Mates that are difficult for engines

Here is a mate in 7

white to move and mate in 7:

Another mate in 7:

white to move and mate in 7:

Here's a double underpromotion that saves the draw for white:

Lovely compact composition by Yochanan Afek. My own next endgame study will be published by him in the Problemist. Also one with an underpromotion but I can't show it yet!

White to mate in 83 moves, can engines like crystal figure this one out?

I see how this one works, the first move is to take the Knight and then work your Queen along the h1-a8 diagonal, if the Bishop on a3 moves you check on the a1-a8 file and then get back on the h1-a8 diagonal until Black runs out of pawn moves or they are captured along the diagonal and then the Queen makes it's way over to c3 to capture the Rook on b2 for the mate. I'll let my computer think about it overnight.

My computer didn't find the mate, it was stuck on playing 1.Qxa3 for the draw.

Here's my solution, but mine looks like a mate in 79, but of course there may be better moves on either side to shorten or lengthen the mate.

This position is not a mate that is difficult for engines. Here is Rybka 2.3.2a mating the stronger rated SF15.1 with NNUE at 40 moves in two hours (repeating) each with the same resources. My guess is it would mate against any Black moves.

It only has trouble with finding a mate in 6, but that is not the purpose of the engine. It's designed to win games against what currently passes as strong competition. If it examined the whole game tree from a given position to the depth of any mate it finds, it wouldn't be able to do that. It would, in that case, be able to find the shortest mate in this position, but that's not the point. Mate in 15 scores the point just as mate in 6 does.

(But I was quite amazed that SF didn't find the mate in 6 anyway.)

But that is not the point of this thread, or the problem. The title is "mates that are difficult for engines", and the mate is in a specific amount of moves, otherwise it would just be specified "white to play and win", or white to play and find any mate".

Yes of course.

I was just pointing out that you shouldn't expect mate in n problems to be easy for engines because they're not what engines are designed for. (Many of the engines are very good at what they are designed for.)

So if the subject of the topic is mainly mates in n it would be rather akin to the topic "Violin concertos that are difficult for Lewis Hamilton".

That is correct, but I think you accidentally posted on the wrong thread.

Edit, actually it is me who is mixing up the threads here, I'm at work so I on my idiotic phone looking at this tiny screen.

The truth is that the vast majority of mate in x problems ARE easy for engines, it was a lot of work for me to finds ones in my databases that are not easy. This thread is for fun and is NOT concerned if engines were or were not designed for mate in x problems.

Probably true, so long as we are talking about composed mate in x problems.

Some very nice problems too, so carry on posting. (Many violin concertos that Lewis Hamilton might find difficult still make interesting listening.)

Here is a mate in 86 (or less) that is in it's simplest form (Kings and Pawns) that illustrates the theme of triagulation and Zugzwang rather well:

kinda annoying ngl to have to randomly choose between Kb1 Ka1 or Ka1 Kb1 in triangulation

Was promoting to a knight the actual fastest win there? LOL!

i think yes

to cover a6 and prevent escape

You of all people should have appreciated the necessary underpromotion for the shortest mate 😁

imo that is appreciation

I'm afraid that happens a lot in triangulation problems, it is unavoidable.