Dear all,
As per Dave's suggestion to the list regarding setting ig = -1 as the
default for random number generator seeds. This would set the random
seed to truly 'random' in the sense of picking up the seed from the
system clock, making it incredibly unlikely that you ever get the same
one twice.
I am all for the change, but let me make clear what the possible cons are:
1. It would make it impossible to reproduce data exactly (let's say for
debugging) unless the user sets him/herself the seed by hand.
2. The test cases MUST have a seed set by hand, otherwise they would
never pass.
These are no problems in general, unless the users copy and paste test
cases and adapt them to their own needs. We do this a lot. When I need
to run a new feature I am not familiar with, I look at the test cases,
learn from them and change what I need to change. If the test cases have
a fix ig value and users do not change it, then we are back to square one.
I do not know how to solve this, but I welcome suggestions. This might
be a perfect time to make the change !
Adrian
--
Dr. Adrian E. Roitberg
Associate Professor
Quantum Theory Project
Department of Chemistry
Senior Editor. Journal of Physical Chemistry
American Chemical Society
University of Florida PHONE 352 392-6972
P.O. Box 118435 FAX 352 392-8722
Gainesville, FL 32611-8435 Email adrian.qtp.ufl.edu
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Apr 13 2010 - 07:00:02 PDT
