nc6
{-0.07/14 2.5s}
6.Nc3
{-0.02/12 2.2s}
6...Be6
{+0.08/13 1.7s}
7.Qxb7
{+0.24/13 1.7s}
7...Ndb4
{+0.86/16 1.6s}
8.Qb5
{+0.21/15 2.0s}
8...Bd7
{+0.77/15 2.1s}
9.Kd1
{+0.08/14 2.0s}
9...e5
{+0.90/16 2.0s}
10.e3
{+0.06/15 3.9s}
10...Be7
{+0.79/14 2.6s}
11.Qe2
{+0.34/13 2.0s}
11...Be6
{+0.85/13 1.5s}
12.h4
{+0.22/12 1.9s}
12...O-O
{+0.85/12 1.7s}
13.Bh3
{+0.07/12 2.0s}
13...Bxh3
{+1.18/14 1.6s}
14.Rxh3
{+0.37/13 2.0s}
14...Qd7
{+1.07/12 1.5s}
15.Rh1
{+0.24/12 1.4s}
15...Rad8
{+0.95/12 3.8s}
16.a3
{+0.42/13 1.7s}
16...Nd3
{+1.00/15 1.2s}
17.Ne1
{+0.43/14 2.0s}
17...c4
{+0.90/16 1.9s}
18.Nxd3
{+0.38/15 2.0s}
18...cxd3
{+0.93/16 2.1s}
19.Qh5
{+0.29/14 1.7s}
19...Rfe8
{+0.92/13 1.6s}
20.b4
{+0.49/12 1.7s}
20...Qe6
{+0.88/12 1.5s}
21.Qf3
{+0.57/13 5.0s}
21...f5
{+1.42/14 1.1s}
22.Qg2
{-0.30/13 2.3s}
22...e4
{+1.62/16 3.3s}
23.Rb1
{-1.14/15 5.2s}
23...Ne5
{+2.19/12 1.1s}
24.h5
{-1.13/14 1.3s}
24...Nf3
{+2.79/14 1.6s}
25.g4
{-2.55/14 2.6s}
25...fxg4
{+3.66/13 1.4s}
26.h6
{-6.91/14 1.9s}
26...Bf6
{+4.33/15 2.9s}
27.Rh5
{-6.95/15 1.9s}
27...Bxc3
{+7.98/14 1.1s}
28.dxc3
{-9.31/14 1.8s}
28...Qc6
{+9.50/17 1.3s}
29.Qxf3
{-9.99/14 1.3s}
29...gxf3
{+9.90/16 1.5s}
30.Bd2
{-12.79/17 1.9s}
30...Qg6
{+12.53/21 1.3s}
31.Rh1
{-12.97/19 1.5s}
31...Qg2
{+13.04/22 1.4s}
32.Re1
{-13.87/20 1.5s}
32...gxh6
{+13.18/22 1.9s}
33.Kc1
{-13.12/19 1.7s, Black wins by adjudication}
0-1[/pgn]

[abik]

Thematic in many of the games is the "poisoned pawn". Bikjumps queen likes that pawns.

Thanks for the games and the analysis. It definitely gives me inspiration to start working on v3.1 right away! Btw, are you sure that SM_2182b is just a "bean counter"? It seems to play positionally quite well?

Also, I look forward to get some positive results in tournament play against other engines, just not to lose motivation

[abik]

Hello. Do you also have the Android version? Thank you

Working on that. I will distribute this as chess engine package (since applications that target API level 29 and run on Android 10 and later, such as my own Chess for Android, no longer can install engines from SD card due to new security-related restrictions on W^X violations; the OEX and Chessbase compatible protocols luckily still work).

[Bart Weststrate]

Thematic in many of the games is the "poisoned pawn". Bikjumps queen likes that pawns.

Thanks for the games and the analysis. It definitely gives me inspiration to start working on v3.1 right away! Btw, are you sure that SM_2182b is just a "bean counter"? It seems to play positionally quite well?

Also, I look forward to get some positive results in tournament play against other engines, just not to lose motivation

Yes it is a beancounter. Except for the well known tapered "Pesto/Rofchade" PST (which kickstarts a pst) and some scoring for pawns, bishoppair and some mobility from movegen and small bonus for right to move it has nothing more.

