units in votca

[Saeed Norouzi]

Dear all,

I am currently working on my CG ibi using VOTCA. The all-atom simulation has been conducted using GROMACS, and I have obtained the target distribution from my AA model. For the ibi part, I am utilizing LAMMPS, which requires a conversion of units. I achieved this by scaling the bond lengths(nm to A) and nonbonded table values (nm to A) by a factor of 10 using a Python script.

As far as my understanding goes, the populations are unitless, and therefore, I believe I do not need to alter the population size. Up to this point, the radial distribution function (RDF) appears satisfactory. However, I have encountered an issue with the bond distribution from ibi iterations not matching the target distribution (please refer to the attached image where the red line represents the target distribution, and the black line depicts the calculated distribution from ibi iterations).

My question is: Should I also divide the populations for bonds by a factor of 10 to ensure consistency?

Thank you for your assistance.

Best regards

Seed

[Christoph Junghans]

Did you calculate the target distributions with VOTCA as well? (with csg_stat)

I have to look at the code again, but I believe VOTCA is normalizing the distribution by some means.

Christoph

Thank you for your assistance.

Best regards

Seed

--
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/CAJZjMJZt0j%2BN%3DeEqa5DY-vx0F50xSeHrr7iQ3%3DFk54UHBfSnYQ%40mail.gmail.com.

--

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

[Saeed Norouzi]

Dear Christoph,

Yes, csg_stat using votca for the distribution. I did not touch the angle and dihedrals. just changed the nm to A for the bond and RDF.

Best regards

Saeed 

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAHG27e5Lfbh4sOXf%3Dfg13Xdzp-a4mz%2BRmJAz1qO%3D-3g0VGTFhw%40mail.gmail.com.

[Cecília Álvares]

So the red and black curves were both calculated using csg_stat? If both curves (the red and black) are an output of the csg_stat, I dont see how it is possible that the curves have different areas underneath them (I would say there should be no normalization factor to be applied in this case).

Your plot made me think of a scenario where you are calculating a reference by means of some python code and you are calculating bond distribution functions and not probability density functions for the bond (they are not the same). So this is not the case?

[Saeed Norouzi]

I calculated the both distributions using cas_stat, the red one is the target which I got from AA simulations in gromacs (in gromacs the units are in nm for bonds so I multiplied the x-data by 10 to use for ibi in lammps), the black one is the new target during the ibi process (again using the csg_stat this time simulations in lammps). I used the python code only to convert the nm to A by x-data*10. 

Best regards

Saeed

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/c4d9f892-7716-42c4-bfa1-2d3992d764edn%40googlegroups.com.

[Christoph Junghans]

On Fri, Aug 18, 2023 at 8:39 AM Cecília Álvares <cecilia...@gmail.com> wrote:

So the red and black curves were both calculated using csg_stat? If both curves (the red and black) are an output of the csg_stat, I dont see how it is possible that the curves have different areas underneath them (I would say there should be no normalization factor to be applied in this case).

Your plot made me think of a scenario where you are calculating a reference by means of some python code and you are calculating bond distribution functions and not probability density functions for the bond (they are not the same). So this is not the case?

I think what Ceillia is asking is if you used csg_stat to also calculate the target distribution.

Additionally a quick back of the envelope calculation:

the IBI update is

delta_U ~= log(d/d_target) = log(d/(d'_target*Norm))=log(d/d'_target)-log(Norm)

so the Norm will just show up as constant in the potential update and hence no actual force.

So it doesn't matter for the update, but it will matter for the convergence number.

Christoph

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/c4d9f892-7716-42c4-bfa1-2d3992d764edn%40googlegroups.com.

[Saeed Norouzi]

The target is also using csg_stat.

Best regards

Saeed

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAHG27e79FPRAGerjEX4CTvu5YRLdy4gvJJL7sn1Gbym8HuLspA%40mail.gmail.com.

[Cecília Álvares]

What do you obtain if you integrate the red curve? And what about the black curve? What is strange is that, even if they are not matching, they should both have the integral from r = 0 to r = 3.9 (angs, I suppose) to be equal to 1. The curve that does not yield an integral equal to 1 should be the wrong one. I suppose it is the red.

[Saeed Norouzi]

Integral for the red one is 10 and for the black one 1. So if I convert unit from nm to A, I need to divide the y-axis by 10 also,

Best regards

Saeed

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/ba0a852a-a216-479d-9b1a-d0e809b369ebn%40googlegroups.com.