Dear all,
I ran a SOM clustering analysis using ptraj from AT1.5 (patch level 27)
on a NetCDF trajectory of a medium sized system (ca. 60 residues) and
obtained a non-reproducible segmentation fault:
When I rerun the ptraj analysis (same machine), the calculation finishes
sometimes successfully, sometimes it crashes again; rerunning for
several times finally yields a result.
Did anyone experience a similar behaviour?
A collegue from the IT did a first analysis and came up with an error
backtrace (see below); that was, however, not the location where the
putative array out of bounds error occured, he told me. Although he'll
try to figure out the exact error location, I wanted to write to the
Amber list, because he's very busy at the moment with other things.
My second question is thus simply, whether I can trust the results from
my clustering analysis or not - and use a different clustering scheme.
Thanks in advance.
Anselm Horn
Bioinformatik
Institut für Biochemie
Friedrich-Alexander-Universität Erlangen-Nürnberg
Fahrstr. 17
91054 Erlangen
Germany
----------------------------------------
#0 0x00000000004f98c6 in PtrajClusteringConvertSOMNodesToClusters
(This=0x1af3be0, SOMNodes=0x1aeb3d0, ClusterCount=28261344) at
cluster.c:6782
6782 SOMNodes[ClusterIndex]->Stddevs[i] =
SOMNodes[ClusterIndex]->Stddevs[i] / PointCount -
(SOMNodes[ClusterIndex]->Means[i]) * (SOMNodes[ClusterIndex]->Means[i]);
(gdb) backtrace
#0 0x00000000004f98c6 in PtrajClusteringConvertSOMNodesToClusters
(This=0x1af3be0, SOMNodes=0x1aeb3d0, ClusterCount=28261344) at
cluster.c:6782
#1 0x00000000004f7a89 in PtrajClusteringClusterSOM (This=0x1af3be0,
ClusterCount=28226512, map=28261344) at cluster.c:6896
#2 0x0000000000470669 in PtrajPerformClustering (trajInfo=0x1af3be0,
action=0x1aeb3d0) at actions.c:2969
#3 0x000000000046ea06 in transformCluster (action=0x1af3be0,
x=0x1aeb3d0, y=0x1af3be0, z=0x0, box=0x1af3e70, mode=28226512) at
actions.c:3821
#4 0x00000000004640e6 in ptraj (filenamep=0x1af3be0
"\255\032\346.R\310\224.\376\337 .") at ptraj.c:7418
#5 0x0000000000441ad5 in interface (mode=28261344, filename=0x1aeb3d0
"\300F\257\001") at interface.c:92
#6 0x0000000000406c21 in main (argCount=28261344, argPointer=0x1aeb3d0)
at main.c:220
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Mar 30 2012 - 10:00:03 PDT
