|
After a period of 14 years, I have resumed the work of computer testing the problems reproduced in Fide Albums. In this period, computer testing is used increasingly by magazines prior to publishing new problems, which of course improves the quality of the magazines themselves, but also lowers the need for testing the Fide Albums.
A variety of programs make it possible to test almost any problem with a few major exceptions:
Computer technology has produced faster and faster chips, making it possible to test problems in an increasing number of moves. But also the versatility of the solving programs has increased:
While the first bullet is all good, the last bullet creates a new kind of testing, allowing for testing between 0% and 100% guarantee of correctness. While "fully tested" problems normally are identified by the symbol C+, one might use the symbol C- for "not tested" problems and the symbol C% for "partially tested" problems.
The first four revisions of the booklet Computer Test Fide Albums include test results of problems that are fully tested (C+), and only my own set of solving programs were used. Now the result of partially tested problems (C%) are included, and two other solving programs have been used. The preference for choosing a solving program has been:
| Own programs | #2->8, s#2->5, h#2->4, r#2->5 and "many fairy problems" | |
| Alybadix | #9->, s#6->, h#5->, r#6-> and "almost any other fairy problem" | |
| Popeye | Fairy problems not solvable by Alybadix |
Partially tested problems (C%): Problems in this category are all tested using Alybadix, and in the test report they are marked with the parameter set used for the testing. Be aware, that if a parameter is used, the result shown is produced under the conditions of this parameter. This is fine for flaws. Normally it is of little interest, whether all cooks or the variations of a cook are 100% precise or not. On the other side, the test result "OK" does not imply, that the problem is correct.
All test results are edited. This means that I have transformed the printout from a solving program to a single cell in a MS-Excel worksheet. In this process, errors can occur.
MS-Excel is chosen as publishing form, because it offers the flexibility to produce reports using sorting and filtering functions. Thus the layout used in the booklets can easily be obtained. Please note the last three columns. They represent sorting sequence within a problem, alternative solving program and solving parameter used.
The parameter setting for Alybadix testing of partially tested problems is the same as the settings used by mr. Ilkka Blom, Finland, in his testing of Fide Albums, published as "Alybase FAS 5 database". I doubt that I will be able to find better settings than the ones he has used. In fact, my work for partially tested problems has been transforming the printout from Alybadix into a uniform layout, suitable for the chosen format.
Unchanged from previous revisions:
2009.03.18
Steen Christensen
Skodsborgvej 350, 2.tv.
2850 Nærum
Denmark
steenchr@comxnet.dk