Hi, here I found two 2nd moves that lead to a straight win, yet one of them has a better computer eval:
lichess.org/training/fhHt3
How is really provided solution any better?
lichess.org/training/fhHt3
How is really provided solution any better?
Did you analyse your other "solution"?
I think the puzzle has only one solution.
So what moves do you have in mind?
I think the puzzle has only one solution.
So what moves do you have in mind?
I followed e3 for 8-10 moves, still having approx. +1.3 eval at moderate depth. f4 is more straightforward, maybe is this really a case of "best move" instead of "the only good move"?
When you evaluate e3 it shows no advantage. As you said you followed some moves when the "advantage" suddenly showed up. And even then it's a tiny and uncertain advantage. So this is a legit puzzle (whether it's nice or instructive is a different matter)
@OctoPinky said in #3:
> I followed e3 for 8-10 moves, still having approx. +1.3 eval at moderate depth
After 2. e3
SF 16
single PV, depth 100, 12B nodes, +0.00
multi PV (3 lines), depth 96, 53B nodes, +0.00
Crystal 5
Single PV, depth 83, 41B nodes, +0.00
Multi PV, depth 80, 36B nodes, +0.00
Leela
BT2 768x15 3250000, depth 46, 58M nodes, +0.07
> I followed e3 for 8-10 moves, still having approx. +1.3 eval at moderate depth
After 2. e3
SF 16
single PV, depth 100, 12B nodes, +0.00
multi PV (3 lines), depth 96, 53B nodes, +0.00
Crystal 5
Single PV, depth 83, 41B nodes, +0.00
Multi PV, depth 80, 36B nodes, +0.00
Leela
BT2 768x15 3250000, depth 46, 58M nodes, +0.07
@corvusmellori said in #5:
> SF 16
> single PV, depth 100, 12B nodes, +0.00
Thanks for the work, no way I could reach depth 100!
> SF 16
> single PV, depth 100, 12B nodes, +0.00
Thanks for the work, no way I could reach depth 100!
@OctoPinky said in #6:
> no way I could reach depth 100
Simple positions like this result in very high nodes per second and very fast deepening. On my desktop, with the SF AVX2 binary, I hit over 100Mnps on this position, easily 6x or more compared to the starting position.
That being said, that many nodes are definitely not required. On SF16, single PV:
Depth 20: +0.36 (0.03 seconds)
Depth 40: +0.04 (0.51 seconds)
Depth 47: +0.00 (0.90 seconds)
Evaluation remains 0.00 for all depths 47 and higher.
> no way I could reach depth 100
Simple positions like this result in very high nodes per second and very fast deepening. On my desktop, with the SF AVX2 binary, I hit over 100Mnps on this position, easily 6x or more compared to the starting position.
That being said, that many nodes are definitely not required. On SF16, single PV:
Depth 20: +0.36 (0.03 seconds)
Depth 40: +0.04 (0.51 seconds)
Depth 47: +0.00 (0.90 seconds)
Evaluation remains 0.00 for all depths 47 and higher.
@Cedur216 said in #4:
> So this is a legit puzzle
Well, every puzzle is legit! Even if there was no clear solution (and there is, I see now) the aim is to practice or learn something, which is a job for the puzzle solver.
> So this is a legit puzzle
Well, every puzzle is legit! Even if there was no clear solution (and there is, I see now) the aim is to practice or learn something, which is a job for the puzzle solver.
I was using my phone SF11 HCE or SF16 NNUE at 24 and getting around +1.3 all the way, but finally reached a position where no further progress seems possible. Still having slightly positive eval (?).
+1.3 isn't really a "straight win."