Hi,
The "energy" calculation from cpptraj will typically only calculate
potential energies (if velocities are present it will also try to
calculate kinetic energies, but that is less reliable due to how the
leapfrog integrator estimates kinetic energy; cpptraj doesn't have all
that information because it's typically not written to trajectory
files). Also, if you want to have the "exact" energies you need to
make sure you're using PME and the same PME parameters (direct sum
tolerance, spline order, etc). Cpptraj uses the same defaults as
sander/pmemd, but you should double check.
As far as speed, you can improve things in two ways. One is to use the
OpenMP-compiled cpptraj; the energy calculation will benefit from this
quite a bit. The other is to use MPI-compiled cpptraj to divide the
input trajectory up. Ideally, use both (the OpenMP and MPI compiled
code) and you can benefit from both, just be sure that the number of
MPI processes times number of OpenMP threads per process doesn't
exceed the number of available cores.
Hope this helps.
-Dan
On Thu, Apr 23, 2020 at 9:01 AM Eduardo R. Almeida
<eduardoe.r.a.hotmail.com> wrote:
>
> Dear Amber users,
>
> I would like to analyze the energies printed in the mdout files referring only to the solute molecules. On the other words, I would like to build and analyze the graphics of these properties (potential energy, kinetic energy, total energy, temperature, pressure, density, volume, ...) excluding the values referring to the solvents molecules. Would it be possible?
>
> In order to do that, I used the energy calculation performed by the cpptraj where I selected the solute atoms (only including the lipids molecules) in a atom mask for this calculation. Specifically, this calculation has taken a long time (seven days for a lipid bilayer). Therefore, I would like to know if this alternative really calculates these macroscopic properties generated in the mdout files referring only to the solute molecule.
>
> Best regards,
>
> Eduardo.
>
>
> ---------------------------------------------------------------------------
> Eduardo R. Almeida
> PhD student, Computational chemistry
> Núcleo de Estudos em Química Computacional (NEQC)
> Universidade Federal de Juiz de Fora (UFJF)
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 24 2020 - 09:30:03 PDT
