Expert Advice needed: PERIODIC_EFIELD with metal nano-particle

Sam Broderick

unread,

Jun 13, 2022, 5:42:26 AM6/13/22

to cp2k

Dear cp2k aficionados,

It seems to me that I have run into a strong limitation of cp2k, but I am not know enough to anything close to sure. My goal: determine spectroscopic response of an organic molecule near a metal nano-particle.

On the one hand, &DIAGONALIZATION is strongly recommended for metals. On the other hand, TRAVIS requires &PERIODIC_EFIELD to determine the polarizability, but this means &OT. I am having severe difficulties with the production run with a thermostat.

Is cp2k the right tool or am I just not doing it right (e.g., something better than DZVP-MOLOPT-SR-GTH and GTH-PBE-q)

Would you please have a look at my .inp and provide me with some tips? My boss is quite dissatisfied with node hour usage while I try to figure this out.

Many, many thanks

Sam

Au79-ot-production.inp

Sam Broderick

unread,

Jun 13, 2022, 6:19:20 AM6/13/22

to cp2k

An additional question: after a global run, where the last step did not converge within the WALLTIME I set in the input:

*** SCF run terminated - exceeded requested execution time: 86340.000 seconds.

*** Execution time now: 86340.037 seconds.

What is stored in the restart data? The last converged MD step or the one that did not converge? Could this be why I get this error when I try to start a production run form this? Sorry for the newbie questions.

READ RESTART : WARNING : nspin is not equal

*******************************************************************************
* ___ *
* / \ *
* [ABORT] *
* \___/ Reducing nspin is not possible. *
* | *
* O/| *
* /| | *
* / \ qs_mo_io.F:708 *
*******************************************************************************

===== Routine Calling Stack =====

8 read_mo_set_from_restart
7 calculate_first_density_matrix
6 scf_env_initial_rho_setup
5 init_scf_run
4 qs_energies
3 qs_forces
2 qs_mol_dyn_low
1 CP2K

Sam Broderick

unread,

Jun 13, 2022, 6:24:03 AM6/13/22

to cp2k

The output from the production one (the global run is large show I include the end here:

MD| ***************************************************************************
MD| Step number 5104
MD| Time [fs] 2552.000000
MD| Conserved quantity [hartree] -0.262582505304E+04
MD| ---------------------------------------------------------------------------
MD| Instantaneous Averages
MD| CPU time per MD step [s] 26.061410 33.822750
MD| Energy drift per atom [K] 0.374503646625E+03 0.205283418423E+03
MD| Potential energy [hartree] -0.262589862674E+04 -0.262592631376E+04
MD| Kinetic energy [hartree] 0.137757326096E+00 0.119719069562E+00
MD| Temperature [K] 376.625724 327.309498
MD| ***************************************************************************

Spin 1

Number of electrons: 435
Number of occupied orbitals: 435
Number of molecular orbitals: 435

Spin 2

Number of electrons: 434
Number of occupied orbitals: 434
Number of molecular orbitals: 434

Number of orbital functions: 1975
Number of independent orbital functions: 1975

Extrapolation method: PS Nth order
Extrapolation order: 2

SCF WAVEFUNCTION OPTIMIZATION

----------------------------------- OT ---------------------------------------
Minimizer : DIIS : direct inversion
in the iterative subspace
using 7 DIIS vectors
safer DIIS on
Preconditioner : FULL_SINGLE_INVERSE : inversion of
H + eS - 2*(Sc)(c^T*H*c+const)(Sc)^T
Precond_solver : DEFAULT
stepsize : 0.08000000 energy_gap : 0.08000000
ortho_irac : CHOL irac_degree : 4
max_irac : 50 eps_irac : 0.10000E-09
eps_irac_switch: 0.10000E-01 eps_irac_quick_exit: 0.100E-04
on_the_fly_loc : F
----------------------------------- OT ---------------------------------------

