Up until now, I've been doing all my engine vs. engine testing from the standard starting position. It seems to me though it would be better to test from a broader range of initial positions, since otherwise my results may be biased by some small peculiarities in how different engines open when no using an opening book.
Are there typical positions people use for this purpose? Ideally I'd just like a set of FEN strings of representative early game positions I could do test runs from. Is there a good set for this sort of situation around, or should I go ahead and generate my own?
Thanks for any pointers,
Matt Gingell
https://sourceforge.net/apps/wordpress/chesley/
Starting positions for engine vs. engine testing.
Moderators: hgm, Rebel, chrisw
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: Starting positions for engine vs. engine testing.
if you go to ftp.cis.uab.edu/pub/hyatt/tests, the file "openings.epd" contains 4000 starting positions that I use for my cluster testing. They are 12 move into the game, with the most popular 4,000 positions chosen. Feel free to use them if you want.
-
- Posts: 2851
- Joined: Wed Mar 08, 2006 10:01 pm
- Location: Irvine, CA, USA
Re: Starting positions for engine vs. engine testing.
ftp://ftp.cis.uab.edu/pub/hyatt/tests works better for me.bob wrote:if you go to ftp.cis.uab.edu/pub/hyatt/tests, the file "openings.epd" contains 4000 starting positions that I use for my cluster testing. They are 12 move into the game, with the most popular 4,000 positions chosen. Feel free to use them if you want.
Re: Starting positions for engine vs. engine testing.
That's exactly what I was looking for. Thanks very much.bob wrote:if you go to ftp.cis.uab.edu/pub/hyatt/tests, the file "openings.epd" contains 4000 starting positions that I use for my cluster testing. They are 12 move into the game, with the most popular 4,000 positions chosen. Feel free to use them if you want.