Loading ...
Sorry, an error occurred while loading the content.

Anyone want to do Millenium Chess for Axiom?

Expand Messages
  • Rich Hutnik
    Apparently the inability to code AI values causes the Zillions adaptation of Millenium Chess to not work right. I was curious if anyone wanted to get ahold of
    Message 1 of 26 , Dec 1, 2007
    View Source
    • 0 Attachment
      Apparently the inability to code AI values causes the Zillions
      adaptation of Millenium Chess to not work right. I was curious if
      anyone wanted to get ahold of the Zillions file, and adapt it to
      Axiom, so that the AI works right. Doing this would show the ability
      to do a chess game in Axiom, which would be a plus.

      On this note, I believe doing games like Millenium chess, checkers and
      other classic games will show people what can be done with Axiom.

      So, can anyone help out?
    • Ed van Zon
      ... and ... How is it not working right? As for implementing it in Axiom, that would be a big challenge. Axiom isn t geared towards chess-like games.
      Message 2 of 26 , Dec 2, 2007
      View Source
      • 0 Attachment
        --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@...>
        wrote:
        >
        > Apparently the inability to code AI values causes the Zillions
        > adaptation of Millenium Chess to not work right. I was curious if
        > anyone wanted to get ahold of the Zillions file, and adapt it to
        > Axiom, so that the AI works right. Doing this would show the ability
        > to do a chess game in Axiom, which would be a plus.
        >
        > On this note, I believe doing games like Millenium chess, checkers
        and
        > other classic games will show people what can be done with Axiom.
        >
        > So, can anyone help out?

        How is it not working right?

        As for implementing it in Axiom, that would be a big challenge. Axiom
        isn't geared towards chess-like games. Everything is possible, of
        course, but I'll pass.

        cheers, Ed
      • Greg Schmidt
        I m in basic agreement with Ed s response, it s doable but would require considerable effort. On the other hand, I expect it would be considerably easier than
        Message 3 of 26 , Dec 2, 2007
        View Source
        • 0 Attachment
          I'm in basic agreement with Ed's response, it's doable but would
          require considerable effort. On the other hand, I expect it would be
          considerably easier than writing a chess program from scratch (I
          admit, that may not be an overly meaningful assessment).

          I'm a little surprised that Zillions doesn't play this game well. If
          the strength issue is limited to the piece values, then I believe the
          values can be altered via "tweaking" (see the posts
          regarding "tweaking" in the ZoG) newsgroup. I now Mats (and others)
          have used this technique with good success.

          BTW, I took a quick look at the zrf and didn't see anything preventing
          the king from moving into check. The king uses a macro name "leap"
          which mainly checks to see if the target square is "not-freend?". Did
          I miss something?

          -- Greg
        • Rich Hutnik
          ... Ed in the version of Mill. chess I did, with two captures of kings, Zillion actually plays the game. In the version implemented with a capture and
          Message 4 of 26 , Dec 2, 2007
          View Source
          • 0 Attachment
            --- In axiom-system@yahoogroups.com, "Ed van Zon" <yahoo@...> wrote:
            >
            > --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@>
            > wrote:
            > >
            > > Apparently the inability to code AI values causes the Zillions
            > > adaptation of Millenium Chess to not work right. I was curious if
            > > anyone wanted to get ahold of the Zillions file, and adapt it to
            > > Axiom, so that the AI works right. Doing this would show the ability
            > > to do a chess game in Axiom, which would be a plus.
            > >
            > > On this note, I believe doing games like Millenium chess, checkers
            > and
            > > other classic games will show people what can be done with Axiom.
            > >
            > > So, can anyone help out?
            >
            > How is it not working right?
            >
            > As for implementing it in Axiom, that would be a big challenge. Axiom
            > isn't geared towards chess-like games. Everything is possible, of
            > course, but I'll pass.
            >
            > cheers, Ed

