Rook vs. Bishop – Keysquares

Here’s a little atomic endgame problem. White to move.

8/p7/P1p1p2k/4P3/2Pp4/3P3b/3K4/1R6 w - - 0 1
White to move

FEN: 8/p7/P1p1p2k/4P3/2Pp4/3P3b/3K4/1R6 w – – 0 1

What’s the evaluation of this position? Is there a winning move?

Basic evaluation

White wins. But he has to find the only winning move right now in this position.


White only wins with 1. c5!!

The other moves will bedeclined by blacks 1…c5 and the position is a draw. Like 1. Rb7? c5!! 2. Rxa7 Bg2 and the bishop easily stops the a-pawn.

But after 1. c5!! the position of the bishop is really bad. The long diagonal h1-a8 is closed now and the bishop is to slow to stop the a-pawn. One variation is 1…Bf1 (spite check) 2. Kc1 Be2 3. Rb7! Bd1 (3…Bh5?? 4. Rh7+ Kg6 5. Rxh5#) 4. Rxa7 Ba4 5. a7 Bb5 6. a8Q +- The new queen wins easy against the bishop.


The darn exception

A rule is a nice thing. Because, if you know something, you dont need to calculate it. So you can save your time for something else.

BUT, there are “No rules without exceptions!”
I had a new idea how to figure out drawish positions in the endgame rook versus bishop. This type of endgame is a draw if the pawnstructure is symetric.

Like in this position


White can not improve his position. The black king is far enough away from his own pieces, so there is no risk to get exploded.

BUT, take a look on the next position

kein Remis
no draw

Whats up here? If you let Atomystica run with the rule implemented the program brings up a 0.00 evaluation. But if you remove the rule a checkmate in seven comes up! Here we have the exception. The bishop position is really bad here. Black can not avoid the rook sacrifice on a7 followed by the queen promotion of the a-pawn.

Piece values

Here some numbers about the piece values Atomystica uses. The program differs between Opening, Middlegame and Endgame.
The values are in centipawns.

Queen    575
Bishop   60
Knight   75
Rook     250
Pawn     100


Queen    575
Bishop   50
Knight   50
Rook     250
Pawn     100


Queen    675
Bishop   50
Knight   50
Rook     250
Pawn     100

These are only the basic material values. In some endgames a rook gets a massiv malus to avoid senseless pawn sacrifices because a rook as more then two pawns more worth then a bishop in example. On the other hand a passed pawn can earn a lot of points by just stepping up the board.

As you can see the values of the minor pieces are very low. But we know, the wrong bishop is sometimes totally worthless, so thats the idea behind that. The base for that is also a document i linked in this “Atomic in science” blog entry.

The program, a slave – easy for a human

Currently I am searching for errors in the opening books Atomystica uses. And i found some of them 🙁

Here is a nice example after 1. Nf3 f6 2. d4 Nh6 3. Ne5 Ng4??

Weiß am Zug
White to move

Of course Atomystica knows 4. Nxd7 and atomic checkmate. But the book offers a second option! But this line is far from good, but was used in one game with hundred percent success. This line is 4. Nxg4?! d5 5. e4 Na6 6. Nc3 Bg4 . Although the better player won the game. The bad thing for Atomystica is that the program knews it better, but has to follow the dictate of his programmer! And in this case the programmer just throws a coin to choose the line Atomy has to use.

It is interesting that we find the final position of the line two more times in the book
1. Nf3 f6 2. Nd4 Nh6 3. Nf5 Nxf5 4. d4 d5 5. e4 Na6 6. Nc3 Bg4
and with an other moveorder
1. Nf3 f6 2. Nd4 Nh6 3. Nf5 Nxf5 4. e4 d5 5. d4 Na6 6. Nc3 Bg4
here White has no checkmate in move four, so these to lines are also bad.

Opening traps

A atomic game ends mostly with a tactic.
Eine Atomicpartie ist meist taktisch geprägt. Through the possibility to blow up a bunch of pieces in only one move a good knowledge of opening tricks is essential.

Here black wins with a pretty little tactical counter.

After 1. Nc3 e6 2. Nf3 Qf6 3. Ne4

nach 3. Ne4
after 3. Ne4

Black played 4…Nh6! and won after 4. Neg5?? (better is 4. g4) 4…Kd8 and White resigned.

nach 4...Kd8!
after 4…Kd8!

Whats the magic here?


a) 6. Nxf7 fails to 6…Ng4 and atomicmate on f2.
So better kill the Black Knight
b) 6. Nxh7 together with an additional rook. Because the black Queen can’t go to f2. But then the black Bishop becomes the hero. 6…Bd6!! 7. d4 the King needs a place to escape after 7…Bg3!! 8. Kd2. But the Bishop stays insistent 8…Bf4+ 9. Ke1 back in jail 9…Bd2!! and Checkmate.

und Matt.
and Checkmate.

the wrong bishop – the ever buggy code

In some positions a extra bsihop is worth nothing if it has the wrong color. I already wrote about that fact:

The method is simple, maybe…for a human…
Chess knowledge …. simple for a human

And I thought Atomysticas detection algorithm works, but it did not.

The following position Atomystica detected as a draw

Schwarz am Zug
Black to move

Atomystica calculated the move order 16…Rf2 17. Rxf2 h5 18. Nxc7 Bxc3 and a draw because of the wrong Bishop rule. All the black pawns are indeed on the “right” color, but White can create a unstoppable passed pawn. And so it happened

Weiß am Zug
White to move

White moved 19. Bd6 and Atomystica suddenly recognized the problem, but it was to late.

The problem, if you put the a-pawn onto the e-file the position is really a draw.

Weiß am Zug
White to move

Here White can not create a passed pawn.