[gmx-users] vacuum right but melt collapse

Tsjerk Wassenaar tsjerkw at gmail.com
Fri Mar 9 11:38:07 CET 2007 

Hi Kitty,

Your english is quite ok.., no need to worry about that too much.
Earlier remarks to someone else also had to do more with poor grammar
structure, making it very hard to follow the question.

First, if I didn't make it clear enough, you have to find the papers
from Jorgenson about the OPLS force field.

Second, the thing with parameterization of force fields is this: a set
of parameters is derived to obtain a good fit of the simulation
behaviour to experimentally or ab initio derived data. If a force
field is parameterized taking only interactions into account within a
sphere of a certain radius, than the fit between the behaviour
obtained with these parameters and the desired behaviour will worsen
if each particle suddenly "sees" particles it wasn't supposed to see.
This particularly applies to C6/C12 Lennard-Jones parameters.
Now if it doesn't become clear to you, please read about force fields
and force field development. That is a vital part of molecular
dynamics simulations, and I wouldn't recommend doing simulations, if
you don't have a basic understanding of these things.

Although highly relevant to your work, I do think that there more
fundamental problems underlying your initial question posted. Changing
the cut-off shouldn't really make the difference between explosion and
proper behaviour.

Cheers,

Tsjerk


On 3/9/07, kitty ji <jiqing at iccas.ac.cn> wrote:
>
>
> Hi Wassenaar:
>
> As mensioned in GMX manual3.3 , the parameter of OPLS force field is listed
> bellow:
>
> Name             FUNCTION
> bond              1
> pair              1
> angle             1
> dihedral          3
> nb                1
> Col               3
>
> In the manual,4.6.3 Treatment of Cutoffs,it is said that "In the neighbor
> list all
> interaction pairs that fall within rlist are stored. Furthermore, the
> interactions between pairs
> that do not fall within rlist but do fall within max(rcoulomb,rvdw) are
> computed during
> NS, and the forces and energy are stored separately, and added to
> short-range forces at every time
> step between successive NS."
>
> From the paragraph, I concluded that when the distance of cutoff is
> increased, the further interaction pair will be considered. So the presion
> will be increased also.
>
> I modified the cutoff from 1.5 nm to 1.2 nm.Till now the system is not
> exploding. Whatever 1.5 nm or 1.2 nm, it has been proved that 0.9 nm is too
> short to simulating with OPLS.
>
> By the way, where do you find the critical value 1.2 nm. I can not found it
> by google. But there is really a lot of works used that values. Thank you
> very much.
>
> For the IE format, these mail was copied from WORD. Because I'm not
> confident with my English.:)
>
>
> Message: 7
> Date: Fri, 9 Mar 2007 10:07:47 +0100
> From: "Tsjerk Wassenaar" <tsjerkw at gmail.com>
> Subject: Re: Re: [gmx-users] vacuum right but melt collapse
> To: "kitty ji" <jiqing at iccas.ac.cn>, "Discussion list for GROMACS
> users" <gmx-users at gromacs.org>
> Message-ID:
> <8ff898150703090107i264e57cdt4ba6647115d284e3 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
>
> Hi Kitty,
>
> OPLS was developed (IIRC) for use with a cut-off of 1.2 nm. Not
> shorter, not longer. If you set the cut-off higher, you introduce
> additional force components; noise, artefacts, not precision. It may
> be good to do some reading about force fields and the parameterization
> thereof.
>
> The reaction field is a correction for the neglect of electrostatic
> interaction beyond the cut off. It should be used when using a GROMOS
> force field, as those are parameterized for use with it. Regarding
> OPLS, I'm not sure. I thought it was not parameterized for use with
> RF, but either a plain cut-off or a cut-off with a switch function.
> You should check the OPLS papers to be sure.
>
> By the way, please keep discussions on the list... In that way, it is
> publicly archived, which potentially saves us a lot of trouble
> answering the same questions over and over.
>
> Cheers,
>
> Tsjerk
>
> On 3/9/07, kitty ji <jiqing at iccas.ac.cn> wrote:
> > Thanks for your reply.
> >
> > The force field I used here is OPLS. Increasing the cutoff means to
> increase the precision of caluculation not "underlying cause". Therefore it
> is acceptable. In addition, it is said that cutoff 9 angstrom is not enough
> for electricity calculation.
> >
> > what is using reaction-field? I thought it was not used here.
> >
> >
> > ----- Original Message -----
> > From: "Tsjerk Wassenaar" <tsjerkw at gmail.com>
> > To: "kitty ji" <jiqing at iccas.ac.cn>; "Discussion list for GROMACS users"
> <gmx-users at gromacs.org>
> > Sent: Thursday, March 08, 2007 10:34 PM
> > Subject: Re: Re: [gmx-users] vacuum right but melt collapse
> >
> >
> > Hi Kitty,
> >
> > Such a solution is unsatisfactory, as it seems that you're masking an
> > underlying cause. Now, I'm not sure of your other parameters, but when
> > using a cut-off (do you use reaction-field) you would usually (gromos
> > force field) use a twin-range of 0.9/1.4, which is what the force
> > field was parameterized for.
> >
> > If you look at the energies in order to find the cause of a crash, you
> > want to look for the energy term first displaying strange behaviour.
> > You also may want to try setting the neighbour list update to every
> > step.
> >
> > And please, don't use html formatted e-mails... :)
> >
> > Hope it helps,
> >
> > Tsjerk
