AMBER Archive (2005)

Subject: Re: AMBER: REM error

From: Carlos Simmerling (carlos_at_ilion.bio.sunysb.edu)
Date: Tue Sep 13 2005 - 12:40:20 CDT


I apologize for the manual not being more clear on this, we are working on
a tutorial.

Anthony Cruz wrote:

>Ok.
>I need to run a short equilibration MD first for all the replicas or just one
>and the use this structure for all the replicas???
>
run short MD (even a few steps is enough, just to get velocities) for
any structure that you want to use for the initial structure. using the
same inpcrd for all replicas is ok, but some people prefer to use
different structures for each. that's up to you.

>This little run at what
>temperature will be if the lower temperature of the replica is 330k, (ex.
>from 0 to 298 for 200ps)???
>
it can be any T, the replicas will equilibrate during the simulation
anyway. Just like normal MD, you will want to discard some initial
period of REMD simulation as equilibration.

>Then I need to set the tempi of all replicas equal
>to the temperature of this littler run ???
>
>
tempi won't be used since it is ignored if ntx=5 and irest=1. just set
temp0 for the replicas.
make sure that the input files are identical except for temp0 - we use a
script to create the mdin files.

>Anthony
>On Tuesday 13 September 2005 5:25 am, Carlos Simmerling wrote:
>
>
>>Anthony Cruz wrote:
>>
>>
>>>Thank you for your answer.
>>>I find an error in the *.in files, :
>>>I have this irest = 1, ntx = 5, and sander cand find the velocities so
>>>I change it to irest = 0, ntx = 1, because I start all the replicas
>>>with the same minimized structure.
>>>
>>>
>>you need to use irest=1, ntx=5. Amber9 will have more
>>extensive checking of REMD inputs. Just run a short equilibration MD first.
>>The reason is that after each exchange the input is re-used and
>>you need to have all of the data read from the restart file.
>>carlos
>>
>>
>>
>
>
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu