msi2lmp does not read and export certain bonds to data file

Hi all,

I have a structure containing different atoms and bonding. The problem is that the data file produced by msi2lmp does not contain certain types of bonds, angles, and dihedrals. I don’t understand the reason because these connections are seen to be correct in the mdf file. But the export to data file is not working. All other properties are correct.
Attached are the .car and .mdf files. I am specifically talking about Zr-Zr bonds and the related angles and dihedrals.

Do you have similar experiences? What can be the reason and solution?

Structure.car (165 KB)

Structure.mdf (179 KB)

Hi all,

I have a structure containing different atoms and bonding. The problem is that the data file produced by msi2lmp does not contain certain types of bonds, angles, and dihedrals. I don’t understand the reason because these connections are seen to be correct in the mdf file. But the export to data file is not working. All other properties are correct.
Attached are the .car and .mdf files. I am specifically talking about Zr-Zr bonds and the related angles and dihedrals.

Do you have similar experiences? What can be the reason and solution?

what is the version of msi2lmp you are using and what is the command line?

axel.

Hi,

I used the msi2lmp version available in lammps stable 12 Dec 2018.
The command I use:
msi2lmp.exe Structure -frc cvff -class I -ignore

Hi,

I used the msi2lmp version available in lammps stable 12 Dec 2018.
The command I use:
msi2lmp.exe Structure -frc cvff -class I -ignore

why use -ignore? aren’t you concerned about getting an incomplete conversion by ignoring missing force field info?

axel.

when i run msi2lmp on my local machine with your files, i get a complete topology and geometry conversion.
however, since your atom types do not match the atom types for cvff AT ALL, all your mass and force field parameters are zero. msi2lmp does not know what to do with them. most certainly, there are not parameters for Zr. that is what -ignore does. it just substitutes 0.0 for unknown masses and force field parameters.
all bonds however are there. you are getting what you are asking for. don’t blame msi2lmp, when you feed it badly prepared files.

axel.

Thanks for your reply.
I am aware what -ignore does. I actually don’t care about the 0.0 values, because I have all the force-field parameters ready. I want to use msi2lmp to build the structure and the data file. Then, I will add the FF parameters to the file later. At the moment, I just want complete topology and geometry conversions.

That’s interesting you got the complete file. I have been working several days on this issue, and haven’t found the solution yet. Which version you used? Could you please send me the data file produced? Thanks.

Thanks for your reply.
I am aware what -ignore does. I actually don’t care about the 0.0 values, because I have all the force-field parameters ready. I want to use msi2lmp to build the structure and the data file. Then, I will add the FF parameters to the file later. At the moment, I just want complete topology and geometry conversions.

without knowing the exact force field, you cannot get a “complete” conversion. how should the software for example know which impropers to define? there may be other issues, too. for those you essentially need to know the type of bonds.

That’s interesting you got the complete file. I have been working several days on this issue, and haven’t found the solution yet. Which version you used? Could you please send me the data file produced? Thanks.

i am using version v3.9.9 get a file which looks complete. i don’t have the time to piece-by-piece verify whether this is exact. again, if you use the -ignore flag, all bets are off and you become subject to internal heuristics. for example, when i do the conversion with VMD, i get a different number of bonds, angles, dihedrals and impropers, because it uses different heuristics.

if you know what bonds etc. there should be, then rather than abusing a tool not made for the kind of process you are looking for, you should write a script that builds the topology and writes out a data file exactly as you need it and to match the force field interactions you need to apply. expecting tools using heuristics for molecular systems to work under such unusual circumstances corresponds to a degree of optimism that i cannot comprehend.

axel.

Could you please guide where I can get the version v3.9.9 from?
I understand you are too busy to examine the file. If you send me your converted file, I can verify it precisely.

I rebuilt the structure and manually added Zr-Zr bonds (assuming probably automatic way in Mat Studio did not work well).
It is interesting that msi2lmp with cvff could convert several of the Zr-Zr bonds! This shows that it can find, read, and convert these bonds. However, it is not clear to me why it does not convert all of them, specifically considering that the structure is symmetric.

Structure_New.car (165 KB)

Structure_New.mdf (179 KB)

Hi,

Just want to check if msi2lmp performs as follows in the following situations:

  1. The bond is defined between two atoms in the mdf file, and cvff includes that bond ==> bond + coeffs in the data file
  2. The bond is defined between two atoms in the mdf file, but cvff does not include that bond ==> bond + 0.0 coeffs in the data file

Thanks.

Hi,

Just want to check if msi2lmp performs as follows in the following situations:

  1. The bond is defined between two atoms in the mdf file, and cvff includes that bond ==> bond + coeffs in the data file
  2. The bond is defined between two atoms in the mdf file, but cvff does not include that bond ==> bond + 0.0 coeffs in the data file

the only way to know for certain is to study the source code. LAMMPS and msi2lmp are open source. so feel free to help yourself.

but if translating all bonds into a data file is your desire and you don’t care about force field assignments, then using VMD and the TopoTools plugin is probably a much better choice. VMD will read and apply the bond information from .mdf files and TopoTools will use whatever information VMD provides. so if all bonds are visible when you load the .mdf + .crd file into VMD, then TopoTools will be able to write them to a data file.

axel.