Re: [ESPResSo] Units query for tabulated potential - clarification and assistance requested

[Owen Hickey]

Hey,

You don't have to alter Espresso source code to use reasonable SI values in your simulation.  If you want distances to be in nanometers and energies to be in kilojoules (just divide the energy by Avagadro's number to go from kJ/mol to kJ) just enter them as numbers.  As long as all the distances and energies you input are in those units you shouldn't have a problem.  Same with the mass.  As for the temperature when you specify it using one of the thermostats it is important to realize the value you feed into the thermostat is kT and not simply temperature.  So you could make some tcl code
set T 300
set kT [expr $T*1.3806503e-23]
thermostat langevin $kT 1.0
should give you a temperature in Kelvin using the Langevin thermostat.

I assume you also want your time in seconds.  In order to figure out how long a time step is simply convert it from the other units:
kJ=1.661e-27kg*nm^2/t^2
t=sqrt(1.661e-27kg*nm^2/1e3J)
t=sqrt(1.661e-27*1e-18/1e3)s
t=1.289e-24s
Which is to say 1 time unit in your MD simulation would be 1.289e-24 seconds...  I think I may have made a mistake somewhere (perhaps inputting energies in kJ/mol makes more sense as you would probably get a more reasonable time step)...  Anyways I hope that at least helps.  Best of luck.

Owen

On Fri, Oct 2, 2009 at 10:52 PM, Steven Kirk <address@hidden> wrote:

Hello,

Currently running Espresso 2.1.2j.

I would greatly appreciate some help from the members of this list on the following question regarding units for a simulation involving identical particles interacting using only a non-bonded 2-body interaction, which is specified in tabulated form.

If I decide that I want that all distances are to be specified to Espresso are in nanometres, all energies in kJ/mol, all times in picoseconds, and all masses in amu (this is the default set of units used in GROMACS, another MD code), then what factors do I need to adjust and set so that I can also, say, define temperature in degrees K? Do I have to hack into the source code to, for example, redefine the Boltzmann constant, and if so, what value should it take?

I would GREATLY appreciate it if somebody could advise me on generating such a self-consistent complete SI-based set of units that I could use for simulations with Espresso, instead of relying on LJ or arbitrary dimensionless units, which are of no use to me. The documentation, which I have read repeatedly, gives no such example of a set of self-consistent units defined in this way.

Very many thanks in advance if you can help!

Steve Kirk

--
Dr. Steven R. Kirk           <address@hidden, address@hidden>
Dept. of Technology, Mathematics & Computer Science  (P)+46 520 223215
University West                                      (F)+46 520 223299
Trollhattan 461 86 SWEDEN                     http://beacon.webhop.org

_______________________________________________
ESPResSo mailing list
address@hidden
https://fias.uni-frankfurt.de/mailman/listinfo/espresso