Re: dummy atom mass changed

From: Jerome Henin (jhenin_at_cmm.chem.upenn.edu)
Date: Tue May 12 2009 - 11:51:55 CDT

Hi,
That is correct. Either your calculations are not fully converged, or
the code is buggy. Which of course it isn't.
Jerome

On Tue, May 12, 2009 at 12:10 PM, yun luo <luoyun724_at_gmail.com> wrote:
> Hello,
>
> I keep the fepVdwLambdaEnd=1.0, but change the fepElecLambdaStart from 0.2
> to 0.8 with decouple on. The energy of disappearing an ion in bulk water
> shift a lot. Are the different fepElecLambdaStart value supposed to only
> change the path but total enegry should converge at the end?
>
>
>
> On Thu, May 7, 2009 at 11:18 AM, yun luo <luoyun724_at_gmail.com> wrote:
>>
>> Thank you, Christ and Floris, that makes sense. I used the same system as
>> the "Charging a spherical ion" example for NAMD2.6. The ion is growing in a
>> box with length of 30 A and I use "switchdist 10  cutoff  12". The ion is
>> fixed in the center of box. So it's 30 A away from it's image. I will test a
>> larger box to see if I get the similar solvation energy.
>>
>> Lyna
>>
>>
>> On Thu, May 7, 2009 at 5:09 AM, Floris Buelens <floris_buelens_at_yahoo.com>
>> wrote:
>>>
>>> Hi, I can confirm what Chris said - it's clear why you'd assume that
>>> 'decouple on' shouldn't affect a single-atom perturbation but there is a
>>> difference arising from PME. With 'decouple off' the end state is an atom
>>> with no non-bonded interactions whatsoever (vdW or electrostatic), while
>>> with 'decouple on' non-bonded interactions across the periodic boundary are
>>> still in effect; vdW interactions will be beyond the cutoff but a charged
>>> atom will still interact with its periodic images.
>>> best regards,
>>>
>>> Floris
>>>
>>> ________________________________
>>> From: yun luo <luoyun724_at_gmail.com>
>>> To: char_at_ks.uiuc.edu
>>> Cc: NAMD <namd-l_at_ks.uiuc.edu>
>>> Sent: Wednesday, 6 May, 2009 17:50:06
>>> Subject: Re: namd-l: dummy atom mass changed
>>>
>>> Could anybody explain me what the "decouple on" do in case of
>>> disappearing one ion in pure water ?
>>>
>>>
>>> On Tue, May 5, 2009 at 3:26 PM, yun luo <luoyun724_at_gmail.com> wrote:
>>>
>>> Hi Christ,
>>>
>>> Thanks for your notice. According to the ug, "decouple" is dealing with
>>> intramolecular interactions. Since I only disappear one ion, there is no
>>> intramolecular interaction. So I think decouple on/off should give similar
>>> result. However, I disappear one sodium ion in TIP3 box with:
>>>
>>>
>>> fepVdwLambdaEnd 1.0
>>> fepElecLambdaStart 0.5
>>> fepVdwShiftCoeff 5.0
>>> decouple on/off
>>>
>>> I got dG=-73 kcal/mol with "decouple on" and dG=-89 kcal/mol with
>>> "decouple off (default)". How do you think those value? Where does the
>>> difference come from?
>>>
>>>
>>>
>>> On Tue, May 5, 2009 at 12:53 PM, Chris Harrison <char_at_ks.uiuc.edu> wrote:
>>>
>>> Ly,
>>>
>>> The code has been changing very quickly in the last several months and
>>> the default values for your softcore parameters will depend on when you
>>> checked out the cvs or downloaded the beta.  I very strongly say this:  set
>>> the parameters yourself.  Until beta2, do not assume the softcore
>>> parameters' default values.  Until beta2, do not assume the softcore
>>> parameters' default values in the users guide are correct.  Because the code
>>> has been changing so rapidly, SET THE SOFTCORE PARAMETERS MANUALLY based on
>>> the below suggestion.
