[gmx-users] Use NVT to mimic NVE

[David van der Spoel]

On 2014-10-12 17:31, Dr. Vitaly Chaban wrote:
> The artifacts are caused by energy leakage. Thermostatting does not
> solve them. The problem exists, but you do not see it clearly with
> thermostat. For instance, I have an example where a liquid droplet in
> vacuum is not stable due to bad energy conservation. Some particles
> just sporadically fly away over the course of the simulation.
We have run 3 microsecond simulations of water droplets in vacuum with 
perfect energy conservation, double precision, no cut-off and SETTLE.
With constraint solvers it is indeed close to impossible to get real 
energy conservation.

Turning off the constraints and using a time step <= 0.5 fs might work 
somewhat better with mixed precision, and these degrees of freedom may 
be important for energy transfer too.


>
> NVE generally does not like constraints. If I remember correctly from
> the previous messages, you use constraints. Novel performance
> improvements in gromacs (accumulated after 4.0 and especially after
> 4.5) sometimes compromise accuracy, though it is a long discussion
> itself. What about downgrading to gmx 4.0 or gmx 3.3, after all? The
> very last time I used gromacs for NVE simulation was in 2009. I guess
> I used gmx 3.3.3 (but maybe 4.0.x). There were no problems for sure.
>
> I would also give a try to some very simple system (e.g. liquid argon)
> to observe what happens to total energy.
>
> Good luck with troubleshooting!
>
>
> Dr. Vitaly V. Chaban
>
> Виталий Витальевич ЧАБАН
>
>
> On Sun, Oct 12, 2014 at 4:53 PM, Johnny Lu <johnny.lu128 at gmail.com> wrote:
>>   Firstly, what are the possible errors/artifacts caused by masking "the
>> energy leakage problem with the thermostat" in this case?
>>
>> Secondly, are there any other mdp setting that I can try to make the mixed
>> precision gromacs NVE run have conserved energy over 100 ns ? (and... is it
>> because of some errors that I made in my systems that make the NVE run not
>> conserved in energy?)
>>
>> The NVE has the following statistics.
>>
>> Statistics over 2747281 steps [ 0.0000 through 1373.6400 ps ], 4 data sets
>> All statistics are over 274729 points
>>
>> Energy                      Average   Err.Est.       RMSD  Tot-Drift
>> -------------------------------------------------------------------------------
>> Disper. corr.              -3961.13          0          0 2.60178e-09
>> (kJ/mol)
>> Total Energy                -309939        160     337.76    478.828
>> (kJ/mol)
>> Temperature                 300.292       0.24    1.49289   0.631428  (K)
>> Pressure                   -64.5409        4.7    148.779    25.1326  (bar)
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se