AMBER: problem in quasih.f diagonalising matrix

From: Phineus Markwick <>
Date: Mon, 29 Nov 2004 14:14:35 +0100

Dear AMBER users,

I am having difficulties with the quasiharmonic analysis routine in AMBER 7.
The program reads in all the coordinates and mass file, generates an
average structure
which it writes to file xyz and constructs the matrix A. However, even
after running for 2 days
no further output (nor error message) is written.
I am running Redhat 8.0, Kernel 2.4 18-14bigmem on an i686 with Intel
compilers v7.0

Details for input:

 From an initial MD simulation of a 61 residue protein (927 atoms) using
explicit water solvent,
I have a crd file with 100 coordinate data sets.

using ptraj, I initially strip the water molecules and Na+ counter ions:

trajin mol.crd
center :1-61
image center familiar
strip :WAT
strip :Na+
trajout moldry.crd

from the restart file from the MD simulation, I get a pdb file for the
molecule (using ambpdb), and remove all the waters.
Then, using tleap I generate a topology file for the "dry" protein (ie.
without water molecules).

Using ptraj, I then generate an average structure:

trajin moldry.crd
center mass origin
rms first mass out rmsdata1.dat
average avg.pdb

and then use this to produce a new trajectory RMS fit to the average:

trajin moldry.crd
trajout moldry_fit.crd
reference avg.pdb
rms reference mass out rmsdata2.dat

The RMSD data seems sensible and I have viewed the resulting trajectory
in VMD, which also appears to be OK.
I have generated a mass file from the pdb (excluding water molecules

In quasih.f I have increased the parameter NATMAX from 800 to 1000 and
AMBER7 recompiled with no problems (no unresolved libraries etc. etc)

I then run quasih:

quasih -natom 927 -f 2 -m mass.dat -x xyz -v vecs -first 1 -last 100
-nmode 10 < moldry_fit.crd > molqh.out

As I noted previously, after a few seconds the file molqh.out is as follows:

Using 100 coordinate sets from 1 through 100
| Time to construct A: 0.13

and the file xyz is generated containing the average coordinates.
However, after 2 days there is no further output.
The program has not crashed and there are no error messages. Maybe I am
being naive, but I expected that the diagonalisation
of a 2781*2781 matrix would only take a few seconds. All the subroutines
in the /src/arpack directory including dsaupd.f are present
and complied.

Am I really being naive, or have I done something wrong?

Any help would be greatly appreciated,

with best regards,
Phineus Markwick.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Mon Nov 29 2004 - 13:53:00 PST
Custom Search