On using i-pi to calculate the D2O system

[ke Lin]

Dear users,

Hi everyone, I have recently been using i-pi for simulations of aqueous systems, and the difference between pimd and MD when calculating pure H2O is consistent with previous studies. But it is the pimd of D2O when the results obtained are consistent with calculating the pimd of H2O.

I'm not sure if there is a problem with the D2O setup. Can you tell me how to calculate the D2O system using i-pi.

The top picture shows my results and the bottom shows someone else's work. You can notice that the MD and PIMD of the H2O are good, it's the PIMD of the D2O that's not right.

Thank you and best regards.

[Venkat Kapil]

Hi! You just need to replace the 'H' atoms with 'D' atoms in the input structure to i-PI (note that ASE doesn't allow using a D symbol, but you can use pdb or the i-PI xyz format. Everything on the client side should be the same as for H2O.

[ke Lin]

Hello, my previous calculation was changing only the mass of the H atom in xyz, should I modify the sign as well? For example from H to D.

[Venkat Kapil]

Hi, I do not fully understand what you mean by changing the mass in the xyz. Conventionally, the mass is not specified in the xyz file. If you change the label of the atoms from H to D in an xyz format that i-PI can read or in the pdb format that I-PI can read, i-PI will automatically associate the correct physical masses.

[ke Lin]

Then I have a problem with my setup, I am using lammps in conjunction with i-pi, and the lammps input file data format has quality in it, and I only changed that. Does that mean that the H in the lammps input file data has to be changed to D? Or just change the D in the xyz of the i-pi input file?

[ke Lin]

I have 6 input files now, ipi.xml and md01.xyz are related to i-pi and md01.data is related to lammps. So I should change H to D in md01.xyz. the others don't need to be modified right?

[Venkat Kapil]

Yes, only the i-PI xyz file. Changing the lammps file doesn't change anything as the potential doesn't depend on the mass and i-PI still uses the masses from md01.xyz. Hope this helps.

[ke Lin]

Thank you for your patience!