Mate-in-1 testing

[BB+]

Here is a small collection of 10 mate-in-1 positions I constructed. I tried to include various "strange" elements (promotions, castling, en passant, captures, discovered/doubled checks).

position fen 2N5/4R3/2k3KQ/R7/1PB5/5N2/8/6B1 w - - 0 1
position fen 4N3/5P1P/5N1k/Q5p1/5PKP/B7/8/1B6 w - - 0 1
position fen 8/4N3/7Q/4k3/8/4KP2/3P4/8 w - - 0 1
position fen 8/8/K7/p7/k2N3R/p7/P7/8 w - - 0 1
position fen 1rk2N2/1p6/8/B1Pp4/B6Q/K7/8/2R5 w - d6 0 1
position fen 8/7B/8/3N4/8/1Q2B3/PPP5/rk2K2R w K - 0 1
position fen 3r4/7b/8/2p5/1pPnn3/1p1K4/2pP1k2/8 b - - 0 1
position fen r3kb2/5p2/8/n2K1p2/1pP5/3P1n2/b7/8 b q c3 0 1
position fen 8/8/3k4/b2n4/7r/3K4/P1ppp3/1q3b2 b - - 0 1
position fen 8/8/8/8/3kPp1q/bpn5/2pK1p2/8 b - e3 0 1

Obviously the list can be expanded. MultiPV testing can also be useful for some engines.

[Jeremy Bernstein]

Here is a small collection of 10 mate-in-1 positions I constructed. I tried to include various "strange" elements (promotions, castling, en passant, captures, discovered/doubled checks).

position fen 2N5/4R3/2k3KQ/R7/1PB5/5N2/8/6B1 w - - 0 1
position fen 4N3/5P1P/5N1k/Q5p1/5PKP/B7/8/1B6 w - - 0 1
position fen 8/4N3/7Q/4k3/8/4KP2/3P4/8 w - - 0 1
position fen 8/8/K7/p7/k2N3R/p7/P7/8 w - - 0 1
position fen 1rk2N2/1p6/8/B1Pp4/B6Q/K7/8/2R5 w - d6 0 1
position fen 8/7B/8/3N4/8/1Q2B3/PPP5/rk2K2R w K - 0 1
position fen 3r4/7b/8/2p5/1pPnn3/1p1K4/2pP1k2/8 b - - 0 1
position fen r3kb2/5p2/8/n2K1p2/1pP5/3P1n2/b7/8 b q c3 0 1
position fen 8/8/3k4/b2n4/7r/3K4/P1ppp3/1q3b2 b - - 0 1
position fen 8/8/8/8/3kPp1q/bpn5/2pK1p2/8 b - e3 0 1

Obviously the list can be expanded. MultiPV testing can also be useful for some engines.

Thanks. 1000ms movetime results in the "Houdini Engine Origins" (viewtopic.php?f=7&t=992&p=8939#p8937) discussion for HIARCS 13.2, Houdini 1.0, Ippolit 0.080a, Robbolito_0085g3 and Rybka 3.

[BB+]

The first versions with MultiPV capability are Houdini 1.03a and IvanHoe 81(?) I think. The only observed differences in the mate-in-1 orderings from a "go depth 1" search were in the 2nd position (where Houdini 1.03a differs from the prior version). Some of this is to be expected (everyone scans bitboards either forwards or backwards for instance, so there's really only two orderings in some cases).