            Ed in the version of Mill. chess I did, with two captures of kings,
            Zillion actually plays the game. In the version implemented with a
            capture and checkmate, Zillions ends up moving a bunch of Bishops
            around, pushing no pawns and ends up in a stalemate (when the AI plays
            itself). Something is wrong here. I am told it isn't the code, it is
            how Zillions evaluates the pieces.
          • Mats Winther
            ... ... ability ... checkers ... Axiom. ... Axiom ... plays ... In all my chess variants I change the evaluation of pieces and I tweak
            Message 5 of 26 , Dec 2, 2007
            View Source
            • 0 Attachment
              --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@.
              ..> wrote:
              >
              > --- In axiom-system@yahoogroups.com, "Ed van Zon" <yahoo@>
              wrote:
              > >
              > > --- In axiom-system@yahoogroups.com, "Rich Hutnik"
              <richardhutnik@>
              > > wrote:
              > > >
              > > > Apparently the inability to code AI values causes the Zillions
              > > > adaptation of Millenium Chess to not work right. I was curious if
              > > > anyone wanted to get ahold of the Zillions file, and adapt it to
              > > > Axiom, so that the AI works right. Doing this would show the
              ability
              > > > to do a chess game in Axiom, which would be a plus.
              > > >
              > > > On this note, I believe doing games like Millenium chess,
              checkers
              > > and
              > > > other classic games will show people what can be done with
              Axiom.
              > > >
              > > > So, can anyone help out?
              > >
              > > How is it not working right?
              > >
              > > As for implementing it in Axiom, that would be a big challenge.
              Axiom
              > > isn't geared towards chess-like games. Everything is possible, of
              > > course, but I'll pass.
              > >
              > > cheers, Ed
              >
              > Ed in the version of Mill. chess I did, with two captures of kings,
              > Zillion actually plays the game. In the version implemented with a
              > capture and checkmate, Zillions ends up moving a bunch of Bishops
              > around, pushing no pawns and ends up in a stalemate (when the AI
              plays
              > itself). Something is wrong here. I am told it isn't the code, it is
              > how Zillions evaluates the pieces.
              >

              In all my chess variants I change the evaluation of pieces and I tweak
              Zillions to move the pawns. One can add dummy-pieces as reward, or
              one can position pawns on the second rank that have less value than a
              moved pawn (that is, the weaker pawn is promoted). I also encourage
              castling, discourage queen moves, and discourage king moves to the
              corner squares. This does wonders to my chess variants.

              By the way, since Axiom is not chess friendly, I wonder whether it's
              possible to use the c2forth program to convert algorithms from any of
              the open source chess programs, like "Gambit Fruit". But that's an
              ambitious idea.

              Mats
            • Keith Carter
              I have tweaked the pre-king values from a little under 1,200 to the range of 127,286 vs a King value in the range of 111472. I believe that 16,000 gap will
              Message 6 of 26 , Dec 2, 2007
              View Source
              • 0 Attachment
                I have tweaked the pre-king values from a little under 1,200 to the
                range of 127,286 vs a King value in the range of 111472. I believe
                that 16,000 gap will mean that the AI will let a pre-King get
                captured to avoid the capture of a rook or queen but not a knight or
                bishop.

                I added an irregular move type that the pre-King will never get to
                use because of move priorities. To adjust the value of the pre-king
                up add a few more "add"s to the irregular move description. To lower
                it delete a few.

                --Keith
                --- In axiom-system@yahoogroups.com, "Rich Hutnik"
                <richardhutnik@...> wrote:
                >
                > --- In axiom-system@yahoogroups.com, "Ed van Zon" <yahoo@> wrote:
                > >
                > > --- In axiom-system@yahoogroups.com, "Rich Hutnik"
                <richardhutnik@>
                > > wrote:
                > > >
                > > > Apparently the inability to code AI values causes the Zillions
                > > > adaptation of Millenium Chess to not work right. I was curious
                if
                > > > anyone wanted to get ahold of the Zillions file, and adapt it to
                > > > Axiom, so that the AI works right. Doing this would show the
                ability
                > > > to do a chess game in Axiom, which would be a plus.
                > > >
                > > > On this note, I believe doing games like Millenium chess,
                checkers
                > > and
                > > > other classic games will show people what can be done with
                Axiom.
                > > >
                > > > So, can anyone help out?
                > >
                > > How is it not working right?
                > >
                > > As for implementing it in Axiom, that would be a big challenge.
                Axiom
                > > isn't geared towards chess-like games. Everything is possible, of
                > > course, but I'll pass.
                > >
                > > cheers, Ed
                >
                > Ed in the version of Mill. chess I did, with two captures of kings,
                > Zillion actually plays the game. In the version implemented with a
                > capture and checkmate, Zillions ends up moving a bunch of Bishops
                > around, pushing no pawns and ends up in a stalemate (when the AI
                plays
                > itself). Something is wrong here. I am told it isn't the code, it
                is
                > how Zillions evaluates the pieces.
                >
              • Rich Hutnik
                ... Ok, thanks for the help. Can someone get this latest version on the Zillions site?
                Message 7 of 26 , Dec 2, 2007
                View Source
                • 0 Attachment
                  --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                  >
                  > I have tweaked the pre-king values from a little under 1,200 to the
                  > range of 127,286 vs a King value in the range of 111472. I believe
                  > that 16,000 gap will mean that the AI will let a pre-King get
                  > captured to avoid the capture of a rook or queen but not a knight or
                  > bishop.
                  >
                  > I added an irregular move type that the pre-King will never get to
                  > use because of move priorities. To adjust the value of the pre-king
                  > up add a few more "add"s to the irregular move description. To lower
                  > it delete a few.
                  >
                  > --Keith