> >
> >
> > On 3/8/07, kitty ji <jiqing at iccas.ac.cn> wrote:
> > >
> > >
> > >
> > >
> > > Tsjerk Wassenaar, thank you very much for you suggestion even if most of
> > > them have been manipulated.
> > >
> > >
> > >
> > > "Write out the energies at every step and check which of the energy
> > > contributions goes astray. This may give you a good clue to where
> > > you're mistake is. Also, run through the topology again, by hand,
> > > checking all bonds, angles, dihedrals, exclusion (implicit ones!),
> > > etc."
> > >
> > > when the explosion happened, the energy of bond, angle and potential
> will
> > > were increase. And the topology has been checked several times.
> > >
> > >
> > >
> > >
> > >
> > > Now I increase the cutoff of VDW & Coulomb from 0.9 to 1.5. The system
> looks
> > > ok from last night to this time. If the problem is fixed, I conclude
> that
> > > cutoff 0.9 is not enough for electric system.
> > >
> > >
> > >
> > > Hope this conclusion is helpful for GMX users.
> > >
> > >
> > >
> > > Message: 2
> > > Date: Wed, 7 Mar 2007 17:49:30 +0100
> > > From: "Tsjerk Wassenaar" <tsjerkw at gmail.com>
> > > Subject: Re: Re: [gmx-users] vacuum right but melt collapse
> > > To: "kitty ji" <jiqing at iccas.ac.cn>, "Discussion list for GROMACS
> > > users" <gmx-users at gromacs.org>
> > > Message-ID:
> > >
> <8ff898150703070849r59cb8a8dr60b894a881fcc331 at mail.gmail.com>
> > > Content-Type: text/plain; charset=GB2312; format=flowed
> > >
> > >
> > > Hi Ji Qing,
> > >
> > > Write out the energies at every step and check which of the energy
> > > contributions goes astray. This may give you a good clue to where
> > > you're mistake is. Also, run through the topology again, by hand,
> > > checking all bonds, angles, dihedrals, exclusion (implicit ones!),
> > > etc.
> > >
> > > Good luck,
> > >
> > > Tsjerk
> > >
> > > On 3/7/07, kitty ji <jiqing at iccas.ac.cn> wrote:
> > >
> > > >
> > > >
> > > > Thanks Mark.
> > > > But when one polymer chain was used, non-bonded interaction has
> already
> > > been
> > > > there. Because one chain means C44H156O22.
> > > > After all, I'd like to try more chains.
> > > >
> > > > Message: 1
> > > > Date: Tue, 06 Mar 2007 17:01:52 +1100
> > > > From: Mark Abraham <Mark.Abraham at anu.edu.au>
> > >
> > > > Subject: Re: [gmx-users] vacuum right but melt collapse
> > > > To: Discussion list for GROMACS users <gmx-users at gromacs.org>
> > > > Message-ID: <45ED03D0.6010803 at anu.edu.au>
> > >
> > > > Content-Type: text/plain; charset=GB2312
> > > >
> > > > kitty ji wrote:
> > > > > Hi GMX user:
> > > > >
> > > > > A model for Poly(vinyl methyl ether) was built with OPLS force field
> (
> > > > > all H ).
> > > > >
> > > > >
> > > > >
> > > > > When the model was run alone (a single chain) in vacuum, anything
> goes
> > > ok.
> > > > >
> > > > >
> > > > >
> > > > > But with 45 chains box, the system will be collapse with a little
> bond
> > > > > elongates abnormal whatever in NVT or NPT.
> > > > >
> > > > >
> > > > >
> > > > > The relax time of pressure and temperature were adjust but no
> helpful.
> > > > >
> > > > >
> > > > >
> > > > > How can I found some possible reason and then fix it ? Any advise
> will
> > > > > be appreciated.
> > > >
> > > > If an isolated molecule doesn't do anything strange, and two more more
> > > > do, then you have a problem with inter-molecular interactions, i.e.
> the
> > > > non-bonded ones. Try with just 2 chains to verify this... and look
> > > > carefully at the atom types and the non-bonded parameters for them.
> > > >
> > > > Mark
> > > >
> > > >
> > > >
> > > > *************************************************
> > > > Ji Qing
> > > > Institute of Chemistry, Chinese Academy of Sciences
> > > > Tel: 0086-10-62562894  £¬82618423
> > > > *************************************************
> > > > _______________________________________________
> > > > gmx-users mailing list    gmx-users at gromacs.org
> > >
> > > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > > 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
> > > >
> > >
> > >
> > > --
> > > Tsjerk A. Wassenaar, Ph.D.
> > > Junior UD (post-doc)
> > > Biomolecular NMR, Bijvoet Center
> > > Utrecht University
> > > Padualaan 8
> > > 3584 CH Utrecht
> > > The Netherlands
> > > P: +31-30-2539931
> > > F: +31-30-2537623
> > >
> > > _______________________________________________
> > > gmx-users mailing list    gmx-users at gromacs.org
> > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > 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
> > >
> >
> >
> > --
> > Tsjerk A. Wassenaar, Ph.D.
> > Junior UD (post-doc)
> > Biomolecular NMR, Bijvoet Center
> > Utrecht University
> > Padualaan 8
> > 3584 CH Utrecht
> > The Netherlands
> > P: +31-30-2539931
> > F: +31-30-2537623
>
>
> --
> Tsjerk A. Wassenaar, Ph.D.
> Junior UD (post-doc)
> Biomolecular NMR, Bijvoet Center
> Utrecht University
> Padualaan 8
> 3584 CH Utrecht
> The Netherlands
> P: +31-30-2539931
> F: +31-30-2537623
>
>
> ------------------------------
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
>
>
> End of gmx-users Digest, Vol 35, Issue 22
> *****************************************
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> 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
>


-- 
Tsjerk A. Wassenaar, Ph.D.
Junior UD (post-doc)
Biomolecular NMR, Bijvoet Center
Utrecht University
Padualaan 8
3584 CH Utrecht
The Netherlands
P: +31-30-2539931
F: +31-30-2537623

More information about the gromacs.org_gmx-users mailing list