rdf does not fall to one

45 views
Skip to first unread message

Victor Nazarychev

unread,
Nov 9, 2022, 1:48:57 PM11/9/22
to votca
Dear Votca users,

I try to parametrize the non-bonded intecations in CG model using the ibi procedure for graphene molecules. However, when I build nanocomposite comprises polymer and graphene flakes is atomostic simulation stacks are formed between the graphene flakes.  In this case, the target pair distribution function between CG atoms in graphene falls to 1 only on the end of box. please see the attached file.

If I undesrtand it is imposible to acurate calculate potential between the atoms in stacks using ibi procedure due to the fact the rdf does not fall to one. However, could I calculate in some case this like potential for CG model with help of the force-matching?

Best regards,
Victor   

rdf_AA.png

Christoph Junghans

unread,
Nov 9, 2022, 2:52:45 PM11/9/22
to vo...@googlegroups.com
On Wed, Nov 9, 2022 at 11:48 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>
> Dear Votca users,
>
> I try to parametrize the non-bonded intecations in CG model using the ibi procedure for graphene molecules. However, when I build nanocomposite comprises polymer and graphene flakes is atomostic simulation stacks are formed between the graphene flakes. In this case, the target pair distribution function between CG atoms in graphene falls to 1 only on the end of box. please see the attached file.
>
> If I undesrtand it is imposible to acurate calculate potential between the atoms in stacks using ibi procedure due to the fact the rdf does not fall to one.
There are re-weighting techniques for these kind of systems, e.g. see
https://doi.org/10.1063/1.4742067

> However, could I calculate in some case this like potential for CG model with help of the force-matching?
I think force-matching could work as well.

Christoph
>
> Best regards,
> Victor
>
> --
> Join us on Slack: https://join.slack.com/t/votca/signup
> ---
> You received this message because you are subscribed to the Google Groups "votca" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/965daf19-3386-4355-8547-e651605e49b8n%40googlegroups.com.



--
Christoph Junghans
Web: http://www.compphys.de

Victor Nazarychev

unread,
Nov 10, 2022, 7:01:38 AM11/10/22
to votca
Dear Cristoph, 

Thank you for your help. 

среда, 9 ноября 2022 г. в 22:52:45 UTC+3, Christoph Junghans:
On Wed, Nov 9, 2022 at 11:48 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>
> Dear Votca users,
>
> I try to parametrize the non-bonded intecations in CG model using the ibi procedure for graphene molecules. However, when I build nanocomposite comprises polymer and graphene flakes is atomostic simulation stacks are formed between the graphene flakes. In this case, the target pair distribution function between CG atoms in graphene falls to 1 only on the end of box. please see the attached file.
>
> If I undesrtand it is imposible to acurate calculate potential between the atoms in stacks using ibi procedure due to the fact the rdf does not fall to one.
There are re-weighting techniques for these kind of systems, e.g. see
https://doi.org/10.1063/1.4742067

I have a couple of clarifying questions. I did not find re-weighting techniques in the VOTCA, as I understand this approach has not been implemented yet?
 
> However, could I calculate in some case this like potential for CG model with help of the force-matching?
I think force-matching could work as well.


If I use bonded potentials calculated by the ibi procedure, will calculating only nonbonded potentials with the help of force-matching be correct for my model? Or should I parametrize bonded and non-bonded potential together by means of force-matching?

Best regards, 
Victor

Christoph Junghans

unread,
Nov 10, 2022, 11:30:55 AM11/10/22
to vo...@googlegroups.com
On Thu, Nov 10, 2022 at 5:01 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>
> Dear Cristoph,
>
> Thank you for your help.
>
> среда, 9 ноября 2022 г. в 22:52:45 UTC+3, Christoph Junghans:
>>
>> On Wed, Nov 9, 2022 at 11:48 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>> >
>> > Dear Votca users,
>> >
>> > I try to parametrize the non-bonded intecations in CG model using the ibi procedure for graphene molecules. However, when I build nanocomposite comprises polymer and graphene flakes is atomostic simulation stacks are formed between the graphene flakes. In this case, the target pair distribution function between CG atoms in graphene falls to 1 only on the end of box. please see the attached file.
>> >
>> > If I undesrtand it is imposible to acurate calculate potential between the atoms in stacks using ibi procedure due to the fact the rdf does not fall to one.
>> There are re-weighting techniques for these kind of systems, e.g. see
>> https://doi.org/10.1063/1.4742067
>>
> I have a couple of clarifying questions. I did not find re-weighting techniques in the VOTCA, as I understand this approach has not been implemented yet?
Correct and most likely it will never get added. You will have to
overwrite the RDF calculator script with something that works for your
special case.