                  Ok, thanks for the help. Can someone get this latest version on the
                  Zillions site?
                • Greg Schmidt
                  ... convert algorithms from any of the open source chess programs, ... I suspect a rewrite might be easier. One reason being that each chess program has it s
                  Message 8 of 26 , Dec 2, 2007
                  View Source
                  • 0 Attachment
                    --- In axiom-system@yahoogroups.com, "Mats Winther" <mwi5@...> wrote:
                    > ... I wonder whether it's possible to use the c2forth program to
                    convert algorithms from any of the open source chess programs, ...

                    I suspect a rewrite might be easier. One reason being that each chess
                    program has it's own unique way of representing the pieces and the
                    board. Some use bitboards, some use explicit arrays etc., chess
                    programs typically rely on very customized and specialized techniques
                    in their implementation. The code would have to be understood, not
                    blindly translated. As I sometimes like to say, there's no substitute
                    for actually knowing what you're doing.

                    Adding a UCT engine would simplify the task of creating a Chess
                    program since an evaluation function would not have to be crafted. I
                    don't know how strong it would play chess though as I'm unable to find
                    any references on applying UCT to Chess. I will say that it is a
                    project that is of interest to me.

                    -- Greg
                  • Rich Hutnik
                    ... I would like to speak here and say, while Axiom isn t really suited for chess, I don t see any more development to Zillions being done here. As a result,
                    Message 9 of 26 , Dec 2, 2007
                    View Source
                    • 0 Attachment
                      --- In axiom-system@yahoogroups.com, "Greg Schmidt" <gschmidt958@...>
                      wrote:
                      >
                      > --- In axiom-system@yahoogroups.com, "Mats Winther" <mwi5@> wrote:
                      > > ... I wonder whether it's possible to use the c2forth program to
                      > convert algorithms from any of the open source chess programs, ...
                      >
                      > I suspect a rewrite might be easier. One reason being that each chess
                      > program has it's own unique way of representing the pieces and the
                      > board. Some use bitboards, some use explicit arrays etc., chess
                      > programs typically rely on very customized and specialized techniques
                      > in their implementation. The code would have to be understood, not
                      > blindly translated. As I sometimes like to say, there's no substitute
                      > for actually knowing what you're doing.
                      >
                      > Adding a UCT engine would simplify the task of creating a Chess
                      > program since an evaluation function would not have to be crafted. I
                      > don't know how strong it would play chess though as I'm unable to find
                      > any references on applying UCT to Chess. I will say that it is a
                      > project that is of interest to me.

                      I would like to speak here and say, while Axiom isn't really suited
                      for chess, I don't see any more development to Zillions being done
                      here. As a result, if Axiom wants to get positioned as an AI playing
                      engine for abstract strategy games, then chess variants will need to
                      be set up so they can work in Axiom.
                    • Keith Carter
                      I will submit it closer to the end of the week and I am sure Ed will post it as part of his weekly Zillions posts. In the meantime anyone who is interested
                      Message 10 of 26 , Dec 2, 2007
                      View Source
                      • 0 Attachment
                        I will submit it closer to the end of the week and I am sure Ed will
                        post it as part of his weekly Zillions posts. In the meantime anyone
                        who is interested can play what is uploaded here to see if it needs
                        any more refinements. For example, some of the things Mats suggested.

                        --Keith


                        --- In axiom-system@yahoogroups.com, "Rich Hutnik"
                        <richardhutnik@...> wrote:
                        >
                        > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@> wrote:
                        > >
                        > > I have tweaked the pre-king values from a little under 1,200 to
                        the
                        > > range of 127,286 vs a King value in the range of 111472. I believe
                        > > that 16,000 gap will mean that the AI will let a pre-King get
                        > > captured to avoid the capture of a rook or queen but not a knight
                        or
                        > > bishop.
                        > >
                        > > I added an irregular move type that the pre-King will never get to
                        > > use because of move priorities. To adjust the value of the pre-
                        king
                        > > up add a few more "add"s to the irregular move description. To
                        lower
                        > > it delete a few.
                        > >
                        > > --Keith
                        >
                        > Ok, thanks for the help. Can someone get this latest version on the
                        > Zillions site?
                        >
                      • Rich Hutnik
                        ... Thanks. I will snag it and see if Zillions actually moves pawns.
                        Message 11 of 26 , Dec 2, 2007
                        View Source
                        • 0 Attachment
                          --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                          >
                          > I will submit it closer to the end of the week and I am sure Ed will
                          > post it as part of his weekly Zillions posts. In the meantime anyone
                          > who is interested can play what is uploaded here to see if it needs
                          > any more refinements. For example, some of the things Mats suggested.
                          >
                          > --Keith

                          Thanks. I will snag it and see if Zillions actually moves pawns.
                        • Keith Carter
                          It will take more tweaking to get that. I only changed the value of the pre-King so the AI would want to protect it. The problem with the pawns is that a
                          Message 12 of 26 , Dec 2, 2007
                          View Source
                          • 0 Attachment
                            It will take more tweaking to get that. I only changed the value of
                            the pre-King so the AI would want to protect it. The problem with the
                            pawns is that a pawn that has not moved is worth a lot more than a pawn
                            that has. A pawn that moves goes from about 3000 to about 1800.

                            --Keith

                            --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@...>
                            wrote:
                            >
                            > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@> wrote:
                            > >
                            > > I will submit it closer to the end of the week and I am sure Ed
                            will
                            > > post it as part of his weekly Zillions posts. In the meantime
                            anyone
                            > > who is interested can play what is uploaded here to see if it needs
                            > > any more refinements. For example, some of the things Mats
                            suggested.
                            > >
                            > > --Keith
                            >
                            > Thanks. I will snag it and see if Zillions actually moves pawns.
                            >
                          • Rich Hutnik
                            ... Ok, it is working now. Well, pretty much. Any way piece recycling is on? I don t recall that you could only promote a pawn to a piece that was
                            Message 13 of 26 , Dec 2, 2007
                            View Source
                            • 0 Attachment
                              --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                              >
                              > It will take more tweaking to get that. I only changed the value of
                              > the pre-King so the AI would want to protect it. The problem with the
                              > pawns is that a pawn that has not moved is worth a lot more than a pawn
                              > that has. A pawn that moves goes from about 3000 to about 1800.
                              >
                              > --Keith

                              Ok, it is working now. Well, pretty much. Any way piece recycling is
                              on? I don't recall that you could only promote a pawn to a piece that
                              was originally captured in the original Mill. Chess. Is there any
                              reason why that is so? I remark out promotion recycling and it is
                              working fine now.
                            • Keith Carter
                              I don t know about recycling. I only changed the code related to the pre-kings to bump their value. I did not look at or change any of the rest of the code.
                              Message 14 of 26 , Dec 2, 2007
                              View Source
                              • 0 Attachment
                                I don't know about recycling. I only changed the code related to the
                                pre-kings to bump their value. I did not look at or change any of
                                the rest of the code. The recycling status is as posted (whatever
                                that is).

                                --Keith

                                --- In axiom-system@yahoogroups.com, "Rich Hutnik"
                                <richardhutnik@...> wrote:
                                >
                                > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@> wrote:
                                > >
                                > > It will take more tweaking to get that. I only changed the value
                                of
                                > > the pre-King so the AI would want to protect it. The problem
                                with the
                                > > pawns is that a pawn that has not moved is worth a lot more than
                                a pawn
                                > > that has. A pawn that moves goes from about 3000 to about
                                1800.
                                > >
                                > > --Keith
                                >
                                > Ok, it is working now. Well, pretty much. Any way piece recycling
                                is
                                > on? I don't recall that you could only promote a pawn to a piece
                                that
                                > was originally captured in the original Mill. Chess. Is there any
                                > reason why that is so? I remark out promotion recycling and it is
                                > working fine now.
                                >
                              • Keith Carter
                                I have looked at the pawn code. The problem is that the pawn is two different pieces creating two different piece values. One has the initial move of one or
                                Message 15 of 26 , Dec 2, 2007
                                View Source
                                • 0 Attachment
                                  I have looked at the pawn code. The problem is that the pawn is two
                                  different pieces creating two different piece values. One has the
                                  initial move of one or two forward and the other has en-passant.
                                  This is really hampering the AI's approach to pawns. There should be
                                  one pawn description such as the way it is done in chess.zrf that
                                  comes with Zillions.

