Re: [AMBER] Entropy calculation

From: Jason Swails <>
Date: Wed, 29 Oct 2014 13:44:29 -0400

On Tue, Oct 28, 2014 at 10:38 PM, asdda asa <> wrote:

> Dear Amber users,
> I am performing entropy calculation using nmode.
> My system consist of a protein with 213 residue and a ligand.
> First, I did a try with only 4 frames and drms=0.1.
> My calculation succeed to reach convergence, but then fail with following
> message:
> Line minimizer aborted: step at upper bound 0.061508653
> Line minimizer aborted: step at upper bound 0.011450028
> Line minimizer aborted: step at upper bound 0.025196144
> Line minimizer aborted: step at upper bound 0.044987034
> Line minimizer aborted: step at upper bound 0.00036005744
> Line minimizer aborted: step at upper bound 0.003992881
> FATAL: allocation failure in vector()
> CalcError: /usr/local/AMBER/amber12/bin/mmpbsa_py_nabnmode failed with
> prmtop complex.prmtop!
> Exiting. All files have been retained.\
> From the massage, the problem might be due to prmtop file,

​No, the error message was simply telling you where the calculation died
(not why it died). It's possible that the prmtop file has problems that
make finding a minimum difficult, but I doubt that is the case.

The actual problem is the "allocation failure", which indicates that a
malloc call failed. As far as I know, memory allocation fails either when
there is not enough memory available or when some memory becomes corrupt
(due to a bug in the program).

> but I was using the same prmtop file to calculate entalphy, and it was
> ​​
> succesfull.
> I am using a single computer with 48 GB memory, so it should be no problem

​That does make the error more puzzling. However, it is not always the
case that a program has access to all of the physical memory in the system
(a 32-bit program, for instance, cannot address more than ~4GB of memory).
Also, depending on how large your system is, it is still possible that 48
GB is not enough. The Hessian matrix is 3N x 3N (since it's symmetric, the
actual requirement is a little more than half that number). At full double
precision (8 bytes per number), 48 GB would be consumed completely by
~20-30 thousand atoms (assuming the full Hessian is stored rather than just
the upper- or lower-triangular portion). While this is very big for an
unsolvated system, it is not unfathomably large.

If you still can't get this working, we would need to be able to reproduce
the problem on our own machine(s) to help further.

​All the best,

Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Wed Oct 29 2014 - 11:00:02 PDT
Custom Search