output polarizability of water and MB-SCAN

[Holo Santis]

Hi,

I would like to calculate dipole and polarizability of water using lammps interface, how can I output polarizabilities? Another question is that I noticed that this paper(https://www.nature.com/articles/s41467-021-26618-9) mentioned MB-SCAN is available in MBX, may I know how to access it? Thanks!

Best,

Santis

[Holo Santis]

May I know is it difficult? Or MD-SCAN is just fake, so it is impossible to be public, what mentioned in that paper is cheating reviewers. 

[Francesco Paesani]

Dear Santis,

Thank you for your email and patience.

MB-DFT is not fake but is still not part of the public version of MBX because we are waiting for a paper that is in press in the Journal of Chemical Physics. It will be part of the next release of MBX. However, if it is urgent, we can have a Zoom meeting where we can guide you through the steps of using the MB-DFT potentials in the current version of MBX. Please let us know and we will schedule the meeting.

Regarding the polarizability, it is currently not outputted by MBX but can be calculated post-processing the trajectories. Also in this case, we are happy to show you how to do that.

All the best,

Francesco

[Francesco Paesani]

Dear Santis,

Thank you for your patience. Our latest paper on ion hydration using the MB-SCAN and MB-SCAN(DC) potentials is now out as part of the John Perdew Festschrift of The Journal of Chemical Physics (https://doi.org/10.1063/5.0174577).

The MB-SCAN and MB-SCAN(DC) potentials for water as well as Na+ and Cl- in water are available in the mb_scan_mb_dcscan branch of MBX that you can clone from the corresponding GitHub repository: https://github.com/paesanilab/MBX/tree/mb_scan_mb_dcscan. Similar to the master branch, the mb_scan_mb_dcscan branch of MBX is compatible with LAMMPS or i-PI, allowing you to run both classical (MD) and quantum (PIMD/CMD/TRPMD) simulations.

As a reference, here is the list of our papers describing our data-driven many-body formalism applied to DFT and DC-DFT:
https://www.nature.com/articles/s41467-021-26618-9
https://doi.org/10.1021/acs.jctc.2c00050
https://doi.org/10.1063/5.0090862
https://doi.org/10.1021/acs.jctc.2c00313
https://doi.org/10.1063/5.0129613

If you have any further questions or need assistance with using our MB-SCAN and MB-SCAN(DC) potentials, please don't hesitate to reach out.

All the best,

Francesco

[Bence Balázs Mészáros]

Dear  Francesco Paesani,

I would like to use the MB-SCAN potential for water. I have found the right GitHub branch, however, I couldn't find instructions, on how to "turrn on" or choose the  MB-SCAN, or MB-SCAN(DC) potentials in the input. I've already used the ipi_driver executable with the original MBX branch according to the examples, what modifications do I need in the case of MB-SCAN?

Best regards,

Bence Mészáros

[Bence Balázs Mészáros]

Dear  Francesco Paesani,

I cloned the specified git branch, but I could not find instructions on how to "turn on" the MB-SCAN or MB-SCAN(DC) potentials instead of the MB-pol. How can I specify in the input, which potential to use? (I'm using the ipi_driver executable.)

Best wishes,

Bence Mészáros

Francesco Paesani a következőt írta (2023. november 11., szombat, 2:41:42 UTC+1):

[Etienne Palos]

Dear Bence Mészáros,

Thank you for your interest in MBX and our many-body models, and for reaching out with your question regarding the MB-SCAN potential for water.

In MBX, the implemented potentials are called using  their “monomer name”.

Using the config.nrg file for MB-pol as an example here, “h2o” is the monomer name, and so in MBX this corresponds to / calls MB-pol.

MB-pol I-PI example:

https://github.com/paesanilab/MBX/blob/mb_scan_mb_dcscan/plugins/i-pi/examples/molecular_dynamics/condensed_phase/npt/256h2o/config.nrg 

1. TO USE MB-DFT MODELS WITH I-PI

To use MB-SCAN instead of MB-pol, you replace  “MONOMER h2o” with “MONOMER h2o-scan” in the config.nrg file.

MB-SCAN config.nrg

SYSTEM WATER 256

MOLECULE

MONOMER h2o-scan

O       2.781038e+00   -3.651332e+00   -1.210302e+00  

H       3.212050e+00   -3.232290e+00   -1.990020e+00  

H       1.875751e+00   -3.261017e+00   -1.183359e+00  

ENDMON

ENDMOL

MOLECULE

MONOMER h2o-scan

.

.

.

Rest of file 

MB-SCAN(DC) config.nrg

SYSTEM WATER 256

MOLECULE

MONOMER h2o-dcscan

O       2.781038e+00   -3.651332e+00   -1.210302e+00  

H       3.212050e+00   -3.232290e+00   -1.990020e+00  

H       1.875751e+00   -3.261017e+00   -1.183359e+00  

ENDMON

ENDMOL

MOLECULE

MONOMER h2o-dcscan

.

.

.

Rest of file 

For I-PI simulations, this is all you need to change. 

2. TO USE MB-DFT MODELS WITH LAMMPS
For reference, the change is similar for LAMMPS. In the lammps.in file, we would define h2o-dcscan as opposed to h2o ; for ions Na+ and Cl-, we would use na+dcscan and cl-dcscan, respectively.

Defining only e.g. cl- and h2o, would use MB-nrg and MB-pol.

MB-SCAN(DC) LAMMPS example:

https://github.com/paesanilab/MBX/tree/mb_scan_mb_dcscan/plugins/lammps/examples/cl-_bulk_dcscan 

As we further develop MB-DFT, MB-DFT(DC) and related methods, we will continue to add more examples for both I-PI and LAMMPS, as well as improving our documentation.

Please let us know if we can be of any assistance, and happy computing!

Best regards,

The MBX team

[Bence Balázs Mészáros]

Dear Etienne Palos,

Thank you for your answer, the specified git branch worked for me with MB-SCAN, however, since it uses MBX 1.0 it has a significant computational overhead compared to MB-pol in new versions. Are you planning to include MB-SCAN(DC) in the new 1.2 release of MBX?

Best regards,

Bence Mészáros

[Etienne Palos]

Dear Bence Mészáros,

Thank you for reaching out and informing us about the performance issues with MB-SCAN and MB-SCAN(DC) compared to MB-pol in MBX v1.2.

We would like to inform you that the performance-enhanced MB-SCAN and MB-SCAN(DC) potentials for water, as well as Na+ and Cl− in water,
are now available in the branch mbx_with_mb-dft_1.2: https://github.com/paesanilab/MBX/tree/mbx_with_mb-dft_1.2

This update replaces the previous (v1.0) mb_scan_mb_dcscan branch, which will soon be deprecated.

We appreciate your interest in MB-SCAN, MB-SCAN(DC), and MB-DFT.
We're pleased to share that in the next release (v1.3), MB-SCAN, MB-SCAN(DC), and representative related MB-DFT models will be included in the master branch.

Please don't hesitate to reach out if you need any assistance. Happy computing!

Best regards,
The MBX Developer Team

[Bence Balázs Mészáros]

Dear Etienne Palos,

Thank you for the quick update, it works fine now!

Best regards,

Bence Mészáros