>>>
>>> The default values in the cvs are currently, and will be in NAMD2.7b2:
>>>
>>> fepVdwLambdaEnd = 1.0
>>> fepElecLambdaStart=0.5
>>> fepVdwShiftCoeff=5.0
>>>
>>> If you want to use softcore in an FEP you should probably start with the
>>> following parameters and values:
>>>
>>> fepVdwLambdaEnd 1.0
>>> fepElecLambdaStart 0.5
>>> fepVdwShiftCoeff 5.0
>>> decouple on
>>>
>>> MAKE SURE to set decouple to on.  It's default, for other reasons, is
>>> currently set to off.
>>>
>>> Thanks for your feedback.  It is valuable input as we try to resolve any
>>> problems during the beta phase
>>>
>>>
>>>
>>> C.
>>>
>>>
>>> --
>>> Chris Harrison, Ph.D.
>>> Theoretical and Computational Biophysics Group
>>> NIH Resource for Macromolecular Modeling and Bioinformatics
>>> Beckman Institute for Advanced Science and Technology
>>> University of Illinois, 405 N. Mathews Ave., Urbana, IL 61801
>>>
>>> char_at_ks.uiuc.edu                            Voice: 217-244-1733
>>> http://www.ks.uiuc.edu/~char               Fax: 217-244-6078
>>>
>>>
>>>
>>>
>>> On Tue, May 5, 2009 at 11:49 AM, yun luo <luoyun724_at_gmail.com> wrote:
>>>
>>> However, in the NAMD2.7 UserGuide, both fepVdWLambdaEND and
>>> fepEleLambdaStart have default value 0.5.  So I guess if we don't specify
>>> those two value, the decoupling or coupling of VdW interaction will only be
>>> carried out from lambda 0 to 0.5, and decoupling or coupling of
>>> Electrostatic interaction will only be carried out during lambda 0.5 to 1.
>>> Is that right?
>>>
>>>
>>>
>>> On Tue, May 5, 2009 at 10:43 AM, Jerome Henin <jhenin_at_cmm.chem.upenn.edu>
>>> wrote:
>>>
>>> Hi all,
>>>
>>> Since this issue has been confusing to many people, here is a figure
>>> illustrating two common types of calculations, and what happens to the
>>> various coupling/decoupling parameters.
>>>
>>> Cheers,
>>> Jerome
>>>
>>>
>>>
>>> On Tue, May 5, 2009 at 11:18 AM, daniel aguayo <bioquimico_at_gmail.com>
>>> wrote:
>>> > Tks Chris for your answer. I now understand the meaning of it.
>>> >
>>> > Best
>>> >
>>> > Daniel Aguayo V.
>>> >
>>> > On Mon, May 4, 2009 at 11:24 PM, Chris Harrison <char_at_ks.uiuc.edu>
>>> > wrote:
>>> >>
>>> >> Daniel,
>>> >>
>>> >> fepVdwLambdaEnd is the lambda value at which the coupling/decoupling
>>> >> of
>>> >> vdW interactions is desired to be completed.  The default is 1.0, so
>>> >> the vdW
>>> >> interactions of annihilated atoms are gradually decoupled from a
>>> >> system from
>>> >> lambda=0 to lambda=1.  In complimentary fashion, the vdW interactions
>>> >> of
>>> >> appearing atoms are gradually coupled to a system from lambda=0 to
>>> >> lambda=1.
>>> >>
>>> >> fepElecLambdaStart in a similar fashion controls the electrostatic
>>> >> coupling/decoupling.  The default is 0.5, which results in the
>>> >> electrostatic
>>> >> interactions of annihilated atoms being gradually attenuated from
>>> >> lambda=0
>>> >> to lambda=0.5 as they are decoupled from the system, while
>>> >> electrostatic
>>> >> interactions involving appearing atoms are gradually coupled to the
>>> >> system
>>> >> from lambda 0.5 to 1.0.
>>> >>
>>> >> A value of 0.5 is perhaps not the best example to use, but does this
>>> >> answer your question?
