A Problem in analyzing chess games with Chess Explorer
Moderators: Watchman, Mark Uniacke, mrudolf
-
- Member
- Posts: 273
- Joined: Fri Jul 02, 2010 7:12 pm
A Problem in analyzing chess games with Chess Explorer
Hi
When analyzing a completed game with Chess Explorer (Deep Hiarcs 14).In every game when I select the option of analyzing only Non Book moves , the program does not respect this. Instead what it does is to only not analyze the first 5 moves and to ANALYZE its own BOOK moves . This has happened in many many games , so it is not a one time event. Of course the Deep Hiarcs 14 local Opening Book goes way past 5 moves in most cases. It would be nice if this issue was solved.
When analyzing a completed game with Chess Explorer (Deep Hiarcs 14).In every game when I select the option of analyzing only Non Book moves , the program does not respect this. Instead what it does is to only not analyze the first 5 moves and to ANALYZE its own BOOK moves . This has happened in many many games , so it is not a one time event. Of course the Deep Hiarcs 14 local Opening Book goes way past 5 moves in most cases. It would be nice if this issue was solved.
- Mark Uniacke
- Hiarcs Author
- Posts: 1459
- Joined: Sun Jul 29, 2007 1:32 pm
- Location: United Kingdom
- Contact:
This is a difficult one, HCE uses its ECO reference to decide what are established book moves. It does not use local opening books since just because a move has been played before does not mean it is firmly established.
If we were to avoid analysing all moves in the local book this would mean some rarely played moves would not be analysed.
The game analysis is performed in reverse order from the end of the game to the beginning so machine learning is most effective. This also means the game analysis can be stopped early by the user.
There are other options to control which moves in the game are analysed.
If we were to avoid analysing all moves in the local book this would mean some rarely played moves would not be analysed.
The game analysis is performed in reverse order from the end of the game to the beginning so machine learning is most effective. This also means the game analysis can be stopped early by the user.
There are other options to control which moves in the game are analysed.
-
- Member
- Posts: 273
- Joined: Fri Jul 02, 2010 7:12 pm
Mark :I sure this is not a bug?? Why is the number 5 a fixed number in hundreds of games I have analyzed so far??? How can the established book moves run out after 5 moves after 1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 d6 5.c3 Nf6 6.d3 g6 7.0-0 Bg7 8.Re1 0-0 9. Nbd2 b5 10.Bc2 Bb7 ????Mark Uniacke wrote:This is a difficult one, HCE uses its ECO reference to decide what are established book moves. It does not use local opening books since just because a move has been played before does not mean it is firmly established.
If we were to avoid analysing all moves in the local book this would mean some rarely played moves would not be analysed.
The game analysis is performed in reverse order from the end of the game to the beginning so machine learning is most effective. This also means the game analysis can be stopped early by the user.
There are other options to control which moves in the game are analysed.
- Mark Uniacke
- Hiarcs Author
- Posts: 1459
- Joined: Sun Jul 29, 2007 1:32 pm
- Location: United Kingdom
- Contact:
-
- Member
- Posts: 273
- Joined: Fri Jul 02, 2010 7:12 pm
-
- Member
- Posts: 273
- Joined: Fri Jul 02, 2010 7:12 pm
Mark :Mark Uniacke wrote:Just checked and it does seem to go to early in some established openings, we will investigate and resolve any issue for the 1.5 update.
May I ask why did updates 1.5 and 1.6 fail to solve this issue ?? Looks like you have acknowleged that it is in fact a problem.... This is very frustrating when analyzing with HCE because it wastes a huge amout of time analyzing its own opening book moves.............
- Mark Uniacke
- Hiarcs Author
- Posts: 1459
- Joined: Sun Jul 29, 2007 1:32 pm
- Location: United Kingdom
- Contact:
We did look into this but I forgot to report back the findings.
HCE uses its ECO reference to decide on where to stop analysing - as a result it will not normally stop at the last move played from the book. The reason we took this step is that many books moves can be overturned by analysis so we went back to our ECO reference points as a safe point. The user is able to adjust this using the move number ranges in analysis if necessary.
This is not optimal I agree but also stopping at the last book move is not good either as the last book move may be a bad move. The analysis is done from the end of the game to the beginning so HIARCS can use its learning to enhance the analysis, this means the last moves analysed are the early moves so that does allow game annotation to be stopped before these earlier moves are analysed.
We are looking at a better solution in a future HCE which takes account of established book lines which are not likely to change, but at present that is what we do.
HCE uses its ECO reference to decide on where to stop analysing - as a result it will not normally stop at the last move played from the book. The reason we took this step is that many books moves can be overturned by analysis so we went back to our ECO reference points as a safe point. The user is able to adjust this using the move number ranges in analysis if necessary.
This is not optimal I agree but also stopping at the last book move is not good either as the last book move may be a bad move. The analysis is done from the end of the game to the beginning so HIARCS can use its learning to enhance the analysis, this means the last moves analysed are the early moves so that does allow game annotation to be stopped before these earlier moves are analysed.
We are looking at a better solution in a future HCE which takes account of established book lines which are not likely to change, but at present that is what we do.