How does GROMACS deal with the external fields properly? Is the simulation correct?

GROMACS version: 2019.4
GROMACS modification: Yes/No

Dear all especially the developers,

I am trying to introduce a high-frequency and a relatively strong (like ~V/nm) electric field to the system and to investigate the effect of the E field on the bio-molecules. But I am worried if GROMCAS could simulate this case properly.

The thing is, once an E field is used, the system basically can’t be treated as equilibrated any more and there will be field energy deposited, which causes thermal effects. However, in the simulations, the thermostat will take extra thermal energy away and force the system to be equilibrated, so that the atom behaviours could be predicted by the proper formulas. If it is this case, does it mean the simulation might hide some physical facts? or the simulation data couldn’t reflect the real physics or even could be wrong?

I believe there must be a proper reason why the field module of GROMCAS (or other MD tools) was developed, for instance, it can reflect the reality under certain conditions. So what are the conditions? Is the field supposed to be weak and the frequency small?

The field effect on the bio-molecules is quite a popular topic. More and more researchers relay on the MD tools to investigate this. But I am really confused if MD tools could do this work correctly and completely? Any replies and comments would be much appreciated.

All the best,
Eva

Hi Eva,

I have encountered a similar issue in my work. To the best of my understanding of statistical physics, all definitions of canonical ensembles, which in MD are imposed via thermostats or barostats, make sense only at tdm equilibium; since I work in hydrodynamics that means no flow or equivalently no deterministic transport, which as you can very well understand should basically undermine all my research since hydrodynamics is the science of fluid flows.

Point is: how large is the error we make by assuming tmd equilibium? Example: the typical thermal velocity of a water molecule in vapour phase at 1 atm and 300K is around 600 m/s; if the deterministic flow of water has a speed of the order of a few meters per seconds then I most probably will be ok with utilizing an equilibirum thermostat (either the effective temperature would be only slighly smaller or the fluid flow will be locally hindered, but only by a negligible amount). However, if I want to simulate a hypersonic gas flow then I will most probably not be ok with an equilibrium themostat.

Is there a way to correctly resolve transport in non-equilibrium MD? Yes! Have a look at https://doi.org/10.1103/PhysRevE.68.046702 (unfortunately it is not implemented in GROMACS).

I am not sure if my example is somehow releted to your specific research, but the nature of issues in non-equilibrium MD is the same. I believe the mindset should be: let’s assess how large is the mistake made by assuming equilibrum conditions and whether it turns out to be negligible or not.

Cheers,
Michele

​Hi Michele,

Thanks for your comments and clarification on this issue. I quite agree with you. The thermal effect brought by the high frequency field is what I should be very careful with in my case. Hopefully, if the introduced light/field energy is kept low enough, maybe the heat being taken away by the thermoststs wouldn’t matter much to the system. I will have a good look at the paper you recommended. Thanks.

Cheers,
Eva

Hi Eva,

I realized there are other two possibly useful observations to make:

  1. Since the E field you are imposing is oscillatory, testing a few different t-coupling or p-coupling time constants can also be important (think for example at the two ‘limit’ cases, either when the relaxation frequency is much larger than the oscillation frequency or when it’s much smaller).

  2. It may be beneficial to t/p-couple only one part of the system (possibly restrained), letting it act on the rest as a heat bath; this may prevent artificial thermal effects induced by esplicit relaxation, but you also want to check if the heat transfer between the ‘bath’ and the rest of the system is large enough to ensure the whole system is actually isothermal.

​Hi Michele,

Great ideas! Thanks for your suggestions. I will definitely check those.

Cheers,
Eva