This is the full eval in Powerbasic:

SUB evaluate

REM tapered pst + materiaal
tot_pes = MIN(white_pesto + black_pesto,pesto_max)

phase = tot_pes
phase = (phase * 256 + (Pesto_max / 2)) / Pesto_max
positional_evaluation = ((pst_eval_eg * (256 - phase)) + (pst_eval_mg * phase)) / 256

pawn_evaluation ''scores isolani, double, tripple, passed, backward
positional_evaluation += pawn_struct_eval/2

'' loperpaar
positional_evaluation += (BIT(white_pesto,twee_lopers)*16)
positional_evaluation -= (BIT(black_pesto,twee_lopers)*16)

IF BIT(flags,white_to_movex) THEN
positional_evaluation +=5
ELSE
positional_evaluation -=5
END IF

positional_evaluation +=wmob*2
positional_evaluation -=bmob*2

''-------------------------------------------------------------------------
''-------------------------------------------------------------------------
''LET OP: DEZE REGEL RETOURNEERT DE EVAL - MOET DUS LAATSTE REGEL ZIJN....

full_eval_result= positional_evaluation
''-------------------------------------------------------------------------
''-------------------------------------------------------------------------

END SUB

[Bart Weststrate]

Aart if i can download Bik 2.1.... Than i run match 1 again. Than we have some clue about the elo difference between 2.1 and 3.0 that is against the beancounter.

[abik]

Aart if i can download Bik 2.1.... Than i run match 1 again. Than we have some clue about the elo difference between 2.1 and 3.0 that is against the beancounter.

I have put v2.0.1 back on the Aart's Chess Page under the "older versions" section.

[Bart Weststrate]

Aart if i can download Bik 2.1.... Than i run match 1 again. Than we have some clue about the elo difference between 2.1 and 3.0 that is against the beancounter.

I have put v2.0.1 back on the Aart's Chess Page under the "older versions" section.

Thanks. Test is running.

[Bart Weststrate]

Aart if i can download Bik 2.1.... Than i run match 1 again. Than we have some clue about the elo difference between 2.1 and 3.0 that is against the beancounter.

I have put v2.0.1 back on the Aart's Chess Page under the "older versions" section.

Aart i think that the link leads to the 3.0 zip... results until now are more or less the same....

[abik]

Aart i think that the link leads to the 3.0 zip... results until now are more or less the same....

It is really late at night here. I changed the visible name, not the actual link Fixed now.

Engine should identify itself as follows:

Code: Select all

uci
id name BikJump v2.01 (64-bit)
id author Aart J.C. Bik
option name Hash type spin default 32 min 1 max 65536
option name NalimovPath type string default <empty>
option name NalimovCache type spin default 8 min 1 max 2048
option name Ponder type check default true
uciok

Latest and greatest version:

Code: Select all

uci
id name BikJump v3.0 (64-bit)
id author Aart J.C. Bik
option name Hash type spin default 64 min 1 max 65536
option name NalimovPath type string default <empty>
option name NalimovCache type spin default 32 min 1 max 2048
option name Ponder type check default true
option name MultiPV type spin default 1 min 1 max 256
option name Processors type spin default 1 min 1 max 8
option name UCI_LimitStrength type check default false
option name UCI_Elo type spin default 1500 min 1000 max 2000
uciok

[Bart Weststrate]

Bikjump 3.0 (60/20+1) first run
Score of SM_2182b vs Bikjump: 320 - 128 - 52 [0.692] 500
... SM_2182b playing White: 162 - 68 - 20 [0.688] 250
... SM_2182b playing Black: 158 - 60 - 32 [0.696] 250
... White vs Black: 222 - 226 - 52 [0.496] 500
Elo difference: 140.6 +/- 31.0, LOS: 100.0 %, DrawRatio: 10.4 %

Bikjump 3.0 (60/20+1) second run (verify)
Score of SM_2182b vs Bikjump: 327 - 123 - 50 [0.704] 500
... SM_2182b playing White: 166 - 61 - 23 [0.710] 250
... SM_2182b playing Black: 161 - 62 - 27 [0.698] 250
... White vs Black: 228 - 222 - 50 [0.506] 500
Elo difference: 150.5 +/- 31.4, LOS: 100.0 %, DrawRatio: 10.0 %

Difference probably due to "timing noise". Now awaiting the real 2.1 Bikjump.

Edit: next match running using the real 2.1.