Step Update method Time Convergence Total energy Change
------------------------------------------------------------------------------
1 OT DIIS 0.80E-01 1.6 0.00000775 -2625.8983262879 -2.63E+03
2 OT DIIS 0.80E-01 1.7 0.00000405 -2625.8983310938 -4.81E-06
3 OT DIIS 0.80E-01 1.7 0.00000353 -2625.8983328495 -1.76E-06
4 OT DIIS 0.80E-01 1.7 0.00000334 -2625.8983350122 -2.16E-06
5 OT DIIS 0.80E-01 1.8 0.00000315 -2625.8983386726 -3.66E-06
6 OT DIIS 0.80E-01 1.7 0.00000296 -2625.8983443709 -5.70E-06
7 OT DIIS 0.80E-01 1.7 0.00000269 -2625.8983545672 -1.02E-05
8 OT DIIS 0.80E-01 1.7 0.00000249 -2625.8983637011 -9.13E-06
9 OT DIIS 0.80E-01 1.7 0.00000233 -2625.8983715220 -7.82E-06
10 OT DIIS 0.80E-01 1.7 0.00000220 -2625.8983782333 -6.71E-06
11 OT DIIS 0.80E-01 1.8 0.00000210 -2625.8983844367 -6.20E-06
12 OT DIIS 0.80E-01 1.7 0.00000203 -2625.8983895822 -5.15E-06
13 OT DIIS 0.80E-01 1.7 0.00000197 -2625.8983941766 -4.59E-06
14 OT DIIS 0.80E-01 1.7 0.00000191 -2625.8983986569 -4.48E-06
15 OT DIIS 0.80E-01 1.7 0.00000185 -2625.8984037404 -5.08E-06
16 OT DIIS 0.80E-01 1.7 0.00000178 -2625.8984097126 -5.97E-06
17 OT DIIS 0.80E-01 1.7 0.00000166 -2625.8984185968 -8.88E-06
18 OT DIIS 0.80E-01 1.6 0.00000154 -2625.8984272390 -8.64E-06
19 OT DIIS 0.80E-01 1.7 0.00000142 -2625.8984357725 -8.53E-06
20 OT DIIS 0.80E-01 1.7 0.00000130 -2625.8984426842 -6.91E-06
21 OT DIIS 0.80E-01 1.7 0.00000121 -2625.8984475524 -4.87E-06
22 OT DIIS 0.80E-01 1.7 0.00000113 -2625.8984512475 -3.70E-06
23 OT DIIS 0.80E-01 1.7 0.00000106 -2625.8984545255 -3.28E-06
24 OT DIIS 0.80E-01 1.7 0.00000099 -2625.8984573944 -2.87E-06

*** SCF run terminated - exceeded requested execution time: 86340.000 seconds.

*** Execution time now: 86340.037 seconds.

*** SCF run converged in 24 steps ***

Electronic density on regular grids: -869.0000000000 0.0000000000
Core density on regular grids: 868.9999999985 -0.0000000015
Total charge density on r-space grids: -0.0000000014
Total charge density g-space grids: -0.0000000014

Overlap energy of the core charge distribution: 0.00000146582976
Self energy of the core charge distribution: -4570.41375350254566
Core Hamiltonian energy: 1458.28031159567945
Hartree energy: 899.55026105138484
Exchange-correlation energy: -412.42098926839282
Dispersion energy: -0.89428873636523

Total energy: -2625.89845739440943

outer SCF iter = 1 RMS gradient = 0.99E-06 energy = -2625.8984573944
outer SCF loop converged in 1 iterations or 24 steps

ENERGY| Total FORCE_EVAL ( QS ) energy [a.u.]: -2625.898460009454993

*** MD run terminated - exceeded requested execution time: 86340.000 seconds.

*** Execution time now: 86343.048 seconds.

Au79-ot-production-error.out

Krack Matthias (PSI)

unread,

Jun 13, 2022, 7:27:30 AM6/13/22

to cp...@googlegroups.com

Hi Sam

it seems that you try to restart from an OT run with spin polarization (LSD/UKS ON) using an input in which LSD/UKS is set OFF or not set (the default is OFF), but it is not possible to change the spin-polarization on restart.

The OT run should have written the last converged (spin-polarized) wavefunction to the wfn restart file.