                                  --Keith


                                  --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                                  >
                                  > It will take more tweaking to get that. I only changed the value
                                  of
                                  > the pre-King so the AI would want to protect it. The problem with
                                  the
                                  > pawns is that a pawn that has not moved is worth a lot more than a
                                  pawn
                                  > that has. A pawn that moves goes from about 3000 to about 1800.
                                  >
                                  > --Keith
                                  >
                                  > --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@>
                                  > wrote:
                                  > >
                                  > > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@>
                                  wrote:
                                  > > >
                                  > > > I will submit it closer to the end of the week and I am sure Ed
                                  > will
                                  > > > post it as part of his weekly Zillions posts. In the meantime
                                  > anyone
                                  > > > who is interested can play what is uploaded here to see if it
                                  needs
                                  > > > any more refinements. For example, some of the things Mats
                                  > suggested.
                                  > > >
                                  > > > --Keith
                                  > >
                                  > > Thanks. I will snag it and see if Zillions actually moves pawns.
                                  > >
                                  >
                                • Rich Hutnik
                                  ... Ok, I will end up testing it more and then resubmit it. If it continues to work fine, I will go with this as the latest update. Thanks for the help
                                  Message 16 of 26 , Dec 2, 2007
                                  View Source
                                  • 0 Attachment
                                    --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                                    >
                                    > I don't know about recycling. I only changed the code related to the
                                    > pre-kings to bump their value. I did not look at or change any of
                                    > the rest of the code. The recycling status is as posted (whatever
                                    > that is).
                                    >
                                    > --Keith

                                    Ok, I will end up testing it more and then resubmit it. If it
                                    continues to work fine, I will go with this as the latest update.
                                    Thanks for the help everyone.
                                  • Alfred Pfeiffer
                                    Hello, I am the guy who did the actual MillenniumChess.zrf at the Zillions site. Thank you for the tweaking of the Pre-King pieces, but the piece nmKing
                                    Message 17 of 26 , Dec 2, 2007
                                    View Source
                                    • 0 Attachment
                                      Hello,

                                      I am the guy who did the actual MillenniumChess.zrf at the
                                      Zillions site. Thank you for the tweaking of the Pre-King
                                      pieces, but the piece 'nmKing' must not be tweaked.

                                      An question with your solution is what will happen in
                                      the case of a stalemate with the two Pre-Kings on board.
                                      I assume the engine will pass or does a null move
                                      (what would be illegal).

                                      --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                                      >
                                      > I have looked at the pawn code. The problem is that the pawn is two
                                      > different pieces creating two different piece values. One has the
                                      > initial move of one or two forward and the other has en-passant.
                                      > This is really hampering the AI's approach to pawns. There should be
                                      > one pawn description such as the way it is done in chess.zrf that
                                      > comes with Zillions.
                                      >
                                      > --Keith

                                      I made a new version with only one type of pawns.
                                      Thank you for the hint.

                                      I uploaded the new version as 'MillenniumChess2.zip'
                                      to the files section.


                                      Alfred Pfeiffer
                                    • Alfred Pfeiffer
                                      ... In the original MillenniumChess site this rule is described at ... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ...
                                      Message 18 of 26 , Dec 2, 2007
                                      View Source
                                      • 0 Attachment
                                        --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@...>
                                        wrote:
                                        > ...
                                        > Ok, it is working now. Well, pretty much. Any way piece recycling is
                                        > on? I don't recall that you could only promote a pawn to a piece that
                                        > was originally captured in the original Mill. Chess. Is there any
                                        > reason why that is so?

                                        In the original MillenniumChess site this rule is described at
                                        "http://www.vipchess.com/MillenniumGameRules.htm":

                                        | When pawns make it to the opposing back row...
                                        | only pieces already captured can be re-instated.
                                        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

                                        and also at "http://www.vipchess.com/millenniumFAQ'S.htm":

                                        | When Pawns Advance To The Opposing Back Row, Can I Reclaim
                                        | A Lost King?
                                        |
                                        | No! You may not reclaim lost Kings. Also, pawns do
                                        | not automatically become Queens.
                                        | You may only reclaim pieces that have been captured.
                                        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

                                        therefore I implemented it in the zrf.

                                        > I remark out promotion recycling and it is working fine now.
                                        >

                                        Alfred Pfeiffer
                                      • Keith Carter
                                        I believe a pre-King will always have a regular move so the irregular null move will never be reached. A stalemate with two pre-kings would be just that.
                                        Message 19 of 26 , Dec 2, 2007
                                        View Source
                                        • 0 Attachment
                                          I believe a pre-King will always have a regular move so the irregular
                                          null move will never be reached. A stalemate with two pre-kings
                                          would be just that. Zillions would move until a 3 position
                                          repetition happened and a draw was declared.

                                          Why must the nmKing piece not be tweaked? At its untweaked value it
                                          is worth less than a pawn and the AI will make no attempt to protect
                                          it.