>>> >>
>>> >>
>>> >> C.
>>> >>
>>> >>
>>> >> --
>>> >> Chris Harrison, Ph.D.
>>> >> Theoretical and Computational Biophysics Group
>>> >> NIH Resource for Macromolecular Modeling and Bioinformatics
>>> >> Beckman Institute for Advanced Science and Technology
>>> >> University of Illinois, 405 N. Mathews Ave., Urbana, IL 61801
>>> >>
>>> >> char_at_ks.uiuc.edu                            Voice: 217-244-1733
>>> >> http://www.ks.uiuc.edu/~char               Fax: 217-244-6078
>>> >>
>>> >>
>>> >>
>>> >> On Mon, May 4, 2009 at 7:53 PM, daniel aguayo <bioquimico_at_gmail.com>
>>> >> wrote:
>>> >>>
>>> >>> Hi Jerome, can you explain more on the use of this new parameters
>>> >>> fepElecLambdaStart and fepVdwLambdaEnd
>>> >>>
>>> >>> Tks
>>> >>>
>>> >>> Daniel Aguayo V.
>>> >>> CBSM UTAL
>>> >>> Chile
>>> >>>
>>> >>>
>>> >>> On Mon, May 4, 2009 at 6:39 PM, Jerome Henin
>>> >>> <jhenin_at_cmm.chem.upenn.edu>
>>> >>> wrote:
>>> >>>>
>>> >>>> Hi,
>>> >>>> As you said, the only reason why the dummy atom is needed is because
>>> >>>> NAMD 2.6 does not have soft-core potentials. The purpose of the
>>> >>>> tutorial is somewhat academic, in that the "charging free energy" is
>>> >>>> only meaningful within some theories of ion solvation, and is not an
>>> >>>> experimental observable.
>>> >>>> If you want a complete solvation free energy, then it is not
>>> >>>> necessary
>>> >>>> to follow the tutorial's "pseudo-single topology" approach.
>>> >>>>
>>> >>>> Note that if for some reason, you do want to use a dummy atom, its
>>> >>>> mass will not affect the thermodynamics of the system. It should
>>> >>>> typically not be less than 1.0, otherwise you may need to use
>>> >>>> smaller
>>> >>>> timesteps to preserve the stability of the simulation.
>>> >>>>
>>> >>>> One more remark: even the charging free energy can now be computed
>>> >>>> without the help of a dummy atom, since NAMD 2.7b1 allows for the
>>> >>>> separate decoupling of electrostatic and L-J interactions, through
>>> >>>> the
>>> >>>> fepElecLambdaStart and fepVdwLambdaEnd parameters.
>>> >>>>
>>> >>>> Best,
>>> >>>> Jerome
>>> >>>>
>>> >>>> On Mon, May 4, 2009 at 5:24 PM, yun luo <luoyun724_at_gmail.com> wrote:
>>> >>>> > Hi Chris,
>>> >>>> >
>>> >>>> > Thank you for your reply.
>>> >>>> > Actually, I did follow the FEP tutorial part 2. Charging a
>>> >>>> > spherical
>>> >>>> > ion
>>> >>>> > using dual-topology paradigm. That's why I want add a dummy atom.
>>> >>>> > In
>>> >>>> > the
>>> >>>> > tutorial one adds a dummy atom with 0 charge but the same radius
>>> >>>> > as
>>> >>>> > sodium
>>> >>>> > because there is no soft-core contribution in NAMD2.6. Since
>>> >>>> > NAMD2.7
>>> >>>> > has
>>> >>>> > soft-core part, I think I need using a dummy atom with 0 charge 0
>>> >>>> > radius to
>>> >>>> > get both elec and vdW energy. But do you think the a dummy atom
>>> >>>> > with
>>> >>>> > nonzero
>>> >>>> > mass will cause problem?
>>> >>>> >
>>> >>>> > Many thanks!
