char set slower in 3.1.0 compared to 3.0.4??

[seelisch]

Example from Viktor, investigate (Hans, Gerhard)

[seelisch]

If this is true, then why did we not notice (by means of general test set when releasing a new version)?

[levandov]

Excerpts from email correspondence of Valery Romanovski and Japanese guys from RISA/Asir team, 2009 are to be found below. As examples of computation nearly any example coming form Romanovski can be taken: there are complicated minimal primes. The most challenging examples can be found in my directory ​http://www.math.rwth-aachen.de/~Viktor.Levandovskyy/filez/chals/ (mckay.tst is the only examples not of this type). A typical one is ​http://www.math.rwth-aachen.de/~Viktor.Levandovskyy/filez/chals/vr_five11.tst .

I remember this change of performance on Romanovski's problems very distinctly, since minAssChar was our default instrument.

Regards, Viktor

Romanovski: ...with the recent version of Singular minAssGTZ found the decomposition in about 2 min, but minAssChar is still running (more than one hour).

Japanese:

1. Unfortunately, your problem was too difficult to solve by our technique. This problem brought about huge coefficients over Q.
2. We have recently found that in our problem, minAssChar in Singular 3-1-0 is about three times slower than that in Singular 3-0-4. We guess that this unfortunate phenomenon arises from a change of `factorization' of Singular 3-1-0.