how to correct jumps in total dipole from Berry phase calculation

[Emma Rossi]

Dear all,

I ran an ab initio molecular dynamics simulation of a molecule in solution in a cubic box, having 15.86 angstrom long side. 

I computed the three components of the total dipole of the box (Mu_x, Mu_y, Mu_z) in periodic conditions, thus using the Berry Phase approach, as reported in the manual at 

CP2K_INPUT / FORCE_EVAL / DFT / LOCALIZE / PRINT / TOTAL_DIPOLE

with PERIODIC TRUE

The plot of the Mu_x, Mu_y and Mu_z in Debye as a function of the time (please find the plots attached) show jumps, which depend on the PBC. 

I read in the literature that one should "remain on the same branch of the polarization lattice" ( A Beginner’s Guide to the Modern Theory of Polarization,  Nicola A. Spaldin , https://arxiv.org/pdf/1202.1831.pdf ) and I think I should multiply my values by a factor, which depends on the box side (15.86 angstrom). 

However, I cannot figure out what is this correction factor? How it depends on the box side?

I would be very grateful if any of you could explain how to proceed.

Thank you very much in advance for your availability and kind support.

Best regards,

Emma Rossi

[Jürg Hutter]

Hi

you have to add/subtract the values by multiples of the quantum of polarization.
For cubic/orthorombic cells that is L/2pi (check for units!).

regards

JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Emma Rossi <emma.r...@studenti.unipd.it>
Sent: Tuesday, January 10, 2023 6:16 PM
To: cp2k
Subject: [CP2K:18306] how to correct jumps in total dipole from Berry phase calculation

Dear all,
I ran an ab initio molecular dynamics simulation of a molecule in solution in a cubic box, having 15.86 angstrom long side.

I computed the three components of the total dipole of the box (Mu_x, Mu_y, Mu_z) in periodic conditions, thus using the Berry Phase approach, as reported in the manual at

CP2K_INPUT<https://manual.cp2k.org/trunk/CP2K_INPUT.html> / FORCE_EVAL<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL.html> / DFT<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT.html> / LOCALIZE<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/LOCALIZE.html> / PRINT<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/LOCALIZE/PRINT.html> / TOTAL_DIPOLE<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/LOCALIZE/PRINT/TOTAL_DIPOLE.html>

with PERIODIC TRUE

The plot of the Mu_x, Mu_y and Mu_z in Debye as a function of the time (please find the plots attached) show jumps, which depend on the PBC.

I read in the literature that one should "remain on the same branch of the polarization lattice" ( A Beginner’s Guide to the Modern Theory of Polarization, Nicola A. Spaldin , https://arxiv.org/pdf/1202.1831.pdf ) and I think I should multiply my values by a factor, which depends on the box side (15.86 angstrom).
However, I cannot figure out what is this correction factor? How it depends on the box side?

I would be very grateful if any of you could explain how to proceed.
Thank you very much in advance for your availability and kind support.

Best regards,
Emma Rossi

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/c3563bc4-535b-493f-921d-708b2ca9444bn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/c3563bc4-535b-493f-921d-708b2ca9444bn%40googlegroups.com?utm_medium=email&utm_source=footer>.