For Marco---possible Stockfish bug

[lech]

By making to SF to solve one specific position, one can justify almost any change. What we need is a logical reasoning: why do you think that algorithmically it's better to allow alpha := beta. Give one (hypothetic) example situation where this could make positive difference.

Becouse there is only one VALUE_DRAW, and two sides.

So? This shouldn't be any kind of problem for negamax. Please give more details.

VALUE_DRAW should be alpha for both sides. In other case a batter move can't appear. It happens with move repeition.

[zamar]

By making to SF to solve one specific position, one can justify almost any change. What we need is a logical reasoning: why do you think that algorithmically it's better to allow alpha := beta. Give one (hypothetic) example situation where this could make positive difference.

Becouse there is only one VALUE_DRAW, and two sides.

So? This shouldn't be any kind of problem for negamax. Please give more details.

VALUE_DRAW should be alpha for both sides. In other case a batter move can't appear. It happens with move repeition.

Sorry, but that's complete nonsense. Please, study the details of Minimax and Negamax.

[lech]

By making to SF to solve one specific position, one can justify almost any change. What we need is a logical reasoning: why do you think that algorithmically it's better to allow alpha := beta. Give one (hypothetic) example situation where this could make positive difference.

Becouse there is only one VALUE_DRAW, and two sides.

So? This shouldn't be any kind of problem for negamax. Please give more details.

VALUE_DRAW should be alpha for both sides. In other case a batter move can't appear. It happens with move repeition.

Sorry, but that's complete nonsense. Please, study the details of Minimax and Negamax.

Why study it? You wrote ".. unpredictable things can happen."
"It is better to have a good feeling or be an astronomer.

[zamar]

Why study it?

Because what you write shows that you don't understand it!

You wrote ".. unpredictable things can happen."

If you break basic negamax, indeed unpredictable things can happen in SMP-mode. Anything can happen if you call search with alpha == beta.
Or some future change could trigger this bug.

"It is better to have a good feeling or be an astronomer.

?

[lech]

Why study it?

Because what you write shows that you don't understand it!

You wrote ".. unpredictable things can happen."

If you break basic negamax, indeed unpredictable things can happen in SMP-mode. Anything can happen if you call search with alpha == beta.
Or some future change could trigger this bug.

"It is better to have a good feeling or be an astronomer.

?

I feel like a boxer in the ring, I like people with a wide horizon in the thinking.
I think that you do not understand me.
I saw that If search returns VALUE_DRAW (by move repetition ?) it is very hard to change it by a better move .
I used alpha = beta to try break it. And I got a good effect. It was an interesting experience. I like it. I hope my knowladge is sufficient to have fun.

[Dave_N]

Why study it?

Because what you write shows that you don't understand it!

You wrote ".. unpredictable things can happen."

If you break basic negamax, indeed unpredictable things can happen in SMP-mode. Anything can happen if you call search with alpha == beta.
Or some future change could trigger this bug.

"It is better to have a good feeling or be an astronomer.

?

I feel like a boxer in the ring, I like people with a wide horizon in the thinking.
I think that you do not understand me.
I saw that If search returns VALUE_DRAW (by move repetition ?) it is very hard to change it by a better move .
I used alpha = beta to try break it. And I got a good effect. It was an interesting experience. I like it. I hope my knowladge is sufficient to have fun.

maybe you could have a statement that detects repetition equality and use the (alpha == beta) code to try to find something better ...

if ( !repetition )
{
// do normal stockfish code
}
else
{
// try to search for a better move with alpha==beta
}

[rvida]

I feel like a boxer in the ring, I like people with a wide horizon in the thinking.

It is good to have a wide horizon. But in this case you are trying to cure the symptoms, not the cause. Please try to rethink how alpha-beta works instead of trial&error approach.

[mcostalba]

I feel like a boxer in the ring, I like people with a wide horizon in the thinking.

It is good to have a wide horizon. But in this case you are trying to cure the symptoms, not the cause. Please try to rethink how alpha-beta works instead of trial&error approach.

2 heavy weights aganst the poor boxer Marek: I see the knock out coming

Don't worry Marek, you fought with honor for almost one round !

[Robert]

Hi Marco

be ironic now can't help to solve the problem pointed .

If other engines solve the problem and Stockfish dont, _AND_ this is not a bug, where the algorithm is failing?


PS.: This is the question...

[lech]

I feel like a boxer in the ring, I like people with a wide horizon in the thinking.

It is good to have a wide horizon. But in this case you are trying to cure the symptoms, not the cause. Please try to rethink how alpha-beta works instead of trial&error approach.

2 heavy weights aganst the poor boxer Marek: I see the knock out coming

Don't worry Marek, you fought with honor for almost one round !

This is a good time to die as an old Indian said, and went to a brothel.
I still come back!