Dear all,
There was indeed a problem in the version 2.0.0-4 of the alps2 Ubuntu Lucid 10.4 package.
S. Aicardy wrote:
The trouble was caused by the RPATH linking in the package which
forced the
tebd program to use a wrong version of lapack. In Ubuntu 10.10 the problem didn't exist because there is only one version of lapack available.
The new version 2.0.0-5 of that package has corrected this bug.
Many thanks to Michael Wall & Stephane Aicardi for their help.
Gregoire Misguich.
Le 27/01/2011 15:14, Grégoire Misguich a écrit :
Dear ALPS community,
As my first attempt to use the TEBD code provided in alps-2.0, I ran a tutorial concerning the simulation of the XX spin chain (tutorial_2a.py), which simulates the dynamics starting from a domain wall spin configuration:
http://alps.comp-phys.org/mediawiki/index.php/ALPS_2_Tutorials:TEBD-02_kink
I got the following figures (attached .png files).
The Wiki says "We see that the magnetization agrees very well to visual accuracy", but in the plot I obtained, it is not quite the case (very "noisy"). I played a bit with some parameters (increasing the number of time steps and the max Schmidt dimension), but it still get some ugly curves.
So, I suspect that something might we wrong...
Any help or suggestion would be very much appreciated !
Gregoire Misguich.