Horizon effect

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

Werewolf
Posts: 1797
Joined: Thu Sep 18, 2008 10:24 pm

Re: Horizon effect

Post by Werewolf »

Nordlandia wrote:Show an position were horizon effect is still present today.

Example position.

[D]2r5/6pk/7p/8/8/7P/1rbR1RPK/8 b - - 0 1
User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

Re: Horizon effect

Post by Nordlandia »

Werewolf wrote:
Nordlandia wrote:Show an position were horizon effect is still present today.

Example position.

[D]2r5/6pk/7p/8/8/7P/1rbR1RPK/8 b - - 0 1
Position is draw right?
User avatar
hgm
Posts: 27816
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Horizon effect

Post by hgm »

No, this is not the horizon effect I am talking about. Classical horizon effect is when the engine starts to make pointless sacrifices to push an unavoidable loss over the horizon. The shown diagram is just a case where the undesired result is so far away (not anywhere near the horizon) that no desparate acts are needed to make it invisible.
User avatar
hgm
Posts: 27816
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Horizon effect

Post by hgm »

Nordlandia wrote:Show an position were horizon effect is still present today.

Example position.
This is a good example of what I mean:

Code: Select all

 11	-19.62	4165.6M	236:44.00	f11i8 i6o12+ o13o12? p11o14! i8i5! i3i5 h13i10 i2i3 i10k10 o14o10 g13h13 g4h4 h13g13
 10	-19.67	3584.1M	201&#58;03.21	<done>
 10	-19.67	1529.1M	88&#58;35.67	f11i8 i6o12+ o13o12? p11o14! i8i5!
  9	-19.67	1368.5M	78&#58;17.46	<done>
  9	-19.67	1216.4M	69&#58;48.14	f11i8 i6o12+ o13o12? p11o14! i8i5!
  9	-23.38	361.8M	21&#58;04.31	k13c5+ c4c5 f11b7! p11p13! m12m11 p13n14! o16n14 c3b2 b7d7!
  8	-17.14	214.1M	12&#58;38.95	<done>
  8	-17.14	105.1M	6&#58;06.43	k13c5+ c4c5 f11i8 p11n12! i8i5! h3i5 m14l13 i5g6 d16e15 n12l11!
  7	-14.04	52.5M	3&#58;05.17	<done>
  7	-14.04	35.1M	2&#58;03.65	k13c5+ c4c5 f11b7! c5c12+ j13j4+ p11n12! b7e4! c2e4
  6	-12.40	25.0M	1&#58;28.34	<done>
  6	-12.40	24.0M	1&#58;24.46	k13c5+ c4c5 f11b7! p11n12! b7c5! i6n11 k14n11? n12o14! c5b3! o14n15! b3e3! g3e3! j13j4+
  6	-16.34	15.0M	0&#58;54.20	k13a3+ p11n12! f11b7! i6n11 k14n11? n12o14! b7c5!
  6	-18.37	4.5M	 0&#58;18.51	f11b7! p11p13! i15g14 p13n12! b7c5! n12l14! j14l14!
  5	 -8.97	989788	0&#58;03.84	f11b7! p11n12! b7c5! n12o14! c5b3! o14l14! j14l14!
  4	-10.56	160487	0&#58;00.67	f11b7! p11n12! b7c5! n12o14! c5b3! o14n15! j13j4+ i3j4,j4i3 b3c2! n15l14! j14l14! g4g13+
  3	 -8.97	82378	0&#58;00.32	f11b7! p11n12! b7c5!
  3	-17.13	50246	0&#58;00.18	f11f5! g3f5! n14o15
  3	-21.01	42639	0&#58;00.17	j13j4+ p11n12! f11b7! n12o14! b7c5! g4g13+
  2	-12.56	9930	 0&#58;00.03	j13j4+ p11n12!
  2	-17.01	8616	 0&#58;00.03	f11f5! g3f5!
  2	-20.16	8609	 0&#58;00.03	g13g12 i4o10 o13o10? p11o14! f11b7! o14n15! b7c5! n15l13! c5e3! g3e3! j14l13!
  1	-10.51	2569	 0&#58;00.01	g13g12
At 6 ply it sees its favorite move f11b7 is badly punished; it then switches to pointless sacrifices like k13c5+ (captures a Pawn with promotion to a powerful piece that cannot be ignored, or it would start checking), which is immediately captured by c4c5, and then the depth is shortened enough to make f11b7 look good again. While in fact the sacrifice did not change anything, other than lowering the score.

At 9 ply it finds that f11b7 fter the sac gets refuted by the same move as that made it plummet iat 6 ply. Only after 70 minutes it then switches to a reasonable move, f11i8.

Image