Chess Test Suite by Kevin Lang and Warren Smith.

The test suite is in the files ending with ".fin". (Warning:
hortjan.fin contains errors and is not really final. If anybody can
get hold of a copy of this book and fix up this dataset, would be
appreciated.) Total of 5551 positions as of June 1993. Some extra
positions are in "oldpos.wds" and "bloss.sje", but they are not been
converted into our standard format. 

Late additions (September 1993): dvor.fin and krabbe.fin.

The files ending ".gnu" are the results of a run of gnuchess 4.0 pl62,
running on SGI Indigos with 50MHz R4000 MIPS RISC processors, at 1
minute CPU time per problem, on the corresponding .fin files. The
".gnu15" files are same but with 15 minutes CPU time per problem.

The files ending ".ps" are documentation explaining everything.  They
need to be printed out on a "postscript" printer.  "diagram1.ps" up to
"diagram3.ps" are chessboard figures. Diagram 1 should be used to fill
the blank spot on page 3 of "testpos.ps", which is the main item of
documentation, in fact is the article accepted by ICCA Journal july
1993 and final revision made on Sept 1 1993. Diagrams 2 and 3 also go
with the article.

Please send known corrections, etc to Warren Smith:
NEC Research Institute (Room 1D04) 4 Independence Way Princeton NJ 08540 USA
Email: wds@research.NJ.NEC.COM

The following impressive paper has just been brought to my attention.
It concerns an automated system for scanning in games from "Chess
Informants".

 Henry S. Baird and Ken Thompson:
 "Reading Chess,"
 IEEE Trans. Pattern Analysis and Machine Intelligence
 12,6 (June 1990) 552-559

Please note any modifications to the test suite in the approproate test file
header with dates, name of modifier etc.

-----------------------------------

