Fruit played hg3 ep. A message appeared saying something like: "Illegal capture rejected by first program". Did anyone noticed (working or not) e.p. capture by Thinker?
There is a bug in the move validation code. Most EP captures are validated correctly, except the ones that are like what you demonstrated. This is when the code thinks that the capturing pawn is pinned, when in fact is is not - the King and the potential attacker are on adjacent files, but the code thinks they are on the same file (that's the bug).
Sorry about that. I have ran thousands of games, and never hit this one.
I will prepare a fix and have it out in the next few days.
Thanks for the update Lance, and Werner thx for the help.
Lance, are you planning to release a SMP-version?
Best wishes,
André
I have had an SMP version for a long time now. I'm just now happy with its performance, so, I have not released it yet. It has been improving over time - I might actually have something out before the end of this year.
Fruit played hg3 ep. A message appeared saying something like: "Illegal capture rejected by first program". Did anyone noticed (working or not) e.p. capture by Thinker?
There is a bug in the move validation code. Most EP captures are validated correctly, except the ones that are like what you demonstrated. This is when the code thinks that the capturing pawn is pinned, when in fact is is not - the King and the potential attacker are on adjacent files, but the code thinks they are on the same file (that's the bug).
Sorry about that. I have ran thousands of games, and never hit this one.
I will prepare a fix and have it out in the next few days.
Well if that happens one game per 10K, I can live without... On the other hand I would immensely appreciate if your engine could output its PV in the usual way (i.e. so that we can analyze with it). Your engine plays in a somewhat unusual way for a program; far more speculative and aware of its attacking chances involving long-term sacrifices and attacking builds up! I've seen more than once Thinker 5.1 being reffered as the "Michael Tal" of computer chess, both on this forum and on the rybkachess.com forum. I understand that you may have others priorities like compatibility with hand held devices or with multi-core, and thank you for the new release, it is always appreciated!
Inside Arena you see the main line of Thinker after the move. The 2nd move is normally the ponder move.
When watching Thinker I often are wondering which move Thinker awaits:
it looks not like the best opponent move, it looks like random. With such ponder-moves Thinker has no advantage from pondering - its always the wrong move I see here.
I would like to see a comment from Lance to this main-line after the move.
I have made an example. In the following game I noted the 2nd move displayed from Thinker an manually wrote it down as comment in the notation (if possible). E.g. have a look at move 11. h4 pondering e8g8?
[D]r3k2r/pppnnppp/8/2P1p1q1/4P2P/2N2B2/PP3PP1/R2QK2R b KQkq h3 0 11
Werner wrote:Inside Arena you see the main line of Thinker after the move. The 2nd move is normally the ponder move.
When watching Thinker I often are wondering which move Thinker awaits:
it looks not like the best opponent move, it looks like random. With such ponder-moves Thinker has no advantage from pondering - its always the wrong move I see here.
I would like to see a comment from Lance to this main-line after the move.
I have made an example. In the following game I noted the 2nd move displayed from Thinker an manually wrote it down as comment in the notation (if possible). E.g. have a look at move 11. h4 pondering e8g8?
[D]r3k2r/pppnnppp/8/2P1p1q1/4P2P/2N2B2/PP3PP1/R2QK2R b KQkq h3 0 11
The PV output in Thinker is never accurate. This is because the PV output is simply from iterating though the hash table (which just gets overwritten with other moves all the time).
Werner wrote:Inside Arena you see the main line of Thinker after the move. The 2nd move is normally the ponder move.
When watching Thinker I often are wondering which move Thinker awaits:
it looks not like the best opponent move, it looks like random. With such ponder-moves Thinker has no advantage from pondering - its always the wrong move I see here.
I would like to see a comment from Lance to this main-line after the move.
I have made an example. In the following game I noted the 2nd move displayed from Thinker an manually wrote it down as comment in the notation (if possible). E.g. have a look at move 11. h4 pondering e8g8?
[D]r3k2r/pppnnppp/8/2P1p1q1/4P2P/2N2B2/PP3PP1/R2QK2R b KQkq h3 0 11
The PV output in Thinker is never accurate. This is because the PV output is simply from iterating though the hash table (which just gets overwritten with other moves all the time).
If the hash is even not accurate enough to give a move to ponder
then it may be better not to use hash in thinker because the hash can confuse thinker by telling it wrong moves to search first.
It also means that if you only fix that bug of bad order of moves thinker may be significantly stronger.
I do not use hash tables to get move to ponder but even in programs that use hash tables for this purpose usually the hash is at least accurate enough to give correct move to ponder because they do not use replace always method and do not replace hash entries with big depths that are more important.
The PV output in Thinker I think is just a hack to simply support that feature. Maybe what it contains is the last line the engine searched that was saved to the hash table. It has no value other than just pure random.
Lance could add PV support correctly in less than 5 minutes but its his choice not to, and I respect that.
