Re: [AMBER] Setting the random seed in Langevin dynamics

From: Adrian Roitberg <roitberg.ufl.edu>
Date: Wed, 03 Oct 2012 09:25:00 -0400

Hi Sajeewa

I really recommend that you upgrade to amber 12 ! The number of changes
and improvements are too many to pass on.

You can set ig to whatever you want, as long as it is different every time.

We have scripts to start or restart runs with ig=1 .... N

They all generate independent random streams, uncorrelated. We tested
this a lot. It makes no difference if you choose ig =1 or 235671 or if
the igs are consecutive or not.

Adrian


On 10/3/12 9:19 AM, Sajeewa Pemasinghe wrote:
> Thank you Robi. The problem is I am using Amber10 so I don't think ig=-1
> option is available, that is why I was asking about the specifics of the
> value for ig.
>
> Sajeewa
>
> On Wed, Oct 3, 2012 at 5:54 AM, Deák Robert <kokumetto.gmail.com> wrote:
>
>> Hi,
>>
>> If you don't want to give the value of ig for every individual run, with
>> "ig=-1" you can set up randomly the seed, calculated from system wallclock.
>>
>> Best,
>> Robi
>>
>> 2012/10/3 Sajeewa Pemasinghe <sajeewasp.gmail.com>
>>
>>> Hi all,
>>>
>>> I am running MD of a protein for 20ns. I am going to use Langevin
>> dynamics
>>> to fix the temperature. "To avoid synchronization artifacts I have to
>> set
>>> a different value for the random seed at every restart." I want to get
>> the
>>> above statement clarified a little more. Does that specifically mean each
>>> time my input file says "irest=1", I have to give a different number for
>>> "ig". If so,about the number I give to "ig" does it have a minimum and a
>>> maximum number of digits or can I just give any number as long as it was
>>> not repeated previously?
>>>
>>> Thank you
>>>
>>> Sajeewa Dewage
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

-- 
                            Dr. Adrian E. Roitberg
                                  Professor
                Quantum Theory Project, Department of Chemistry
                            University of Florida
                              roitberg.ufl.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 03 2012 - 06:30:03 PDT
Custom Search