SIMPLEX

[Alireza Moradzadeh]

Dear VOTCA Users,

For a system composed of two beads , I want to optimize pressure and interaction based on RDF but I want to stick to Lorentz-Berthelot combination rule. I want to know if I define my function as follows would it be right or not? and should I define pressure as target in each interaction. 

CG1-CG1

target rdf presssure

function  4*eps1*[(sig1/x)**12-(sig1/x)**6]

parmeters eps1 sig1

CG1-CG2

target rdf pressure

function 4*(eps1*eps2)**2*[((sig1+sig2)/2/x)**12-(sig1+sig2)/2/x)**6]

parameters eps1 sig1 eps2 sig2

CG2-CG2

target rdf presssure

function  4*eps2*[(sig2/x)**12-(sig2/x)**6]

parmeters eps2 sig2

Thanks,

[Christoph Junghans]

2016-04-07 19:10 GMT-06:00 Alireza Moradzadeh <a.m...@gmail.com>:
> Dear VOTCA Users,
>
>
> For a system composed of two beads , I want to optimize pressure and
> interaction based on RDF but I want to stick to Lorentz-Berthelot
> combination rule. I want to know if I define my function as follows would it
> be right or not? and should I define pressure as target in each interaction.

Parameters are always global so, you should not list any parameter
twice, but you need to list at least one parameter for each
interaction.
e.g. list eps1 sig1 for CG1-CG1, eps2 for CG1-CG2 and sig2 for CG2-CG2

>
> CG1-CG1
> target rdf presssure
> function 4*eps1*[(sig1/x)**12-(sig1/x)**6]
> parmeters eps1 sig1
>
> CG1-CG2
> target rdf pressure
> function 4*(eps1*eps2)**2*[((sig1+sig2)/2/x)**12-(sig1+sig2)/2/x)**6]

This should be square root and not Power 2.

Christoph

> parameters eps1 sig1 eps2 sig2
>
>
> CG2-CG2
> target rdf presssure
> function 4*eps2*[(sig2/x)**12-(sig2/x)**6]
> parmeters eps2 sig2
>
> Thanks,
>

> --
> 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 post to this group, send email to vo...@googlegroups.com.
> Visit this group at https://groups.google.com/group/votca.
> For more options, visit https://groups.google.com/d/optout.



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

[Alireza Moradzadeh]

Thank you for your answer. 

I am also wondering if there is a way to keep eps and sigma positive during simulation time. Because the optimization method is blind to physical meaning of eps and sigma I got negative value using other methods in votca package.

Is it possible to enforce this condition using this method or RE. Also is there anyway in votca package to treat charge of beads as a variable during simulation so I could control sigma and epsilon by variable charge.

Thanks,

Alireza

[Alireza Moradzadeh]

By the way I would be glad if you could provide more information on how to choose parameters initials 

I could not find so much information on how I should choose those value or even its organization

Thanks,

Alireza

[Christoph Junghans]

2016-04-08 21:21 GMT-06:00 Alireza Moradzadeh <a.m...@gmail.com>:
> Thank you for your answer.
>
> I am also wondering if there is a way to keep eps and sigma positive during
> simulation time. Because the optimization method is blind to physical
> meaning of eps and sigma I got negative value using other methods in votca
> package.
> Is it possible to enforce this condition using this method or RE. Also is
> there anyway in votca package to treat charge of beads as a variable during
> simulation so I could control sigma and epsilon by variable charge.

No, we haven't implemented any range restrictions for simplex. For RE
I am not sure but Sikandar will know.
However you could just use the absolute values of eps and sigma in the
parameterization of the interaction, which will effectively force them
to be positive.

Christoph

[Christoph Junghans]

2016-04-08 22:26 GMT-06:00 Alireza Moradzadeh <a.m...@gmail.com>:
> By the way I would be glad if you could provide more information on how to
> choose parameters initials
> I could not find so much information on how I should choose those value or
> even its organization

Have a look at the methanol-water simplex tutorial there is a
simplex.in file for each interaction.
The values in there are given in the same order as in the xml file.
You will need n+1 sets to start with (lines in simplex.in) These sets
can be generated by randomly perturbing some values from a single set,
which you got from e.g. fitting an IBI potential.

Christoph

[Alireza Moradzadeh]

I looked at simplex tutorials. I don,t now why we need n+1 line, furthermore, when i tried to run my settings.xml based on the a combination rule I got the error mentioning there is no parameters for interaction between two beads. I am trying simplex on a four bead model so I don't have enough parameters to allocate for all of the interaction.

> 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/ISshiaVA4mc/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to votca+un...@googlegroups.com.

[Alireza Moradzadeh]

How can i impose absolute value selection on my simulation. By the way, using sigma and epsilon there is no need for table, wouldn't it be better to change topology file at each step. So implementation of combination rule would be automatic. I already wrote a simple scripts to do this after each iteration but it has some limitation for example i have to run my setting file step by step. This approach slow downs my work.

Alireza

> 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/ISshiaVA4mc/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to votca+un...@googlegroups.com.

[Christoph Junghans]

2016-04-09 16:38 GMT-06:00 Alireza Moradzadeh <a.m...@gmail.com>:
> I looked at simplex tutorials. I don,t now why we need n+1 line,

n+1 sets is a requirement of the simplex algorigtm, see our recent paper:
<http://dx.doi.org/10.1371/journal.pone.0131754>
or
<https://en.wikipedia.org/wiki/Nelder%E2%80%93Mead_method>

> furthermore, when i tried to run my settings.xml based on the a combination rule I got the error mentioning there is no parameters for
> interaction between two beads. I am trying simplex on a four bead model so I don't have enough parameters to allocate for all of the
> interaction.

Like I said in my last email you need at least one parameter per
interaction, you have 3 interactions and 4 parameters so you should be
fine. However looking at the code, I think it could be rewritten to
list this constraint.

Christoph

[Christoph Junghans]

2016-04-09 16:45 GMT-06:00 Alireza Moradzadeh <a.m...@gmail.com>:
> How can i impose absolute value selection on my simulation.

You can use the abs() function in the functional form.

>By the way, using sigma and epsilon there is no need for table, wouldn't it be better to change topology file at each step. So implementation of combination rule would be automatic. I already wrote a simple scripts to do this after each iteration but it has some limitation for example i have to run my setting file step by step. This approach slow downs my work.

Sure, that would be another way to implement it, but then VOTCA
wouldn't be able to use any other functional form that LJ, which would
be a huge restriction. When you search the mailing list you will see
that this topic came up before and there was the idea to write a
post-update script to convert the simplex state file back to an LAMMPS
parameter input, so this approach is surely possible.
But we opted to use tables to keep VOTCA more flexible (and simpler)
beside the fact that I am not really seeing any downside to it. In
addition you still need to implement a range check for the parameters
like e.g. suggested in sec 4.1 of:
<https://www.cs.usfca.edu/~brooks/papers/amoeba.pdf>

Christoph

[Sikandar Mashayak]

Hi,

Currently RE implementation does not support constraint optimization. So only way to ensure physically meaningful parameters (C12, C6 or sigma epsilon) is to provide good initial guess, check other MD simulation parameters such as cut-off, dielectric permittivity in case of charges, and using small enough relaxation/mixing parameter.

Best,

Sikandar