Question about PDB file format for molecular H2 simulation in i-PI

[BO LIU]

Hello Developer,

I am a beginner in i-PI and I am learning how to use it through the ph2 example in /i-pi/examples/ph2/nvt/. 

Now, I want to use i-PI to run classical MD to study the properties of liquid hydrogen using silvera golden potential. However, when I run the input.xml file, I encounter errors due to using a self-created PDB file, whereas the example PDB file (atomic style) runs without issues. Since I am simulating a molecular hydrogen system, I believe using the atomic style is inappropriate. Does i-PI have specific requirements for the style of PDB files? Could this be the source of the error? 

As a beginner, would it be possible for you to provide me with a simple tutorial on this topic?

Thanks very much!

All the best,

BO LIU

[Mariana Rossi]

Hi Bo,

PDB files are very restrictive and reading all possible categories that can be contained in one such pdb file is not possible in i-PI. We only read the simple ones, so this could be the source of the error (you can look at the description in the manual). The number of decimal cases that are allowed and so on is also restrictive from our point of view.

Nevertheless, for liquid hydrogen I am sure you don't really need a pdb file, right? Please check our documentation https://docs.ipi-code.org/  (or https://docs.ipi-code.org/input-tags.html#file) and use instead xyz or ase (extended xyz) formats, which we also support. 

All the best,

Mariana

[BO LIU]

Hello  Mariana  ,

Thanks so much for your explanation and suggestion. I will try the xyz format and check the documentation.

All the best,

Bo

[BO LIU]

Hello Mariana,

I have successfully performed classical MD simulations of liquid hydrogen using the xyz format file in i-PI. However, some issues have emerged during this process.

My system consists of 50 H₂ molecules at 1 MPA and 25K, with the box size calculated based on the corresponding density under these conditions. During the simulation, I observed that while the conserved quantity remains essentially constant, both the temperature and pressure become excessively high (in an NVT ensemble, although there's no specific requirement for pressure, the density condition should constrain the pressure). This has led to hydrogen molecules dissociating into atoms. Since I only made some simple modifications to the tutorial case, are you familiar with this abnormal situation? Do you have any better suggestions for improvement?

Thanks very much!

All the best,

Bo

[BO LIU]

there are some files about my simulation.

[Mariana Rossi]

Hi Bo,

In the SG potential, each H2 molecule is a single entity. See the discussion and detailed newer tutorial here: https://github.com/sabia-group/i-PI-tutorial -- therefore you need to describe each H2 molecule by its center of mass. Can that be the problem?

All the best,

Mariana

[BO LIU]

Hi  Mariana,

I have found the corresponding PDB file in the tutorial you provided, and I observed through OVITO that the PDB file displays as individual particles. Combined with the information you shared, does this mean that if I want to use the SG potential to simulate liquid hydrogen, I only need to use a particle resembling an atom (containing the mass of two hydrogen atoms and the center-of-mass information of the hydrogen molecule) to represent the hydrogen molecule? This way, I can successfully complete the simulation?  
Thank you very much for your advice and the tutorial information you provided.

All the best,

Bo