[gmx-users] box size changing during isotropic pressure coupling

[Berk Hess]

Hi,

Hmm... this is a (seemingly( really systematic decrease in x and y and increase in z.

But is x at the start really EXACTLY identical to y?
Because in that case I would think they should always stay identical.
But looking at the code I see that it is always uses triclinic scaling.
Thus x and y are treated differently (although adding only zeros),
so depending on the compiler, the results could differ by a bit.
What system and compiler are you using?

Berk.


> Date: Fri, 18 Jul 2008 13:54:19 -0400
> From: chris.neale at utoronto.ca
> To: gmx-users at gromacs.org
> Subject: [gmx-users] box size changing during isotropic pressure coupling
> 
> Hello,
> 
> I have a system of 750,000 atoms consisting of detergent in water. 
> Although this run has been carried out using isotropic pressure 
> coupling, I see significant drift in the box dimensions over 150 ns. 
> This run was carried out using gromacs version 3.3 and 3.3.1. I used the 
> -shuffle and -sort options and an in-house g_desort. This means that my 
> restarts are via grompp, not tpbconv, although I do use the .trr and 
> .edr with gen_vel=no and unconstrained-start=yes for a proper restart.
> 
> My system is rectangular, starting with an x=y!=z and a=b=c=90. When I 
> view the .xtc file, it is clear that the system is shrinking in the x/y 
> and growing in the z. However, while turning on PBC images it looks like 
> a totally normal simulation in that there is no net drift or anything 
> else obviously problematic.
> 
> I have posted some images here 
> http://www.pomeslab.com/CN_GromacsPost_071808.html that show the box 
> dimensions over time and also the total volume. The initial rapid volume 
> change, and its asymptotic continuation, is an aggregation effect and is 
> expected. Other than this, the volume is generally stable to the changes 
> in the box lengths along single dimensions. I can not explain any of the 
> features of the individual x/y/z plots, nor can I explain why they level 
> out, why y levels out before x and z, or guess if this will continue to 
> stay (relatively) flat.
> 
> I am interested to know if anybody else has experienced this. Also, it 
> seems to me like this is simply an output format thing and will not 
> cause problems unless the x/y shrinks so much that cutoff effects cause 
> me problems.
> 
> my .mdp file follows.
> 
> title               =  seriousMD
> cpp                 =  cpp
> integrator          =  md
> nsteps              =  25000
> tinit               =  153200
> dt                  =  0.002
> comm_mode           =  linear
> nstcomm             =  1
> comm_grps           =  System
> nstxout             =  25000
> nstvout             =  25000
> nstfout             =  25000
> nstlog              =  5000
> nstlist             =  10
> nstenergy           =  500
> nstxtcout           =  5000
> ns_type             =  grid
> pbc                 =  xyz
> coulombtype         =  PME
> rcoulomb            =  0.9
> fourierspacing      =  0.12
> pme_order           =  4
> vdwtype             =  cut-off
> rvdw_switch         =  0
> rvdw                =  1.4
> rlist               =  0.9
> DispCorr            =  no
> Pcoupl              =  Berendsen
> pcoupltype          =  isotropic
> compressibility     =  4.5e-5
> ref_p               =  1.
> tau_p               =  4.0
> tcoupl              =  Berendsen
> tc_grps             =  DPC     SOL
> tau_t               =  0.1     0.1
> ref_t               =  300.    300.
> annealing           =  no
> gen_vel             =  no
> unconstrained-start =  yes
> gen_temp            =  300.
> gen_seed            =  9896
> constraints         =  all-bonds
> constraint_algorithm=  lincs
> lincs-iter          =  1
> lincs-order         =  4
> ;EOF
> 
> #########
> 
> many thanks,
> Chris.
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080724/3da73820/attachment.html>