                                          --Keith

                                          --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@...>
                                          wrote:
                                          >
                                          > Hello,
                                          >
                                          > I am the guy who did the actual MillenniumChess.zrf at the
                                          > Zillions site. Thank you for the tweaking of the Pre-King
                                          > pieces, but the piece 'nmKing' must not be tweaked.
                                          >
                                          > An question with your solution is what will happen in
                                          > the case of a stalemate with the two Pre-Kings on board.
                                          > I assume the engine will pass or does a null move
                                          > (what would be illegal).
                                          >
                                          > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@> wrote:
                                          > >
                                          > > I have looked at the pawn code. The problem is that the pawn is
                                          two
                                          > > different pieces creating two different piece values. One has
                                          the
                                          > > initial move of one or two forward and the other has en-passant.
                                          > > This is really hampering the AI's approach to pawns. There
                                          should be
                                          > > one pawn description such as the way it is done in chess.zrf that
                                          > > comes with Zillions.
                                          > >
                                          > > --Keith
                                          >
                                          > I made a new version with only one type of pawns.
                                          > Thank you for the hint.
                                          >
                                          > I uploaded the new version as 'MillenniumChess2.zip'
                                          > to the files section.
                                          >
                                          >
                                          > Alfred Pfeiffer
                                          >
                                        • Rich Hutnik
                                          ... Ok, silly me. I will need to change it back and send it to Ed. ARGH! So much for me presuming.
                                          Message 20 of 26 , Dec 2, 2007
                                          View Source
                                          • 0 Attachment
                                            --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@...> wrote:
                                            >
                                            > --- In axiom-system@yahoogroups.com, "Rich Hutnik" <richardhutnik@>
                                            > wrote:
                                            > > ...
                                            > > Ok, it is working now. Well, pretty much. Any way piece recycling is
                                            > > on? I don't recall that you could only promote a pawn to a piece that
                                            > > was originally captured in the original Mill. Chess. Is there any
                                            > > reason why that is so?
                                            >
                                            > In the original MillenniumChess site this rule is described at
                                            > "http://www.vipchess.com/MillenniumGameRules.htm":
                                            >
                                            > | When pawns make it to the opposing back row...
                                            > | only pieces already captured can be re-instated.
                                            > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                            >
                                            > and also at "http://www.vipchess.com/millenniumFAQ'S.htm":
                                            >
                                            > | When Pawns Advance To The Opposing Back Row, Can I Reclaim
                                            > | A Lost King?
                                            > |
                                            > | No! You may not reclaim lost Kings. Also, pawns do
                                            > | not automatically become Queens.
                                            > | You may only reclaim pieces that have been captured.
                                            > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                            >
                                            > therefore I implemented it in the zrf.

                                            Ok, silly me. I will need to change it back and send it to Ed. ARGH!

                                            So much for me presuming.
                                          • Alfred Pfeiffer
                                            ... No, you could build a position where also the two pre-kings like the other own pieces will be blocked. Do you need an example? White: pKing a15, pKing a14,
                                            Message 21 of 26 , Dec 3, 2007
                                            View Source
                                            • 0 Attachment
                                              --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                                              >
                                              > I believe a pre-King will always have a regular move so the
                                              > irregular null move will never be reached. A stalemate with
                                              > two pre-kings would be just that.

                                              No, you could build a position where also the two pre-kings
                                              like the other own pieces will be blocked. Do you need an
                                              example?

                                              White: pKing a15, pKing a14, Rook b15, Bishop c15,
                                              Pawn a13, b14, b13, d14;
                                              Black: King d15, Bishop e14;
                                              White to move is stalemated.

                                              > Zillions would move until a 3 position
                                              > repetition happened and a draw was declared.

                                              I tried it. If the Zillions engine has to move in the
                                              stalemated position it will pass. If you as human has to
                                              play in a such situation you cannot do anything (the 'Pass'
                                              button will not be activeted!) except ask for a hint, what
                                              give 'pass' as the result.

                                              Maybe we could include a test in the move block for the
                                              move-type 'irregular' that checks whether moves for the
                                              pre-king are possible.

                                              >
                                              > Why must the nmKing piece not be tweaked? At its untweaked
                                              > value it is worth less than a pawn and the AI will make no
                                              > attempt to protect it.
                                              >
                                              > --Keith
                                              >

                                              The 'nmKing' is a usual King that never moved, means it has
                                              the right to castling, and is a target for checks etc.
                                              It has the nearly same value as a King. It will occur when
                                              a pre-king was captured and the second pre-king did not move
                                              until then (it is still a 'nmpKing') and therfore will be
                                              changed into a 'nmKing'.


                                              Alfred Pfeiffer
                                            • Alfred Pfeiffer
                                              I have to correct the example for the stalemate. Sorry when I was confusing you. ... Please read White: pKing a8, pKing a7, Rook b8, Bishop c8, Pawn a6, b7,
                                              Message 22 of 26 , Dec 3, 2007
                                              View Source
                                              • 0 Attachment
                                                I have to correct the example for the stalemate.
                                                Sorry when I was confusing you.