And right, PERIODIC_EFIELD is only implement for OT.

HTH

Matthias

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/1cb69247-862f-4dba-85a8-4953bed216e9n%40googlegroups.com.

Sam Broderick

unread,

Jun 13, 2022, 8:27:12 AM6/13/22

to cp2k

Thank you so much on the UKS! I don't see the forest for the trees by now, with all the trying out.

Jürg Hutter

unread,

Jun 13, 2022, 8:34:57 AM6/13/22

to cp...@googlegroups.com

Hi

if this is a cluster, you don't need the periodic Efield.

This is the input for a diagonalization run. What are your OT settings?

regards

JH

________________________________________

From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Sam Broderick <wave...@gmail.com>

Sent: Monday, June 13, 2022 11:42 AM
To: cp2k
Subject: [CP2K:17144] Expert Advice needed: PERIODIC_EFIELD with metal nano-particle

--
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/2d334de9-f0ab-4c24-920a-52c664e06d74n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/2d334de9-f0ab-4c24-920a-52c664e06d74n%40googlegroups.com?utm_medium=email&utm_source=footer>.

Sam Broderick

unread,

Jun 13, 2022, 8:47:29 AM6/13/22

to cp2k

Hello Prof. Hutter

Thank you for looking into this! It is truncated octohedral Au-79 particle/cluster based on a literature search.

Do I understand correctly, that you are suggesting to use a &VIBRATIONAL_ANALYSIS? Am I missing something and this includes Raman (which is the reason that I am currently using TRAVIS and &PERIODIC_EFIELD? I was under the impression it is IR reflection/absorption only.

Kind Regards

Sam Broderick

Jürg Hutter

unread,

Jun 13, 2022, 9:22:38 AM6/13/22

to cp...@googlegroups.com

Hi

I'm not an expert in this, but it looks to me you have the following options:

1) Use PERIODIC NONE, without smearing and a constant
field (EFIELD keyword). Do MD and use TRAVIS to analyze.

2) Do MD and calculate the polarizability tensor along the trajectory and
derive Raman spectra from the auto-correlation.

3) Calculate the IR spectra and perform finite difference calculations
of the polarizability (without smearing) along (selected) normal modes.
Calculate the Raman intensity from this data.

Maybe some spectra experts can help to find the most efficient method.

regards

JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Sam Broderick <wave...@gmail.com>

Sent: Monday, June 13, 2022 2:47 PM
To: cp2k
Subject: Re: [CP2K:17152] Expert Advice needed: PERIODIC_EFIELD with metal nano-particle

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/c206960f-95a6-4f80-861f-b68672908ebfn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/c206960f-95a6-4f80-861f-b68672908ebfn%40googlegroups.com?utm_medium=email&utm_source=footer>.

Sam Broderick

unread,

Jun 13, 2022, 10:25:49 AM6/13/22

to cp2k

Is there anyone you would recommend?

Sam Broderick

unread,

Jun 15, 2022, 6:00:15 AM6/15/22

to cp2k

My saga continues! I have been able to get reasonable performance with:

&OT

ALGORITHM IRAC

PRECONDITIONER FULL_SINGLE_INVERSE

MINIMIZER DIIS

However, I cannot convince &E_DENSITY_BQB or &E_DENSITY_CUBE to produce any data/files. It just runs through and does not show any density information. Does anyone have any ideas? I know that there is a warning about the symmetry module, which I don't have installed, but the production run worked OK.

Many, many thanks,

Sam

Au79-production.inp

Au79-Ex.inp

Au79-Ex.out

Au79-production.out

Marcella Iannuzzi

unread,

Jun 15, 2022, 6:49:40 AM6/15/22

to cp2k

Dear Sam,

The print statement is only in the REFTRAJ run, where

EVAL_ENERGY_FORCES

is set "false". Since no electronic structure calculation is done, no density cube file can be printed.

Regards

Marcella

Sam Broderick

unread,

Jun 15, 2022, 7:29:32 AM6/15/22

to cp2k

How stupid of me. Thank you!

Reply all

Reply to author

Forward