[QE-users] [EXT] Environ Bands Not Converging
lance.kavalsky at mail.utoronto.ca
Sat Nov 3 21:13:47 CET 2018
Dear Professor Andreussi,
Thank you so much for your detailed response! I changed continuum interface to using the soft sphere approach and it appears to have resolved the scf convergence issue.
It was able to complete a full scf cycle on the initial structure, but before starting the second cycle it crashes now giving an error of:
%%%%%%%%%%%%%%%%%%%%%%%%%%% task # 5 from update_environ_charges : error # 1 Inconsistent integral of total charge %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Interestingly, setting env_static_permittivity=1.0 in environ.in and removing assume_isolated = 'slabz' did not reproduce this error, and the calculation was able to completely finish. I also noticed in the output file just before the crash, it has the line "extrapolated charge 131.09828, renormalised to 131.00000" and I'm wondering if this may be related. I would very much appreciate any additional suggestions to address this new error.
University of Toronto
From: users <users-bounces at lists.quantum-espresso.org> on behalf of Andreussi, Oliviero <Oliviero.Andreussi at unt.edu>
Sent: Thursday, November 1, 2018 6:46 PM
To: Quantum Espresso users Forum
Subject: Re: [QE-users] [EXT] Environ Bands Not Converging
Thanks for reporting the problem. I am not sure of its source, but the defaults of Environ should be good enough for simple systems like yours and if you tested different numerical parameters without success this would suggest me that there is a physical reason why the calculation is not converging. My main guess is that the problem resides with the lithium ion. Have you tested the calculation without lithium? Have you tried the calculation with lithium and without the solvent? What pseudo potential are you using for lithium (does the lithium ion have just one valence electron or does it also have its core electrons)?
If lithium has only its valence electron and loses it to the substrate, this is probably going to be a problem for SCCS. In SCCS the continuum boundary is defined on the electronic density, so it means that the continuum goes on top of the lithium ion. To avoid this you could either use full potential for lithium or use an alternative approach for the continuum interface, namely the soft-sphere solvation approach (Fisicaro et al. JCTC 2017). To use the latter you can just specify solvent_mode = ‘ionic’. In this case, you may want to check that the soft-sphere radius for lithium is reasonable.
Hope this helps.
Department of Physics
University of North Texas
Email: oliviero.andreussi at unt.edu<mailto:oliviero.andreussi at unt.edu>
On Nov 1, 2018, at 5:28 PM, Lance Kavalsky <lance.kavalsky at mail.utoronto.ca<mailto:lance.kavalsky at mail.utoronto.ca>> wrote:
I am currently trying to use Environ and am experiencing difficulty. When running vc-relax I am receiving the error "bands not converged". This is despite changing the environ_thr, tol, solvent_mode, conv_thr, diagonalization, and mixing_mode. I have yet to see it complete a single full scf cycle, and watching the estimated scf accuracy it will periodically explode hitting values as large as 1700 Ry before finally giving the error message.
Any additional suggestions would be very much appreciated. Both the pw.x and environ.in files have been pasted below. I am using QE 6.2.1 and Environ 1.0.
University of Toronto
tprnfor = .TRUE.
restart_mode = 'from_scratch',
prefix = 'all',
outdir = './all',
Pseudo_dir = '~/qe_pseudo'
nstep = 1000000
assume_isolated = 'slabz'
conv_thr = 1.D-5
18.638982356 0.000000000 0.000000000
-9.319491169 16.141832210 0.000000000
0.000000000 0.000000000 43.442244346
C 12.0107 C.pbe-n-kjpaw_psl.0.1.UPF
Li 6.94 Li.pbe-s-kjpaw_psl.0.2.1.UPF
Li 2.465903098 2.847312956 7.998767559
C -0.001222067 -0.002150713 5.748168621
C -0.006894107 1.419668805 5.748711907
C -1.235385692 2.135486140 5.748168054
C -1.235385692 3.559103273 5.748168054
C -2.465827629 4.271813207 5.747820607
C -2.465827488 5.694593503 5.747451858
C -3.697991730 6.405978452 5.747822309
C -3.699995777 7.827928911 5.748170451
C 2.465832575 -0.007964030 5.748711231
C 2.465830973 1.415317747 5.733405805
C 1.225702627 2.131306376 5.733413904
C 1.225702627 3.563283037 5.733413904
C -0.006894107 4.274920608 5.748711907
C -0.001222067 5.696740126 5.748168621
C -1.233668925 6.405976111 5.747821630
C -1.231667255 7.827928014 5.748169206
C 4.932884056 -0.002150957 5.748168156
C 4.938555745 1.419665120 5.748705064
C 3.705961632 2.131305141 5.733395741
C 3.705961632 3.563284272 5.733395741
C 2.465830973 4.279271666 5.733405805
C 2.465832575 5.702553443 5.748711231
C 1.231662375 6.408544130 5.748170451
C 1.233666422 7.830494589 5.747822309
C 7.397488816 -0.000004090 5.747451858
C 7.397488675 1.422776206 5.747820607
C 6.167048845 2.135488891 5.748165854
C 6.167048845 3.559100522 5.748165854
C 4.938555745 4.274924293 5.748705064
C 4.932884056 5.696740370 5.748168156
C 3.699990897 6.408545027 5.748169206
C 3.697989227 7.830496930 5.747821630
4 4 1 0 0 0
pbc_correction = 'parabolic'
pbc_dim = 2
pbc_axis = 3
tol = 1.D-13
users mailing list
users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the users