My own engine (no name yet).OliverBr wrote:It's because they prune, razor and nullmove like maniacs. In standard positions this might work well but in extraordinary position like a trapped queen (!) it's a flaw.Guetti wrote:No idea why the "professional" engines need depth 16 to find Rxg2. With only material evaluation my toy selects Rxg2 at depth 10; it wants to get on a pawn hunt before it switches to go after the queen.
Probably advanced search and eval stuff is a drawback in this position.
Pruning always has the danger that something is overlooked but it's is generally no problem, because such positions are rare and things will be seen in greater depth nontheless.
OliThink does not prune at all in standard search, so ply 12 already delivers the right move.
BTW, what "tool" are you using?
Testposition-(Medium). Which engine can find it?
Moderator: Ras
-
Guetti
Re: Testposition-(Medium). Which engine can find it?
-
smirobth
- Posts: 2307
- Joined: Wed Mar 08, 2006 8:41 pm
- Location: Brownsville Texas USA
Re: Testposition- 2 more.
I think 1...b6? and 1...a6? both lose to 2.Kg2 followed by 3.Rf3. This plan does not work against 1...a5!, since then White's king will not have enough time to get to the queen side to stop Black's a-pawn.George Tsavdaris wrote: Only 1...a5! **draws(the line that draws for white is just a single one between hundreds of possible replies). All other moves lose.
[d]5kbQ/pp3qp1/4p2p/3p4/8/4P1RP/1P3P2/6K1 b - - 0 1
*The source i had must be wrong. As i see 1...b6! or 1...a6! must draw too. But which engines will play any of these moves?
- Robin Smith