- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Jason Swails <jason.swails.gmail.com>

Date: Sun, 5 Oct 2014 22:24:56 -0400

On Oct 5, 2014, at 1:21 PM, CHAMI F. <fatima.chami.durham.ac.uk> wrote:

*> Dear all
*

*>
*

*>
*

*>
*

*> in order to calculate the heat of evaporation I run a gas and bulk phase of a given solute eg. Heptane
*

*>
*

*> using the same thermostat conditions
*

*>
*

*> the RMSF of bulk phase (NPT) ( 1000 MOLECULES ) is 1.15 degrees
*

*>
*

*> the RMSF of gas phase ( 1 molecule) is 50 degress
*

Sounds about right. As you will see just about everywhere in statistical mechanics, fluctuations scale as 1/sqrt(N), where N is the number of particles.

Take a “particle” to be a “molecule”, and you can estimate that 1000 molecules will have the fluctuations ~1/sqrt(1000) as large as 1 molecule. Take 50 degrees and divide that by sqrt(1000) and you get 1.58 degrees… not too far off of your values.

*> is this normal or do I have to adjust the friction coefficient gamma_ln for the gas phase
*

gamma_ln really shouldn’t affect equilibrium distributions of thermodynamic properties, assuming that you’re sampling the ‘correct’ distribution with each simulation. I don’t think that changing gamma_ln will change your temperature fluctuations at all (assuming that both simulations are converged in that property, of course).

HTH,

Jason

Date: Sun, 5 Oct 2014 22:24:56 -0400

On Oct 5, 2014, at 1:21 PM, CHAMI F. <fatima.chami.durham.ac.uk> wrote:

Sounds about right. As you will see just about everywhere in statistical mechanics, fluctuations scale as 1/sqrt(N), where N is the number of particles.

Take a “particle” to be a “molecule”, and you can estimate that 1000 molecules will have the fluctuations ~1/sqrt(1000) as large as 1 molecule. Take 50 degrees and divide that by sqrt(1000) and you get 1.58 degrees… not too far off of your values.

gamma_ln really shouldn’t affect equilibrium distributions of thermodynamic properties, assuming that you’re sampling the ‘correct’ distribution with each simulation. I don’t think that changing gamma_ln will change your temperature fluctuations at all (assuming that both simulations are converged in that property, of course).

HTH,

Jason

-- Jason M. Swails BioMaPS, Rutgers University Postdoctoral Researcher _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Sun Oct 05 2014 - 19:30:02 PDT

Custom Search