Hi,
does anybody (Bob?) have more informations about the program Chaos and how it worked? It apparently was a quite strong Best-First searcher, keeping the search tree inside memory. 50nps and that strong! I would like to learn more about that program and Iterative Widening. Was it a parallel program?
A second question. Are the two Chaos operators in Belle vs. Chaos Ira Rubin and Fred Swartz?
I recall that Chaos was described as a Shannon Type B program; i.e., selective search using a plausible move generator. Perhaps the plausibility score was in some way path dependent as was the case for Greenblatt's Mac Hack VI.
Only a few programs have used a whole-tree-in-memory approach as did Chaos. However, some algorithms like best-first and Berliner's B* require such. The Lisp side of Symbolic stores the entire tree and this is being ported to the revised C++ version. With a RAM space of up to 32 GB, there's a lot of room for experimentation with whole tree searching.
Gerd Isenberg wrote:Hi,
does anybody (Bob?) have more informations about the program Chaos and how it worked? It apparently was a quite strong Best-First searcher, keeping the search tree inside memory. 50nps and that strong! I would like to learn more about that program and Iterative Widening. Was it a parallel program?
A second question. Are the two Chaos operators in Belle vs. Chaos Ira Rubin and Fred Swartz?
Thanks,
Gerd
To the best of my knowledge, Chaos was a traditional alpha/beta search (depth-first) but with the traditional (at the time) forward-pruning to cull moves that appeared to be unimportant. It was never parallel. Generally ran on a largish Amdahl-type box although we all moved around architecture-wise back in the late 70's. I do not remember Ira at all, but yes, the other is Fred. We used to communicate all the time in fact...
The "best-first" idea you mentioned doesn't ring a bell. (belle? ) but it certainly could be correct. I remembered Chaos as a "Coko-type" program which was just a shannon-type-b search with selectivity similar to what Greenblatt did in his program.
Can't help you much on the technical details of CHAOS since I have not yet seen the source for it.
I can say, however, that if you want, you could contact the authors (Fred S. or Mike A.) directly and get their direct say. (Probably Mike A. might be more responsive.)
Last I heard, Mike was trying to revive his old CHAOS program under an emulator but was having some trouble. (Old OS he was trying to remember how to use & program & he was having to do it under an emulator, which is rather tedious.)
If you want Mike or Fred's email address, PM me. I think I still have them somewhere back from when I contacted them about releasing their program to the public.
Oh, I should also point out that Mike A. is aware of the CPW and I have suggested he add some details about the CHAOS program, selective searching, etc. to the wiki. I don't know if he has had the time or urge to do that, though.
And Tony M. (of AWIT, the other strong selective search program) also knows about CPW. Again, I don't know if he's actually had the urge to contribute his knowledge of selective search stuff to the wiki.
Carey wrote:Can't help you much on the technical details of CHAOS since I have not yet seen the source for it.
I can say, however, that if you want, you could contact the authors (Fred S. or Mike A.) directly and get their direct say. (Probably Mike A. might be more responsive.)
Last I heard, Mike was trying to revive his old CHAOS program under an emulator but was having some trouble. (Old OS he was trying to remember how to use & program & he was having to do it under an emulator, which is rather tedious.)
If you want Mike or Fred's email address, PM me. I think I still have them somewhere back from when I contacted them about releasing their program to the public.
Oh, I should also point out that Mike A. is aware of the CPW and I have suggested he add some details about the CHAOS program, selective searching, etc. to the wiki. I don't know if he has had the time or urge to do that, though.
And Tony M. (of AWIT, the other strong selective search program) also knows about CPW. Again, I don't know if he's actually had the urge to contribute his knowledge of selective search stuff to the wiki.
Carey
Of course, it would be great, if original authors may provide technical or brief biographical content for CPW, directly or via pm or mail. I would gladly do some formatting. But I don't like to bother people personally with requests to do so.
) but it certainly could be correct. I remembered Chaos as a "Coko-type" program which was just a shannon-type-b search with selectivity similar to what Greenblatt did in his program.