Hi,
whenever you start scidvspc, open a database, open the tree window and push the "train" button you get always the same moves. The random number/move generator is not working on linux. (scid 4.26).
On Windows it seems to work.
Did you try it a couple of times? When I delete the .scidvspc directory I get at the first and the second time a different variation. To test that you need to close scidvspc and start it again at least times.
Maybe it was a misunderstanding. When you just restart the training you get different variations. But when you close scidvspc and start it again, you get the same openings in exact the same order.
I get when startig with e2e4: sicilian, aliekhin, ruy lopez ....
Every time the exact openngs and moves.
But you have to start and close the program at least three times.
I think, the weighted random is what is most interesting. A workaround what be to build a opening file but I cant find polyglot anymore.
Last edit: Thomas Gnielka 4 days ago
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Maybe it was a misunderstanding. When you just restart the training you get different variations. But when you close scidvspc and start it again, you get the same openings in exact the same order.
I think you may be right.
I will commit this change to svn
Index: src/tkscid.cpp===================================================================--- src/tkscid.cpp (revision 3586)+++ src/tkscid.cpp (working copy)@@ -507,6 +507,7 @@ scratchPos = new Position;
scratchGame = new Game;
db = &(dbList[currentBase]);
+ srandom32 (time(NULL)); // Seed rand num gen. Only used by tree training ? - S.A.#ifndef TCL_ONLY
#ifndef __APPLE__
I thing, the weighted random is what is most interesting. A workaround what be to build a opening file but I cant find polyglot anymore.
Hmmm... Seems to work for me, though computers are bad at random numbers.
But it's probably weighted random. Maybe pure random move would be more desirable?
Did you try it a couple of times? When I delete the .scidvspc directory I get at the first and the second time a different variation. To test that you need to close scidvspc and start it again at least times.
Maybe it was a misunderstanding. When you just restart the training you get different variations. But when you close scidvspc and start it again, you get the same openings in exact the same order.
I get when startig with e2e4: sicilian, aliekhin, ruy lopez ....
Every time the exact openngs and moves.
But you have to start and close the program at least three times.
I think, the weighted random is what is most interesting. A workaround what be to build a opening file but I cant find polyglot anymore.
Last edit: Thomas Gnielka 4 days ago
I think you may be right.
I will commit this change to svn
Maybe this is what you want ?
https://sourceforge.net/projects/scidvspc/files/support%20files/polyglot1.5.zip/download