From: Axel Kohlmeyer (
Date: Mon Jan 02 2012 - 12:21:06 CST

On Mon, 2012-01-02 at 15:52 +0330, sarah k wrote:
> Dear Axel Kohlmeyer
> yes, in principle that is possible, but it will
> require some programming or scripting. a big question
> is to what level of detail and accuracy (and what
> strength of magnetic field) do you want to do the
> study.
> I'm going to study the effect of low frequency electromagnetic fields
> (30-100 Hz) at several temperatures. Actually, It's a part of my MSc.
> thesis and I want to perform both experimental and simulation studies.
> The results will be compared to each other so I need high precision
> and accuracy and reproducibility of results.

rather than worrying about high precision and
accuracy or MD codes, you should do a reality
check on time scales and size scales.

a 100Hz process has a period of 10 milliseconds,
while atomistic MD simulations have time steps on
the order of 1 femto second. have you considered
how long it will take and how many time steps to
model your process. not to mention, that macroscopic
properties are often only detectable through doing
an average over many objects. so have you thought
about how much compute power that would need and
how large a sample with how many events you have
to simulate?

another question you should answer up front:
what is your observable? and what would be
the connection to the experiment? if you don't
know up front (and don't make any tests on
simple systems), you may do a lot of simulation
work without recording the important information
for post-processing and thus waste a lot of time
and effort.

> classical MD just uses classical mechanics,
> so all you need to do is to compute the force
> vector on a moving charge in the magnetic field.
> in the simplest case of a homogeneous, strong field
> it would simply be an additional force term that
> is orthogonal to the field and velocity vector
> and scaled by the velocity and magnitude of the field.
> Dr. Axel Kohlmeyer
> College of Science and Technology,
> Institute for Computational Molecular Science,
> Temple University, Philadelphia PA, USA.

> I know that simulation is not like in vitro studies but can I hope
> that if electric field be applied to the system, magnetic field will
> be consequently produced? The proteins have polar residues and as a

this is nonsense. please read up on the elementary
physics of the effect of electrostatic and magnetic
fields on charged classical particles. also, *no*
scientific study should hinge on "but can i hope"?

> logical result external electrical field will make the electrical
> dipole momentum rotate and respectively magnetic field will be formed.

you are mixing up a magnetic field generated by molecules
and applying a magnetic field. and no. applying an electrostatic
field is not the same as applying a magnetic field.
for simplicity, lets assume an electrostatic field now, have
you considered that at the length of the time steps of an MD
simulation, a 100HZ varying E-field is quasi-static.
and at which strength, btw? try computing the resulting
additional force relative to typical forces. it may be
barely within the range of numerical accuracy that you
have with floating point numbers.

> So is namd trustable for my purpose if I only apply electric field?

it is not about namd being reliable, but as i wrote above,
your entire simulation project is an ill conceived study
and full of problems beyond the capability of MD software.
unless you resolve those, there is no point in even
trying to get started.


> Thanks.
> Regards,
> Fatemeh

Dr. Axel Kohlmeyer
College of Science and Technology,
Institute for Computational Molecular Science,
Temple University, Philadelphia PA, USA.