[gmx-users] Heat capacity calculations from fluctuations and using g_dos

[David van der Spoel]

Den 2018-01-10 kl. 09:04, skrev Srinath Kompella:
> Hello all,
>                I've simulated Methylcyclohexane (MCH) in NVT and NPT
> ensembles using GROMACS-5.1.4, from which I want to calculate the heat
> capacities (Cv and Cp respectively) of MCH. If I use the fluctuation
> properties, I get the following note: "Heat capacities obtained from
> fluctuations do *not* include quantum corrections. If you want to get a
> more accurate estimate, please use the g_dos program." After going through
> a few discussions on GROMACS mailing lists, I used "gmx dos" module
> thinking that I would get the quantum corrections for the heat capacities.
> But the heat capacity values written at the end of the "dos.log" are larger
> than that of those calculated using fluctuation properties.
> 
> *Temperature(K)                             Cv (J/mol K)
>                            Cp (J/mol K)              *
>                                  *Fluctuation Properties          g_dos  *
>          *Fluctuation Properties        g_dos  *
>         298.15                            165.564                 275.747
>                   193.939                  274.531
> 
>                I noticed that the numbers fluctuation properties give are
> relatively much closer to the experimental values.
> 
>                Please help me understand why the numbers are very different
> and which method to go for. Thank you in advance.
> 
Correspondence to experiment is not a way to check your results.

In order for the corrections to work you need to perform a 100 ps NVT 
simulation saving velocities and positions every 4-5 fs and without 
using constraints.

-- 
David van der Spoel, Ph.D., Professor of Biology
Head of Department, Cell & Molecular Biology, Uppsala University.
Box 596, SE-75124 Uppsala, Sweden. Phone: +46184714205.
http://www.icm.uu.se