Two Knights vs Pawn (Advanced)

MaximeAndCie

Hi everyone,

Recently, I've been struggling with the Two Knights vs Pawn (Advanced) "Checkmates" drill from chess.com.

In this position, White has the initiative and is supposed to push Black's king in the h8 corner in order for the h3 Knight to help deliver checkmate in time. Whatever I do with King+Knight, Black's king seems to always slip away and avoid being trapped. Does anybody know a systematic way to succeed at this drill ? Or the type of position I am supposed to reach before involving the second Knight ?

Furthermore, does anybody have a good ressource to learn some theory about two-knights endgames and the Troitsky lines ? Wikipedia is too vague about this.

Thanks in advance !!! happy.png

MARattigan

Collection of Chess Studies A.A.Troitzky, Ishi Press. Try Amazon.

Final section is two knights v pawn.

First get perfect on KNK game and (relevant in the position you give) KNKP game.

KNNKQ is also a prerequisite for the ending in a small percentage of positions. but not  that particular diagram.  That endgame is harder. I don't know a comprehensive coverage; practice against an EGTB is probably best. 

MaximeAndCie

Thank you for your answer ! Found the book on Amazon, I'll consider buying it. 

When you say "get perfect on KNK game", is there a way to practice it with some guidelines ? Is it in the book ?

Finally, if anyone is interested, I found an EGTB that solves this position in 57 moves with perfect play from both sides. Enjoy !

MARattigan

Chapter II in the supplement of the above book covers KNK. You can really only practice it in a KNNKP game (and it can occasionally come in useful in KBNK), because any computer program will just say "dead position" and you're unlikely to find a human opponent who's very keen. But it's the whole basis of the KNNKP ending. You need to have it automatic. If you've covered Two Knights vs Pawn (not Advanced) or whatever you should already be familiar with it.

The exact position you show will occur frequently when you get into practicing the endings with a pawn on the rook's file, so it's a good example to study.

 

I have a program called Wilhelm and the 3-4-5 man Nalimov EGTBs. Wilhelm will generate a list of endgame positions for KNNKP (or any) that are mate in anywhere between m and n. So you can start with low m and n and analyse/practice those positions, then increase m and n.  You can also do things like limiting the pawn position to a particular file or blocked by a knight on a particular square. It will probably take a long time to cover the ending and you will probably never get into it in a chess game, so I assume you're doing it out of interest rather than to improve your chess results.

You can adopt the same approach with KNNKQ.

Wilhelm used to be available here https://web.archive.org/web/20081208015339/http://www.geocities.com/rba_schach2000/overview_english.htm. It tends to move around so you might need to track it. The Nalimov EGTBs should be easy to locate.  You should be able to download both without forking out.

By the way the Nalimov EGTBs don't take any account of the 50 move rule. Only the Sysygy/Lomonosov EGTBs do that. But I think trying to take the 50 move rule into account is impossibly difficult for mortals, especially if you're also attempting mate in the shortest number of moves.

 

Numquam

I am curious how white is supposed to push the black king in the h8 corner. I don't see a way to force the king to walk to the other side of the board and Syzygy doesn't suggest something like that.

 

MARattigan

@Numquam

then continue from after move 58 in Syzygy v Martin, post #67 here (but reversing flanks)

https://www.chess.com/forum/view/endgames/is-checkmate-with-2-knights-possible?page=4

I think Syzygy may just fail to solve it under the cursed rule. (But doesn't it give you a game whether or not? And anyway I don't think it should fail under the cursed rule from there - I don't think it should get to 50 moves all told. In fact in your example it's playing traditionally accurate moves as should be expected. You haven't taken it far enough.)

You can probably fill in the gaps. Black should finish up mated on h8, but if he's not accurate it could also be a1 or h1. It's the threats of those that push him to h8.

Numquam

@MARattigan, thanks. That is pretty clever to force black to go for the position with the knight on h2. Now the black king can no longer stay in the corner, because the knight can cover a1 in 3 moves.

 

MARattigan
Numquam wrote:

@MARattigan, thanks. That is pretty clever to force black to go for the position with the knight on h2. Now the black king can no longer stay in the corner, because the knight can cover a1 in 3 moves.

 

Not my idea, Troitsky came up with it first. But yes.

silverseal

There is a youtube video about this which explains this type of endgame in detail (just search two knights vs a pawn (or h pawn)). This is also the key position where almost all initial positions where the pawn is securely blockaded at h4 will go through with perfect play (or similar positions). It's actually much easier than many other positions. I also recommend practicing with database.

The only way to make progress is to force pawn push to h3, where the passive knight is better placed and checkmate with one knight is possible (so you don't need to directly defend that knight). This also makes h1 a mating corner, and you need to force the defending king to choose which corner to go.

syedkurirul

this game is draw

Gomer_Pyle

For endgames like this I like to plug the position into an endgame database and see how it's supposed to be played. Then it just takes repetition until I remember how. (It's not a draw)

https://www.shredderchess.com/online-chess/online-databases/endgame-database.html