An opening book of 111 million computer analyzed positions?

[Dann Corbit]

We should add the name Les Fernandez to the credits for this data. He is my data mining partner in crime.

[Rebel]

Consider it done.

[Rebel]

Is it an actual opening book or simply a set of analysed positions?

It's both.

Just study the left screenshot for a while, it's a rich variation of all.

[Rebel]

You can use the opening book with ProDeo 2.1, in fact I am going to use it tomorrow in the remaining rounds of the YAT-2 tournament.

Will be nice to view it in action!

Got 2½ points more against Andsacs than in YAT-1

Even better, it won the match against Tornado rated 127 elo higher.

Also against Gaviota (rated 180 elo higher) ProDeo lost with a small margin, 14-16. In YAT-1 it lost to Gaviota with 6.5-23.5

Statistic of the Dann Corbit book:

Code: Select all

Andscacs    94 = 3.1 extra book moves after the normal book
Gaviota    153 = 5.1 extra book moves after the normal book
Tornado    100 = 3.3 extra book moves after the normal book

I am going to replay the first 5 rounds to see if these super results are consistent.

[MikeB]

Can it be true?

Sure, see:

http://rebel13.nl/misc/ebf/dann%20corbit.html

Dann Corbit is the main contributor of the EBF project. When he saw the potential of the project he created a selection of his huge collection of analyzed positions and donated 111 million deep analyzed (20+ plies) positions in EPD format including best move, score and depth.

There are no typos in the above, Dann since decades analyzes positions with top-engines, about the contributed 111 million he said: You got the tip of the iceberg, but it should be the premium ice.

From his words I understood he must have more than half a billion analyzed positions, a life time work.

So a big thanks Dann for making the top of your (life time) work available for public consumption.

You can use the opening book with ProDeo 2.1, in fact I am going to use it tomorrow in the remaining rounds of the YAT-2 tournament.

Interesting ...here's a position from the 111 million that has some distinction...

[d]1bb2rk1/2qrnppp/2n1p3/1Q6/4N3/1B2BN2/1P3PPP/R2R2K1 w - -

I currently have a modified crafty crawling through the 111,239,727 positions to see which one has the highest perft 1 - so far it's the position above with 71 possible moves ... ( I appreciate that someone can craft a position with 218 possible moves , but I was curious to see how high it would get in a normal chess game...)

[zullil]

Interesting ...here's a position from the 111 million that has some distinction...

[d]1bb2rk1/2qrnppp/2n1p3/1Q6/4N3/1B2BN2/1P3PPP/R2R2K1 w - -

I currently have a modified crafty crawling through the 111,239,727 positions to see which one has the highest perft 1 - so far it's the position above with 71 possible moves ... ( I appreciate that someone can craft a position with 218 possible moves , but I was curious to see how high it would get in a normal chess game...)

Two questions come to mind immediately:

1) How far is "so far"?

2) Why are you doing this at all? (I suppose you answered this one in your post.)