[QE-users] hybrid pseudopotentials
Giuseppe Mattioli
giuseppe.mattioli at ism.cnr.it
Wed Jul 17 17:20:47 CEST 2024
On the top of this, your calculation, with proper cutoffs, is actually
a big one and will require significant resources. Moreover, if you
want to save time&resources I always suggest to perform hybrid EXX
calculations using norm-conserving pseudopotentials. They are much
more stable in my experience.
Make also wise use of all the available tricks (read the pw manual)...
&system
ecutwfc=80.0,
ecutfock=120.0, <---------
input_dft='vdW-DF2-ahbr'
/
&electrons
diagonalization='david',
mixing_mode='plain'
mixing_beta=0.1
conv_thr=1.0d-7
electron_maxstep=100
adaptive_thr=.true. <---------
/
ATOMIC_SPECIES
Zn 65.380 Zn_ONCV_PBE-1.2.upf
O 15.999 O_ONCV_PBE-1.2.UPF
N 14.007 N_ONCV_PBE-1.2.UPF
C 12.011 C_ONCV_PBE-1.2.UPF
H 1.008 H_ONCV_PBE-1.2.UPF
HTH
Giuseppe
Quoting Matic Poberznik <matic.poberznik at ijs.si>:
> Dear David,
>
> after a quick glance there are several problems with your
> input/parallelization setup. Regarding the parallelization, you are
> using 2 nodes, but only one core per node (effectively running on 2
> cores), I would suggest to request the "cpus-per-task equal to the
> number of cpu's on one node".
>
> On the input side, your ecutrho/ecutwfc cutoffs are way to small for
> any meaningful result, and the scf convergence threshold is way too
> large (the default is 10^-6 and yours in 1.0?). In any case, I
> suggest you carefully check that you understand the meaning of each
> input parameter you specified:
>
> https://www.quantum-espresso.org/Doc/INPUT_PW.html#idm814
>
> Best,
>
> Matic Poberznik
>
> --
> Jozef Stefan Institute, Ljubljana, Slovenia
>
> On 7/17/24 16:32, dlduran at uco.es wrote:
>> Dear QE users:
>>
>> I am working with a perovskite and I would like to obtain the
>> Projected DOS (projwfc) with hybrid pseudopotentials. As far as I
>> know, first a scf calculation must be done and afterwards the own
>> projwfc calculation (scf with large number of kpoints, though).
>>
>> I am still in the scf step because I'm having a lot of problems.
>> Any simple calculation takes me a lot of time in the best of the
>> cases, or the calculations directly crashes for paralellization
>> reasons. I have tried hse and gau-pbe pseudopotentials.
>>
>> This is part of my .in file:
>>
>>
>> &CONTROL
>> calculation = 'scf',
>> restart_mode = 'from_scratch',
>> outdir = './tmp',
>> pseudo_dir = '/home/dlduran/CALCULOS/PSEUDOPOTENCIALES/QE',
>> prefix= 's2',
>> etot_conv_thr=1e0,
>> forc_conv_thr=1e0
>> /
>>
>> &SYSTEM
>> ibrav = 14,
>> celldm(1)=24.8964,
>> celldm(2)=0.8654,
>> celldm(3)=1.4874,
>> celldm(4)=0.06589425,
>> celldm(5)=0.279374894,
>> celldm(6)=-0.160074146,
>> nat = 140,
>> ntyp = 6,
>> ecutwfc = 10,
>> ecutrho = 40,
>> occupations = 'tetrahedra',
>> input_dft= 'hse',
>> nqx1 = 2, nqx2 = 2, nqx3 = 2,
>> x_gamma_extrapolation = .true.,
>> exxdiv_treatment = 'gygi-baldereschi'
>> /
>>
>> &ELECTRONS
>> conv_thr = 1e0
>>
>> /
>>
>> ATOMIC_SPECIES
>> C 12.01060 C.pbe-n-kjpaw_psl.1.0.0.UPF
>> H 1.00750 H.pbe-kjpaw_psl.1.0.0.UPF
>> S 32.06750 S.pbe-n-kjpaw_psl.1.0.0.UPF
>> I 126.90400 I.pbe-n-kjpaw_psl.1.0.0.UPF
>> N 14.00650 N.pbe-n-kjpaw_psl.1.0.0.UPF
>> Pb 207.20000 Pb.pbe-dn-kjpaw_psl.1.0.0.UPF
>>
>> ATOMIC_POSITIONS (crystal)
>> Pb 0.742890000000000 0.764150000000000 0.970330000000000
>> Pb 0.813460000000000 0.227210000000000 0.976720000000000
>> Pb 0.257110000000000 0.235850000000000 0.029670000000000
>> Pb 0.186540000000000 0.772790000000000 0.023280000000000
>>
>> ...
>>
>> H 0.744170000000000 0.472670000000000 0.818640000000000
>> H 0.768130000000000 0.621830000000000 0.789940000000000
>> H 0.647140000000000 0.560850000000000 0.847470000000000
>> H 0.025900000000000 0.955450000000000 0.800600000000000
>> H 0.134390000000000 0.949430000000000 0.831100000000000
>> H 0.056530000000000 0.052470000000000 0.859210000000000
>>
>> K_POINTS (automatic)
>>
>> 2 2 2 0 0 0
>>
>>
>> And this is my paralellization setup:
>>
>>
>> #SBATCH --nodes=2 # Number of nodes
>> #SBATCH --ntasks=2 # Total number of MPI processes
>> #SBATCH --cpus-per-task=1 # Number of CPU (cores) per MPI process
>>
>>
>> This calculation is running for 2 days!
>>
>>
>> I would appreciate any help or suggestion. Do I have to try
>> another easier functional? Any suitable set of kpoints and nqx
>> parameters? Any other flag that I am missing or using wrongly?
>>
>>
>> Thank you very much for your help, David.
>>
>>
>>
>> --------------------------
>> David López Durán
>> Department of Physics
>> University of Córdoba, Spain
>> Phone: +34 957 21 20 32
>> https://es.linkedin.com/in/davidlopezduran
>>
>> _______________________________________________
>> The Quantum ESPRESSO community stands by the Ukrainian
>> people and expresses its concerns about the devastating
>> effects that the Russian military offensive has on their
>> country and on the free and peaceful scientific, cultural,
>> and economic cooperation amongst peoples
>> _______________________________________________
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
>> users mailing list users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
> _______________________________________________
> The Quantum ESPRESSO community stands by the Ukrainian
> people and expresses its concerns about the devastating
> effects that the Russian military offensive has on their
> country and on the free and peaceful scientific, cultural,
> and economic cooperation amongst peoples
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
GIUSEPPE MATTIOLI
CNR - ISTITUTO DI STRUTTURA DELLA MATERIA
Via Salaria Km 29,300 - C.P. 10
I-00015 - Monterotondo Scalo (RM)
Mob (*preferred*) +39 373 7305625
Tel + 39 06 90672342 - Fax +39 06 90672316
E-mail: <giuseppe.mattioli at ism.cnr.it>
More information about the users
mailing list