Re: [AMBER] iwrap=1 or netcdf

From: Marek Maly <marek.maly.ujep.cz>
Date: Thu, 16 Jan 2014 18:43:46 +0100

ntxo=2 should be OK (ntxo=0 is NOT recommended in manual)

see here

http://archive.ambermd.org/201301/0382.html

I am using:

iwrap=0,ioutfm=1,ntxo=2,

without any issues.


One more "bonus" tip for you :))

It is a good idea to set explicitly
ntwr=50000 or to similar number otherwise
restart file will be by default written
each 500 steps which will slow down
your calculation.

  Best,

     M.




Dne Thu, 16 Jan 2014 18:01:23 +0100 Jio M <jiomm.yahoo.com> napsal/-a:

> Hi Marek, Justin, Daniel
>
>
> manual says not recommended.
>
> ntx0 = 0 Unformatted (no longer recommended or allowed: please use
> formatted restart files)
>
> I run job in loops so need restart files for every loop
>
>
> What about using, as Daniel suggested, now with iwrap=1,ioutfm=1 then
> using unwrap with ptraj for analysis?
>
> thanks,
>
>
>
>
> On Thursday, January 16, 2014 4:52 PM, Marek Maly <marek.maly.ujep.cz>
> wrote:
> iwrap=0,ioutfm=1,ntxo=2,
>
> is combination you are looking for.
>
> If I am not wrong "ioutfm=1" ensures
> NetCDF format just for trajectory files not
> for the restart files for which ntxo
> flag should be set to value 2.
>
> Best,
>
> M.
>
>
>
> Dne Thu, 16 Jan 2014 10:46:20 +0100 Jio M <jiomm.yahoo.com> napsal/-a:
>
>> Hi
>>
>> Even I used ioutfm=1 for netcdf I got **** in restart and job killed. Is
>> that normal?So should I use iwrap=1 also?
>>
>> P.S. I will come to Marek's point later
>> thanks
>>
>>
>>
>> On Wednesday, January 15, 2014 5:18 PM, Marek Maly <marek.maly.ujep.cz>
>> wrote:
>> Hi Jio,
>>
>> actually I did no diffusion study yet :))
>>
>> But if you do not set nscm parameter, by default
>> the COM movement corrections will be done each
>> 1000 steps (see the Amber12 manual).
>>
>> See my last comments regarding my view (understanding) of that
>> corrections in the framework of the diffusion studies.
>>
>> Perhaps much more important is to know that different
>> water models (TIP3P, TIP4P, SPC ...) have also remarkable differences
>> comparing to real (experimentally measured) diffusion which is connected
>> with just limited ability of those models to imitate some real water
>> parameters here e.g. water viscosity.
>>
>> So especially if you are going to compare your simulation results
>> with experiments you should to take this in account.
>>
>> There will be perhaps some articles available describing (quantifying)
>> this
>> differences for the given water model.
>>
>> You can perhaps eventually do your own "calibration" experiments by
>> comparing
>> experimentally measured water self-diffusion with that obtained from the
>> simulation using the given water model under the same thermodynamic
>> conditions
>> as in experiment.
>>
>> Best wishes,
>>
>> Marek
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> Dne Wed, 15 Jan 2014 17:28:49 +0100 Jio M <jiomm.yahoo.com> napsal/-a:
>>
>>> Dear Marek
>>>
>>> what value of nscm flag you use for trajectories for diffusion
>>> analysis?
>>> just for curiousity did you observe any difference ?
>>>
>>> thanks
>>>
>>>
>>>
>>>
>>>
>>> On Wednesday, January 15, 2014 4:12 PM, Jason Swails
>>> <jason.swails.gmail.com> wrote:
>>> On Tue, Jan 14, 2014 at 11:35 AM, Marek Maly <marek.maly.ujep.cz>
>>> wrote:
>>>
>>>> Hello,
>>>>
>>>> regarding simulations for the diffusion study should one
>>>> care about the
>>>>
>>>> nscm flag ( Flag for the removal of translational and rotational
>>>> center-of-mass (COM) motion ) ?
>>>>
>>>>
>>>> ---------------------------------------------------------------------------------------
>>>> "...For Langevin dynamics, the position of the center-of-mass of the
>>>> molecule is reset
>>>> to zero every NSCM steps, but the velocities are not affected. ...."
>>>>
>>>> "...The only reason to even reset the coordinates is to prevent the
>>>> molecule from diffusing so far away from the origin that its
>>>> coordinates
>>>> overflow the format used in restart or trajectory files. ..."
>>>>
>>>> -----------------------------------------------------------------------------------------
>>>>
>>>> If I understand well this flag is here e.g. to compensate perhaps very
>>>> small (especially for bigger systems) but nonzero translational
>>>> velocity
>>>> of the center of mass of the simulation box content as the result of
>>>> the
>>>> random assigning of the initial atom velocities. Am I right ?
>>>>
>>>
>>> As I understand it, the flag was originally introduced to solve the
>>> 'flying
>>> block of ice' effect introduced by the Berendsen thermostat on
>>> explicitly
>>> solvated systems (it would basically convert internal vibrational
>>> energy
>>> into rapid translational energy as a consequence of the algorithm
>>> itself).
>>>
>>> It does serve the purpose you describe as well, but stochastic
>>> thermostats
>>> should not have this problem since the _average_ translational momenta
>>> will
>>> be zero. Even if translational momentum is not exactly zero after each
>>> random 'hit', the average over a long trajectory will be zero. It's
>>> also
>>> unnecessary for NVE simulations (that display good energy
>>> conservation),
>>> since the law of conservation of momentum holds for Newtonian dynamics.
>>>
>>> HTH,
>>> Jason
>>>
>>
>>
>
>


-- 
Tato zpráva byla vytvořena převratným poštovním klientem Opery:  
http://www.opera.com/mail/
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 16 2014 - 10:00:03 PST
Custom Search