My old Development log that later on was converted to a
Blog
When
|
What
|
Mars 14, 2005
|
|
Feb 09, 2005
|
No programming at all on Alaric. I had this annoying UCI bug in Terra and decided to
fix that. See more at Terra's News
|
Jan 22, 2005
|
Not much programming lately but KRPKR and KPK seems to work fairly ok now.
Did a new test with IID without any noticably improvement. Checked Tords idea
"Probably FL" pruning. No results yet. Most time spent on the final Terra release
that will come today.
|
Dec 9, 2004
|
Oops! I wrote Oct 2 on the table row below. It is of course Dec 2 - now corrected.
My first endgame suite tests are performed without the endgame knowledge
included. "Endgame Nunn" games are currently ongoing.
Results will be published here when endgame knowledge is included and new tests
to compare with are done.
|
Dec 2, 2004
|
Why does everything take more time than expected??? Finally, I will start with the
endgame part. I have a variety of test suites and some "Nunn" positions, endgame
positions to use as start positions in games with other engines.
|
Nov 16, 2004
|
The "not sorting" idea was really interesting. Will develop it further later on. Timing is
now working. Pondering is ok as well. Still some tweaking to use the time more
efficient, though.
|
Nov 1, 2004
|
Timing had to wait a few days. Will test a new idea about move generation without
really sorting but the moves will be sorted... Next: implement timing before starting
with the specialised Endgame engine.
|
Oct 28, 2004
|
Pawn hash and nullmove implemented. Games with Gerbil more even now with
some wins but still in Gerbils favour. Next: fix the timing.
|
Oct 16, 2004
|
My first playing version! It played 10 games with Gerbil. No hangs - no problems...
The result? 10-0 to Gerbil. Without quiescence, nullmoves and an inferior evaluation
that's no surprise at all!
|
Sept 20, 2004
|
Changed my mind... Went back to C++ again and made the first cut move generator
together with a perft function for testing reasons. Will continue with the move
generator for a while (which includes of course things like board representation). I
also will make the first Alpha/Beta function together with some evaluation stuff... and
then it's time for the end game engine!
|
Aug 26, 2004
|
Started to code in Ocaml. The "end game engine" is phase 1.
|
Beginning Aug 2004
|
Started to do GA tests in order to find out how to tune search parameters. Tests still
within Terra.
|
July 2004
|
vacation
|
June 2004
|
I started to make my first design attempts.
How could I use learning integrated with the developing of the engine?
Was GP or GA useful fir this?
I got collected quality pgn files from Dann Corbit that could be used for learning
Started to do some tests with GA in order to find out a good setup for tuning
evaluation parameters. Tests were done to Terra.
|
Winter 2003/2004
|
I decided to stop with Terra and start all over with a new engine.
I wanted to try out some other ideas that didn't fit in to the bitboard structure of Terra
and was bit bored of using bitboards for decades...
|
|
|
|
|