>
>>
>> > However, could I calculate in some case this like potential for CG model with help of the force-matching?
>> I think force-matching could work as well.
>>
>
> If I use bonded potentials calculated by the ibi procedure, will calculating only nonbonded potentials with the help of force-matching be correct for my model? Or should I parametrize bonded and non-bonded potential together by means of force-matching?

That depends on the system, so I don't know.
We mixed IB/IBI with force-matching in this paper
http://dx.doi.org/10.1002/mats.201100011 and it worked, but that isn't
a guarantee it will work for your system.
IBI reproduces the effective PMF while FM reproduces the average
force, which is not always the same.

Christoph

>
> Best regards,
> Victor
>
>>
>> Christoph
>> >
>> > Best regards,
>> > Victor
>> >
>> > --
>> > Join us on Slack: https://join.slack.com/t/votca/signup
>> > ---
>> > You received this message because you are subscribed to the Google Groups "votca" group.
>> > To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
>> > To view this discussion on the web visit https://groups.google.com/d/msgid/votca/965daf19-3386-4355-8547-e651605e49b8n%40googlegroups.com.
>>
>>
>>
>> --
>> Christoph Junghans
>> Web: http://www.compphys.de
>
> --
> Join us on Slack: https://join.slack.com/t/votca/signup
> ---
> You received this message because you are subscribed to the Google Groups "votca" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/fc03698e-076d-4702-8a3a-3c956bc3e079n%40googlegroups.com.

Victor Nazarychev

unread,
Nov 11, 2022, 2:35:24 AM11/11/22
to vo...@googlegroups.com
Dear Christoph, 

Thank you very much.  

чт, 10 нояб. 2022 г. в 19:30, Christoph Junghans <jung...@votca.org>:
On Thu, Nov 10, 2022 at 5:01 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>
> Dear Cristoph,
>
> Thank you for your help.
>
> среда, 9 ноября 2022 г. в 22:52:45 UTC+3, Christoph Junghans:
>>
>> On Wed, Nov 9, 2022 at 11:48 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>> >
>> > Dear Votca users,
>> >
>> > I try to parametrize the non-bonded intecations in CG model using the ibi procedure for graphene molecules. However, when I build nanocomposite comprises polymer and graphene flakes is atomostic simulation stacks are formed between the graphene flakes. In this case, the target pair distribution function between CG atoms in graphene falls to 1 only on the end of box. please see the attached file.
>> >
>> > If I undesrtand it is imposible to acurate calculate potential between the atoms in stacks using ibi procedure due to the fact the rdf does not fall to one.
>> There are re-weighting techniques for these kind of systems, e.g. see
>> https://doi.org/10.1063/1.4742067
>>
> I have a couple of clarifying questions. I did not find re-weighting techniques in the VOTCA, as I understand this approach has not been implemented yet?
Correct and most likely it will never get added. You will have to
overwrite the RDF calculator script with something that works for your
special case.
>>
>> > However, could I calculate in some case this like potential for CG model with help of the force-matching?
>> I think force-matching could work as well.
>>
>
> If I use bonded potentials calculated by the ibi procedure, will calculating only nonbonded potentials with the help of force-matching be correct for my model? Or should I parametrize bonded and non-bonded potential together by means of force-matching?

That depends on the system, so I don't know.
We mixed IB/IBI with force-matching in this paper
http://dx.doi.org/10.1002/mats.201100011 and it worked, but that isn't
a guarantee it will work for your system.
IBI reproduces the effective PMF while FM reproduces the average
force, which is not always the same.

If I understand the procedure csg_fmatch is not paralleled and can be used in single processor mode only? 


