[gmx-users] specific heat calculation using "gmx dos"

[Christian Blau]

Hi Pragati,


Answers in short:

1. It's only c_v, you'll have to add VT \alpha_T^2/ \kappa_T to obtain c_p

2. I suggest you use the approach in the paper, running an NVT simulation and base all on the c_v estimate.

3. To check convergence, I suggest running the analysis again on shorter parts of your trajectory (1/2 of the data, 1/4 
of the data etc.)

4. You'll not see the quantum correction with this approach, your force-field might rely on entropy- enthalpy 
compensation. I would guess that you would obtain worse results with such an approach.


The DOS tool follows the supplementary material S(1)-S(17) in "Force Field Benchmark of Organic Liquids: Density, 
Enthalpy of Vaporization, Heat Capacities, Surface Tension, Isothermal Compressibility, Volumetric Expansion 
Coefficient, and Dielectric Constant" by

Carl Caleman, Paul J. van Maaren, Minyan Hong, Jochen S. Hub, Luciano T. Costa, David van der Spoel. It's output "Heat 
capacity ... J/mol K" corresponds to equation (14) in above publication, the constant volume heat capacity c_v *with* 
quantum correction. To obtain c_p you would still need to apply (eqn 12).


Best,

Christian


On 2020-01-03 07:33, Pragati Sharma wrote:
>   Dear all,
>
> I am trying to calculate heat capacity of a polymeric system (30k atoms)
> using gromacs.
> I tried the classical way from the fluctuation dependent properties using
> "gmx energy", but the values are not matching experimental values.
>
> I want to use "gmx dos" program, but I have some confusions.
>
> 1. Does it gives only the quantum corrections in heat capacity, which
> should be added to the one calculated from 'gmx energy' or  Does it gives
> full heat capacity value including quantum corrections.
>
> 2. Should I have to run NVT simulations, run gmx dos, find Cv and the add
> then calculate Cp, as given in the paper ( J. Chem. Theory Comput. 2012, 8,
> 61–74), *OR*
> I can run a NPT simulation and calculate Cp from gmx dos.
>
> 3. gmx dos gives the Cp/Cv value in "dos.log" file. but is there any output
> file to check its convergence. I mean how should I decide, how long the
> trajectory I should take.
>
> 4. Cp can be calculated from the slope of enthalpy vs temp plot (dH/dT) (in
> a small range of temp), what limitation does this approach have.
>
> Thanks in advance.