Author: Dann Corbit
Date: 18:32:22 07/20/04
Go up one level in this thread
On July 20, 2004 at 16:37:18, Steven Edwards wrote: >Symbolic: Status report 2004.07.20 > >Symbolic's underlying toolkit has two new classes for connecting to other >programs: CTMediator and CTXboard. > >CTMediator is for running automated games. Initially, the class will support >only a single, N-game match per invocation; eventually it will be able to run an >entire tournament across the network. Also, it will use the xboard protocol but >will likely have additional protocols added later. Crafty will be Symbolic's >initial automated sparring partner, just as it was Spector's last sparring >partner back in 1994. > >CTXboard is an interfacing class between xboard and the rest of the toolkit. >Instead of adding xboard commands to the current interactive command processor >(as is done in some other programs like Crafty), an instance of CTXboard will >replace an instance of CTCommProc and have its own xboard specific command >interpreter. This provides a good design firewall between two changing command >sets, one of which I do not control. The CTXboard class may also serve as a >template of sort for additional protocols; one of these would be the CTICS class >for ICS play. > >Auto competition run by the mediator is indicated by the -Mediator command line >option; the -Xboard option selects the xboard interface to Symbolic. If neither >option is present, the program runs a single instance of the interactive command >processor (CTCommProc) attached to the standard I/O streams. > >Testing the CTOppo class (opportunity search; i.e, ponder search) is also being >performed. > >Structurally, the CTCommProc, CTMediator, CTOppo, and CTXboard classes are >similar in that thay all inherit the CTContext class as a direct base class. >CTContext is the class that keeps track of the current values of the position, >the game, the control flags, the opening library interface, the tablebase cache >set, and various other objects. The idea is keep the toolkit design reasonably >well compartmentalized without it getting too bulky. > >At this point, I've gotten the CTXboard class to the point where it connects >reliably to xboard using xboard 's protocol version 2. Enough of the >functionality is present to run automated matches; however, I haven't yet fully >implemented the ponder features. As proof, here's some sample output, verbatim, >from xboard running under the X11 package under Mac OS X 10.3 on a 700 MHz G3 >PowerPC notebook: > >[Event "Computer chess game"] >[Site "jenn.local"] >[Date "2004.07.18"] >[Round "12"] >[White "Crafty-19.13"] >[Black "Symbolic"] >[Result "0-1"] >[TimeControl "300"] > >1. e4 e6 2. d4 d5 3. Nd2 c5 4. Ngf3 Nc6 5. exd5 exd5 6. Bb5 Bd6 7. O-O Ne7 >8. dxc5 Bxc5 9. Nb3 Bd6 10. Re1 O-O 11. Bg5 Bg4 12. Be2 h6 13. Bxe7 Bxe7 >14. h3 Bh5 15. c3 Bg6 16. Bd3 a5 17. Bxg6 fxg6 18. a4 Kh7 19. Nbd4 Qd7 20. >Ne6 Rf6 21. Qe2 Bd6 22. Rad1 Nd8 23. Ned4 Qxa4 24. Nb5 Qa2 25. c4 Bb4 26. >Nc7 Bxe1 27. Nxa8 Bb4 28. Rxd5 Ne6 29. Qc2 Nf4 30. Rd7 Nxh3+ 31. Kh2 Ng5 >32. Nxg5+ hxg5 33. Rxb7 Bd6+ 34. g3 Qa1 35. Qe2 Rf8 36. Nb6 Bc5 37. Nd7 >Rxf2+ 38. Qxf2 Bxf2 39. Nf6+ Kh6 40. Ng4+ Kh5 41. Nxf2 Qf1 42. Rf7 g4 43. >b3 Qe2 44. Kg2 g5 45. Rf5 Qe3 46. Rb5 Qf3+ 47. Kf1 Kg6 48. Rxa5 Qxb3 49. >Re5 Qxc4+ 50. Kg2 Qc6+ 51. Re4 Kf5 52. Kg1 Qxe4 53. Nxe4 Kxe4 54. Kf2 Kd3 >55. Kf1 Ke3 56. Ke1 Kf3 >{White resigns} 0-1 A win against crafty even at blitz speed is impressive at this early stage of development. Is this game a fluke or representitive of Symbolic's ability?
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.