                                                --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@...> wrote:
                                                >
                                                > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@> wrote:
                                                >>
                                                >> I believe a pre-King will always have a regular move so the
                                                >> irregular null move will never be reached. A stalemate with
                                                >> two pre-kings would be just that.
                                                >
                                                > No, you could build a position where also the two pre-kings
                                                > like the other own pieces will be blocked. Do you need an
                                                > example?
                                                >
                                                > White: pKing a15, pKing a14, Rook b15, Bishop c15,
                                                > Pawn a13, b14, b13, d14;
                                                > Black: King d15, Bishop e14;
                                                > White to move is stalemated.

                                                Please read

                                                White: pKing a8, pKing a7, Rook b8, Bishop c8,
                                                Pawn a6, b7, b6, d7;
                                                Black: King d8, Bishop e7;
                                                White to move is stalemated.

                                                > ...

                                                Alfred Pfeiffer
                                              • Mats Winther
                                                ... chess ... techniques ... I don t know what UCT is, but the UCI engines are amazing. UCI stands for universal chess interface. They are exe-files (open
                                                Message 23 of 26 , Dec 3, 2007
                                                View Source
                                                • 0 Attachment
                                                  --- In axiom-system@yahoogroups.com, "Greg Schmidt"
                                                  <gschmidt958@...> wrote:
                                                  >
                                                  > --- In axiom-system@yahoogroups.com, "Mats Winther" <mwi5@>
                                                  wrote:
                                                  > > ... I wonder whether it's possible to use the c2forth program to
                                                  > convert algorithms from any of the open source chess programs, ...
                                                  >
                                                  > I suspect a rewrite might be easier. One reason being that each
                                                  chess
                                                  > program has it's own unique way of representing the pieces and the
                                                  > board. Some use bitboards, some use explicit arrays etc., chess
                                                  > programs typically rely on very customized and specialized
                                                  techniques
                                                  > in their implementation. The code would have to be understood, not
                                                  > blindly translated. As I sometimes like to say, there's no substitute
                                                  > for actually knowing what you're doing.
                                                  >
                                                  > Adding a UCT engine would simplify the task of creating a Chess
                                                  > program since an evaluation function would not have to be crafted. I
                                                  > don't know how strong it would play chess though as I'm unable to find
                                                  > any references on applying UCT to Chess. I will say that it is a
                                                  > project that is of interest to me.
                                                  >
                                                  > -- Greg
                                                  >

                                                  I don't know what UCT is, but the UCI engines are amazing. UCI stands
                                                  for universal chess interface. They are exe-files (open source) with
                                                  which you communicate by sending strings. The best have grandmaster
                                                  strength. They include time management, setting up of positions, etc. If
                                                  Zillions is used as an interface, then it can only play regular chess, and
                                                  this is no great gain as such. But somebody could later work on this
                                                  implementation and make other variants by changing the board size
                                                  etc. in the C code.
                                                  The UCI protocol can be downloaded here:
                                                  http://download.shredderchess.com/div/uci.zip

                                                  Mats
                                                • Keith Carter
                                                  No problem. Your good example arrived before there was any confusion. I see the problem. Stalemated by being blocked in. I have something I want to try.
                                                  Message 24 of 26 , Dec 3, 2007
                                                  View Source
                                                  • 0 Attachment
                                                    No problem. Your good example arrived before there was any
                                                    confusion. I see the problem. Stalemated by being blocked in. I
                                                    have something I want to try. The pre-king has multiple sets of
                                                    extra moves, each set being identical to its regular move. I will
                                                    give it a try some time today and get back to you.

