Re: [AMBER] Multi-Dimensional REMD: rem.log shows exchanges along Hamiltonian dimension, but .mdout files do not

From: Jason Ku Wang <jwang198.stanford.edu>
Date: Fri, 27 Apr 2018 10:54:23 -0700

Hi Vinícius,

Thank you so much for the reply!

I'm currently basing the Hamiltonian dimension on different parameters of
accelerated MD. It makes sense to exchange coordinates and velocities when
the Hamiltonian dimension is based off of varying topology files.

But in the case of aMD-based Hamiltonian exchange, where only the input
parameters differ (much like temperature-REMD), does this also hold?

Thanks for the clarification - I just want to make sure I understand the
implementation!

Best,
Jason


On Wed, Apr 25, 2018 at 6:29 PM, Cruzeiro,Vinicius Wilian D <
vwcruzeiro.ufl.edu> wrote:

> Hello Jason,
>
> In T-REMD, a given replica will have different target temperatures
> throughout the simulation. Therefore, when you sort all replicas to filter
> only the trajectories corresponding to a given temperature, you will get
> chunks of data from many replicas (possibly from all if your exchange rate
> is good and you run long enough).
>
> In H-REMD, a given replica exchanges coordinates and velocities when an
> exchange attempt is accepted, but the replica index will remain the same
> until the end of the simulation.
>
> Therefore, for your T,H-REMD simulation, the fact that the hamiltonian
> index is not changing but the temperature index is actually is then
> expected. If you are obtaining good exchange rates in the rem.log files,
> then everything should be fine.
>
> I hope this helps,
> Best,
>
>
> Vinícius Wilian D Cruzeiro
>
> PhD Candidate
> Department of Chemistry, Physical Chemistry Division
> University of Florida, United States
>
> Voice: +1(352)846-1633<tel:+1(352)846-1633>
>
> On Apr 25, 2018, at 8:56 PM, Jason Ku Wang <jwang198.stanford.edu<mailto:
> jwang198.stanford.edu>> wrote:
>
> Hi AMBER Community,
>
> I am currently running multi-dimensional REMD using a temperature and
> Hamiltonian dimension (my Hamiltonian dimension is defined by varying
> parameters for accelerated MD applied only to torsional potential
> energies). To see if exchanges are occurring at a reasonable rate along
> either dimension, I examine the rem.log files.
>
> Looking just at rem.log files, it appears that both temperature and
> Hamiltonian dimensions are mixing well. In particular, we see *very clear
> evidence* of exchanges occurring along the Hamiltonian dimension (note the
> success and success rate columns):
>
> # Rep#, Neibr#, Temp0, PotE(x_1), PotE(x_2), left_fe, right_fe, Success,
> Success rate (i,i+1)
> # exchange 64 REMD group 45
> 32 33 302.97-461315.96-460510.32 6.77 -6.77 T
> 0.06
> 10 30 302.97-460764.81-460769.68 6.46 -6.50 F
> 0.06
> 34 37 302.97-460835.06-460731.85 6.89 -6.72 T
> 0.12
> 1 10 302.97-460532.32-461501.22 -285.50 -6.32 F
> 0.06
> 17 16 302.97-461139.12-461040.58 6.59 -6.46 T
> 0.06
> 2 15 302.97-461393.48-461490.99 6.19 -6.38 T
> 0.12
> 26 20 302.97-461362.27-460938.11 6.70 -6.60 T
> 0.12
> ...
>
> However, when I attempt to examine *.mdout files*, I see no evidence of
> exchange. If I look at the GROUP INDEX corresponding to the Hamiltonian
> dimension, I see that it stays constant across every step. For example, for
> any given replica, the first number corresponding to the Hamiltonian value
> never changes, but the second number corresponding to temperature does
> change. Yet, both Hamiltonian and temperature dimensions show reasonable
> exchange rates in rem.log files.
> ...
> MULTI-D REMD. GROUP INDEXES: * 1* 2
> ...
>
> I checked this systematically for all my replicas and my frequency of
> printing is the same as my frequency of exchange.
>
> *Is this a potential bug? Why would the Hamiltonian group printed out in
> the .mdout file remain unchanging even though the rem.log shows
> Hamiltonians being exchanged?*
>
> Thanks so much for the help!
> Jason
>
> --
> Jason Wang
> Stanford University, Class of 2018
> jwang198.stanford.edu<mailto:jwang198.stanford.edu>
> ᐧ
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org<mailto:AMBER.ambermd.org>
> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.
> ambermd.org_mailman_listinfo_amber&d=DwIGaQ&c=
> pZJPUDQ3SB9JplYbifm4nt2lEVG5pWx2KikqINpWlZM&r=LIQu8OlVNKmzfbMg9_5FnKrt9-
> DrdQBJXyFyocKAWXc&m=kVgrxuNP7GP6yjJQ4lPBhNLKuU--W7lYRcYAmFXunbM&s=
> RqIiaqLJCVeQXouvLlv5I0_YIhJrdCwE0YHzmNQNMLU&e=
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>



-- 
Jason Wang
Stanford University, Class of 2018
jwang198.stanford.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 27 2018 - 11:00:02 PDT
Custom Search