Dear Amber developers and users,
Recently we reported a strange memory allocation problem about PTRAJ (please
see: http://archive.ambermd.org/201108/0721.html). In our knowledge because
this memory problem we reported for PTRAJ and because of "vector principal"
option hasn't been implemented for CPPTRAJ, it is not possible to extract
the principal axes vectors by using any of them.
Therefore, we decided to extract relevant information from trajectory files
with our own python code. Our code basically creates a pdb file for each
frame in trajectory. When this finishes, it calculates the principal axes of
inertia for the relevant residues in pdb file. As a sanity check we compared
its outputs with the old extracted data which we used PTRAJ routine of
AmberTools1.4. We unfortunately observed following behavior;
1) Center of mass calculations of PTRAJ and our code gave same results up to
some round off differences.
2) Eigenvalue, eigenvector calculations were dramatically different such
that if PTRAJ creates a vector with the components (a, b, c) our code
creates a vector with the components
(a, -c, -b). Results were same for all 3 eigenvectors.
We tested those results with 2 different Python codes and 1 MATLAB code.
Those codes were written by different people. Yet results were same up to
some round off differences, again.
We believe this is most probably a bug in PTRAJ, vector principal code.
Therefore we would like to let you know about it.
regards
peker milas
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Sep 01 2011 - 13:00:03 PDT