                                                    --Keith


                                                    --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@...>
                                                    wrote:
                                                    >
                                                    > I have to correct the example for the stalemate.
                                                    > Sorry when I was confusing you.
                                                    >
                                                    > --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@>
                                                    wrote:
                                                    > >
                                                    > > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@>
                                                    wrote:
                                                    > >>
                                                    > >> I believe a pre-King will always have a regular move so the
                                                    > >> irregular null move will never be reached. A stalemate with
                                                    > >> two pre-kings would be just that.
                                                    > >
                                                    > > No, you could build a position where also the two pre-kings
                                                    > > like the other own pieces will be blocked. Do you need an
                                                    > > example?
                                                    > >
                                                    > > White: pKing a15, pKing a14, Rook b15, Bishop c15,
                                                    > > Pawn a13, b14, b13, d14;
                                                    > > Black: King d15, Bishop e14;
                                                    > > White to move is stalemated.
                                                    >
                                                    > Please read
                                                    >
                                                    > White: pKing a8, pKing a7, Rook b8, Bishop c8,
                                                    > Pawn a6, b7, b6, d7;
                                                    > Black: King d8, Bishop e7;
                                                    > White to move is stalemated.
                                                    >
                                                    > > ...
                                                    >
                                                    > Alfred Pfeiffer
                                                    >
                                                  • Keith Carter
                                                    The multiple sets of regular moves works. Your stalemate example now produces a stalemate instead of a pass move. With four extra sets of regular moves the
                                                    Message 25 of 26 , Dec 3, 2007
                                                    View Source
                                                    • 0 Attachment
                                                      The multiple sets of regular moves works. Your stalemate example now
                                                      produces a stalemate instead of a pass move. With four extra sets of
                                                      regular moves the pre-King is worth about 58,500 and a King is worth
                                                      11380 (a queen is worth 34,500). I am not sure how far to push up
                                                      the pre-King's value. Is more than a queen enough? Does it need to
                                                      be about the same as a King?

                                                      I will post my changes. It is based on your post with the single
                                                      pawn type. If nothing else, take the file as a proof of concept and
                                                      add or subtract more sets of moves till you get the value you think
                                                      works best.

                                                      --Keith


                                                      --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@...> wrote:
                                                      >
                                                      > No problem. Your good example arrived before there was any
                                                      > confusion. I see the problem. Stalemated by being blocked in. I
                                                      > have something I want to try. The pre-king has multiple sets of
                                                      > extra moves, each set being identical to its regular move. I will
                                                      > give it a try some time today and get back to you.
                                                      >
                                                      > --Keith
                                                      >
                                                      >
                                                      > --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@>
                                                      > wrote:
                                                      > >
                                                      > > I have to correct the example for the stalemate.
                                                      > > Sorry when I was confusing you.
                                                      > >
                                                      > > --- In axiom-system@yahoogroups.com, "Alfred Pfeiffer" <apfde@>
                                                      > wrote:
                                                      > > >
                                                      > > > --- In axiom-system@yahoogroups.com, "Keith Carter" <keith@>
                                                      > wrote:
                                                      > > >>
                                                      > > >> I believe a pre-King will always have a regular move so the
                                                      > > >> irregular null move will never be reached. A stalemate with
                                                      > > >> two pre-kings would be just that.
                                                      > > >
                                                      > > > No, you could build a position where also the two pre-kings
                                                      > > > like the other own pieces will be blocked. Do you need an
                                                      > > > example?
                                                      > > >
                                                      > > > White: pKing a15, pKing a14, Rook b15, Bishop c15,
                                                      > > > Pawn a13, b14, b13, d14;
                                                      > > > Black: King d15, Bishop e14;
                                                      > > > White to move is stalemated.
                                                      > >
                                                      > > Please read
                                                      > >
                                                      > > White: pKing a8, pKing a7, Rook b8, Bishop c8,
                                                      > > Pawn a6, b7, b6, d7;
                                                      > > Black: King d8, Bishop e7;
                                                      > > White to move is stalemated.
                                                      > >
                                                      > > > ...
                                                      > >
                                                      > > Alfred Pfeiffer
                                                      > >
                                                      >
                                                    • Greg Schmidt
                                                      ... stands UCT (Upper Confidence bounds applied to Trees) is a new algorithm for game tree search (see post #144 in this group). It was used in a Go program
                                                      Message 26 of 26 , Dec 3, 2007
                                                      View Source
                                                      • 0 Attachment
                                                        --- In axiom-system@yahoogroups.com, "Mats Winther" <mwi5@...> wrote:
                                                        > I don't know what UCT is, but the UCI engines are amazing. UCI
                                                        stands

                                                        UCT (Upper Confidence bounds applied to Trees) is a new algorithm for
                                                        game tree search (see post #144 in this group). It was used in a Go
                                                        program called "MoGO" and is ranked first on the 9x9 Computer Go
                                                        Server since August 2006. MoGo won the computer Go KGS tournaments
                                                        of October and November 2006 (9x9 and 13x13) and was second in
                                                        December 2006 (19x19).

                                                        > The UCI protocol can be downloaded here:
                                                        > http://download.shredderchess.com/div/uci.zip

                                                        I took a quick look at this. I see no reason why a Zillions plug-in
                                                        couldn't be developed to support this protocol. In fact, Zillions
                                                        has already been adapted to the WinBoard protocol via a plug-in.

                                                        See:
                                                        http://www.geocities.com/andreas_kaufmann2000/winboard_adapter.htm

                                                        -- Greg
                                                      Your message has been successfully submitted and would be delivered to recipients shortly.