>>> >>>> >
>>> >>>> > On Mon, May 4, 2009 at 3:33 PM, Chris Harrison <char_at_ks.uiuc.edu>
>>> >>>> > wrote:
>>> >>>> >>
>>> >>>> >> Ly,
>>> >>>> >>
>>> >>>> >> With 0 charge, 0 mass, and 0 radius the dummy particle is
>>> >>>> >> effectively
>>> >>>> >> the
>>> >>>> >> same as no atom .... which is of course the desired effect
>>> >>>> >> usually.
>>> >>>> >> Assuming you're not doing something unusual, then you don't need
>>> >>>> >> the
>>> >>>> >> dummy
>>> >>>> >> particle.  You should be able to just make the Na disappear.
>>> >>>> >>
>>> >>>> >> The FEP tutorial may be of help in setting up calculations.  This
>>> >>>> >> tutorial
>>> >>>> >> is for NAMD 2.6 but the system setup process should be the same
>>> >>>> >> as
>>> >>>> >> NAMD2.7b1.
>>> >>>> >>
>>> >>>> >> NAMD2.6 FEP tutorial:
>>> >>>> >>
>>> >>>> >>
>>> >>>> >> http://www.ks.uiuc.edu/Research/namd/tutorial/fep/AlchemicalFEP-Mar2008.pdf
>>> >>>> >>
>>> >>>> >> Required files for tutorial:
>>> >>>> >>
>>> >>>> >>
>>> >>>> >>
>>> >>>> >> http://www.ks.uiuc.edu/Research/namd/tutorial/fep/AlchemicalFEP-Mar2008.zip
>>> >>>> >>
>>> >>>> >>
>>> >>>> >> C.
>>> >>>> >>
>>> >>>> >>
>>> >>>> >> --
>>> >>>> >> Chris Harrison, Ph.D.
>>> >>>> >> Theoretical and Computational Biophysics Group
>>> >>>> >> NIH Resource for Macromolecular Modeling and Bioinformatics
>>> >>>> >> Beckman Institute for Advanced Science and Technology
>>> >>>> >> University of Illinois, 405 N. Mathews Ave., Urbana, IL 61801
>>> >>>> >>
>>> >>>> >> char_at_ks.uiuc.edu                            Voice: 217-244-1733
>>> >>>> >> http://www.ks.uiuc.edu/~char               Fax: 217-244-6078
>>> >>>> >>
>>> >>>> >>
>>> >>>> >>
>>> >>>> >> On Mon, May 4, 2009 at 2:54 PM, yun luo <luoyun724_at_gmail.com>
>>> >>>> >> wrote:
>>> >>>> >>>
>>> >>>> >>> Hi,
>>> >>>> >>>
>>> >>>> >>> I'm using NAMD2.7 for running FEP. I need to disappear a sodium
>>> >>>> >>> ion
>>> >>>> >>> in my
>>> >>>> >>> membrane. So I overlay a dummy atom with 0 mass 0 charge 0
>>> >>>> >>> radius on
>>> >>>> >>> a
>>> >>>> >>> sodium ion. But I got a warning below:
>>> >>>> >>>
>>> >>>> >>> Warning: FOUND 1 ATOMS WITH ZERO OR NEGATIVE MASSES!  CHANGED TO
>>> >>>> >>> 0.001
>>> >>>> >>>
>>> >>>> >>> I'm wondering if the nonzero mass will effect the energy? If
>>> >>>> >>> yes,
>>> >>>> >>> how to
>>> >>>> >>> stop this automatic changing?
>>> >>>> >>>
>>> >>>> >>> Thanks!
>>> >>>> >>>
>>> >>>> >>> Ly
>>> >>>> >>
>>> >>>> >
>>> >>>> >
>>> >>>>
>>> >>>
>>> >>>
>>> >>>
>>> >>> --
>>> >>> saludos desde el fin del mundo
>>> >>
>>> >
>>> >
>>> >
>>> > --
>>> > saludos desde el fin del mundo
>>> >
>>>
>>>
>>>
>>
>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:50:57 CST