??Dear users,
After searching in the source code of Amber 14, I saw that it is possible to use the results of a PCA to perform a steered MD by specifying the type 'PCA' in the ncsu_smd section of my mdin file (see below).
I ran my mdin file with sander and for some reasons, it fails when calling the subroutine colvar_mdread in ncsu-colvar.F90:
[Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7F8C7B7DD6F7
#1 0x7F8C7B7DDCC4
#2 0x7F8C7ACFF0DF
#3 0x77B966 in __ncsu_colvar_MOD_colvar_mdread
#4 0x7A57C8 in __ncsu_smd_hooks_MOD_on_sander_init
#5 0x7A1B19 in __ncsu_sander_hooks_MOD_on_sander_init
#6 0x47D452 in sander_
#7 0x4766C3 in MAIN__ at multisander.F90:?
Segmentation fault]
I am not a Fortran expert but I wonder how can I know where it fails exactly?
Thanks for your help in advance,
Best regards,
G. Roellinger
---------------------------------------------
ncsu_smd section of the mdin file
---------------------------------------------
ncsu_smd
output_file = 'smd.txt'
output_freq = 1000
variable
type = PCA
refcrd = 'refcrd.inpcrd'
avgcrd = 'avgcrd.inpcrd'
evec = 'evecs.dat'
end variable
end ncsu_smd
----------------------
Details on files used
----------------------
evecs.dat contains two header lines (which are ignored when calling the read_evec subroutine from ncsu-read-pca.F90) and the 3N eigenvector of interest from the PCA
avgcrd.inpcrd contains two header lines (also ignored) and my average coordinates (3N)
refcrd.inpcrd contains two header lines (also ignored) and my reference coordinates (3N)?
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Dec 11 2014 - 01:00:02 PST