Best regards.
Victor
 
You received this message because you are subscribed to a topic in the Google Groups "votca" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/votca/N0ETXMTYpBw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to votca+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAHG27e6nbZXSt17gPHgsHnOe8ySouq-1OS2BB4CsgANGwYiBwA%40mail.gmail.com.

Victor Nazarychev

unread,
Nov 14, 2022, 7:20:37 AM11/14/22
to votca
Dear Christoph, 

I tried parameterising non-bonded interaction in composite systems with stacks containing the polymer (C-type beads) and graphene-like molecules (A-type beads).

To understand in what range, I should conduct force calculations, first I calculated the target distribution A-A, A-C, C-C, please see fig.SYS_all_targets.jpg. 

From these distributions, I understand that I should calculate the forces in the range from 0 to 4 ns. Then, using settting_fm2.xml I started a force-matching procedure. However, for C-C interactions, I got infinity 

#### Force matching residual ####
     Chi_2[(kJ/(mol*nm))^2] = -nan
#################################

This should be a number: -nan
This should be a number: -7.13506
This should be a number: -6.01957

I changed the ranges in which I calculated the forces (from 0 to 2, 3 or 3.5 nm). Nevertheless, for the C-C interactions, the number was NAN.   

Maybe I should change some parameters in the setting file? Or should I increase the box size? 

Best wishes, 

Victor

<cg>

  <fmatch>
    <!--Number of frames for block averaging -->
    <frames_per_block>6</frames_per_block>
    <!--Constrained least squares?-->
    <constrainedLS>0</constrainedLS>
  </fmatch>
   
    <non-bonded>
        <name>C-C</name>
        <type1>C</type1>
        <type2>C</type2>
        <fmatch>
            <min>0.0</min>
            <max>4.0</max>
                  <step>0.01</step>
                  <out_step>0.005</out_step>
        <res_output_coeff>5</res_output_coeff>
        </fmatch>
    </non-bonded>

    <non-bonded>
        <name>A-A</name>
        <type1>A</type1>
        <type2>A</type2>
        <fmatch>
            <min>0.0</min>
            <max>4.0</max>
                  <step>0.01</step>
                  <out_step>0.005</out_step>
        <res_output_coeff>5</res_output_coeff>
        </fmatch>
    </non-bonded>
   
        <non-bonded>
        <name>A-C</name>
        <type1>A</type1>
        <type2>C</type2>
        <fmatch>
            <min>0.0</min>
            <max>4.0</max>
                  <step>0.01</step>
                  <out_step>0.005</out_step>
        <res_output_coeff>5</res_output_coeff>
        </fmatch>
    </non-bonded>


</cg>

пятница, 11 ноября 2022 г. в 10:35:24 UTC+3, Victor Nazarychev:
SYS_all_targets.jpg

Christoph Junghans

unread,
Nov 14, 2022, 12:33:44 PM11/14/22
to vo...@googlegroups.com
On Mon, Nov 14, 2022 at 5:20 AM Victor Nazarychev <nazar...@gmail.com> wrote:
>
> Dear Christoph,
>
> I tried parameterising non-bonded interaction in composite systems with stacks containing the polymer (C-type beads) and graphene-like molecules (A-type beads).
>
> To understand in what range, I should conduct force calculations, first I calculated the target distribution A-A, A-C, C-C, please see fig.SYS_all_targets.jpg.
>
> From these distributions, I understand that I should calculate the forces in the range from 0 to 4 ns. Then, using settting_fm2.xml I started a force-matching procedure. However, for C-C interactions, I got infinity
>
> #### Force matching residual ####
> Chi_2[(kJ/(mol*nm))^2] = -nan
> #################################
>
> This should be a number: -nan
> This should be a number: -7.13506
> This should be a number: -6.01957
>
> I changed the ranges in which I calculated the forces (from 0 to 2, 3 or 3.5 nm). Nevertheless, for the C-C interactions, the number was NAN.
>
> Maybe I should change some parameters in the setting file? Or should I increase the box size?
I am not the force-matching expert, so I will have to think about this
a bit more carefully, but I am assuming the error is an artifact from
the fact that the system isn't uniform.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/d142f494-d0e7-4138-a6c6-faedfedcb961n%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages