Recent Forum Posts
From categories:
page 1123...next »
convergence problem with magnetic systems
S. MohammadHossein Modarresi (guest) 22 Apr 2018 14:27
in discussion Forum / Support Questions » convergence problem with magnetic systems

Dear Colleagues
I am using the DFT-1/2 ability of the exciting package for calculation of the electronic structure of CaS+impurity. I first relax the system with normal DFT calculations and then apply the DFT-1/2 technique to obtain accurate electronic structure of the system.

In non-magnetic cases, the calculations run very well and with no problem. But relaxation of the spin polarized systems is a very elaborate task for me and even in some cases I am not able to relax the system after playing with computational parameters. Even in the simple case of a free S atom, I could not perform a spin polarized scf calculations with the "Carbon" version of EXCITING and hence I used the "Boron" version and a high Gmax value to be able to converge the spin polarized calculations of a free S atom.

Increasing Gmax is not able to solve the convergence problem of my spin polarized calculations in the CaS+impurity system.
Your kind comment and helps are highly appreciated.

convergence problem with magnetic systems by S. MohammadHossein Modarresi (guest), 22 Apr 2018 14:27

Hello,

I have a big supercell whose groundstate calculation can't be done within a time limit of my computer system. Is there a way to continue the partially completed calculation?

Hi dear,
when we create an input file for a sheet, like Graphen, we put for ngridk ,for exampel,8 8 1. and executet this command:SETUP-convergence-test.py 8 20 4 4.but when it creat workdire ,i have 7 directory that the first of them start from 8 8 8.why this hapen?why it changed for ngridk in Z direction?

Beast regards,
Sahel loushab

Installation
Sahar Qavi (guest) 18 Apr 2018 17:06
in discussion Forum / Controversy » Installation

Hello,
I want to install the software on my PC but I am not sure how to do that, would you please help me?
Thanks

Installation by Sahar Qavi (guest), 18 Apr 2018 17:06
Re: MPI Run
sahel loushabsahel loushab 08 Apr 2018 16:28
in discussion Forum / Support Questions » MPI Run

I finde this paper
, and use this command befor serial installation: export OMP_NUM_THREADS=32
is it true? and befor MPI i most use it?

beast regards,
sahel loushab

Re: MPI Run by sahel loushabsahel loushab, 08 Apr 2018 16:28
Re: MPI Run
sahel loushabsahel loushab 08 Apr 2018 16:01
in discussion Forum / Support Questions » MPI Run

Dear caterina,

when i installing exciting,after command "make" we have two question:1.Enter the number of the platform that suites your system best: 9 and 2.Build MPI binary ? (yes/No) no but i want build MPI binary and choice "yes" but it cant do this. can you help me?

Best regards,
sahel loushab

I get this error:

libbzint is ready.
make[3]: Entering directory '/opt/exciting-mpi/exciting/build/mpi'
mpif90-ifort -O3 -ip -unroll -scalar_rep -openmp -DUSEOMP -cpp -DXS -DISO -DLIBXC -DIFORT -DFFTW -DMPI -DMPIRHO -DMPISEC -DMPI1 -Ifinclude -c ../../src/src_libbzint/kgen_internals.f90
make[3]: mpif90-ifort: Command not found
Makefile.libbzint:80: recipe for target 'kgen_internals.o' failed
make[3]: * [kgen_internals.o] Error 127
make[3]: Leaving directory '/opt/exciting-mpi/exciting/build/mpi'
../Make.common:80: recipe for target 'libbzint.a' failed
make[2]: *
[libbzint.a] Error 2
make[2]: Leaving directory '/opt/exciting-mpi/exciting/build/mpi'
Makefile:6: recipe for target 'all' failed
make[1]: * [all] Error 2
make[1]: Leaving directory '/opt/exciting-mpi/exciting/build/mpi'
Makefile:17: recipe for target 'mpi' failed
make: *
[mpi] Error 2

Re: MPI Run by sahel loushabsahel loushab, 08 Apr 2018 16:01
Requirment
sahel loushabsahel loushab 08 Apr 2018 08:31
in discussion Forum / Support Questions » Requirment

Hi dear caterina,

I want make a super cell of Phosphoren and then put on it some amino acide molcule for cheeking change in electronic and optice propertice as a sensore.I create a unit cell of phosphoren(as sheet) that have 4 atom.I excute convergence for rgkmax and ngridk to find optimize k-pointe. but after 3day it just run for 1directory.my system have 32core. can you help me? is it my input file incorect? or my computer is it not soficeant for this work?

Beast regurds,
Sahel loushab

Requirment by sahel loushabsahel loushab, 08 Apr 2018 08:31

Hi dear

I want make a super cell of Phosphuren.I prepared unit cell for it.I wante execute convergence test for it.it make directory for this.but run very slow.after 24h just run for 1 directory.can you help me? and then can you say to me haw i can change my unit cell to super ceel?

Best regards,
sahel loushab

Re: running slow by sahel loushabsahel loushab, 08 Apr 2018 08:10
running slow
hegdemanuhegdemanu 21 Mar 2018 06:17
in discussion Forum / Support Questions » running slow

Hello,
I am running 72 (with 2*2*2 k-grid) atom supercell geometrical optimization. I am just doing the atomic position relaxation not the lattice parameter. I have done the complete relaxation in quantum espresso and it is converging .( I am just checking whether forces on atom converging on exciting from QE output). What parameters i should reduce when i am using supercell compared to unit cell on exciting?. any tips to speed up the calculation?
Thanks and Regards,
Manu

running slow by hegdemanuhegdemanu, 21 Mar 2018 06:17
Levi keller (guest) 19 Mar 2018 08:02
in discussion Forum / Support Questions » exciting carbon hanging after G_0W_0

EVALQP is the last file updated before GW_INFO and its contents seem reasonable. The failure to terminate, however, means that reserved computational resources are not freed for other users until I manually kill the process.

The compiler is gfortran (GCC 4.9.2) and mpif90 (OpenMPI 1.8.4).

the input file is:

<?xml version="1.0" encoding="utf-8"?>
<input>

<title>alpha-alumina</title>

<structure speciespath="$EXCITINGROOT/species">
<crystal scale="8.19773017300199">
<basevect> 0.634951269874242 0.000000000000000 1.0 </basevect>
<basevect> -0.317475634937121 0.549883929967765 1.0 </basevect>
<basevect> -0.317475634937121 -0.549883929967765 1.0 </basevect>
</crystal>
<species speciesfile="O.xml" rmt="1.4500">
<atom coord="0.2018805155 0.5923620044 0.8971212599"/>
<atom coord="0.8971212599 0.2018805155 0.5923620044"/>
<atom coord="0.5923620044 0.8971212599 0.2018805155"/>
<atom coord="0.0923620044 0.7018805155 0.3971212599"/>
<atom coord="0.3971212599 0.0923620044 0.7018805155"/>
<atom coord="0.7018805155 0.3971212599 0.0923620044"/>
</species>
<species speciesfile="Al.xml" rmt="2.0000">
<atom coord="0.0000000000 0.0000000000 0.0000000000"/>
<atom coord="0.7942425199 0.7942425199 0.7942425199"/>
<atom coord="0.2942425199 0.2942425199 0.2942425199"/>
<atom coord="0.5000000000 0.5000000000 0.5000000000"/>
</species>
</structure>

<groundstate
do="skip"
gmaxvr="13.0"
ngridk="6 6 6"
outputlevel="normal"
xctype="GGA_PBE_SOL">
</groundstate>

<gw
taskname="g0w0"
ngridq="2 2 2"
nempty="100"
ibgw="1"
nbgw="44"
>
<mixbasis
lmaxmb="3"
epsmb="1.0e-4"
gmb="1.0"/>
</gw>

</input>

by Levi keller (guest), 19 Mar 2018 08:02
Ioan (guest) 16 Mar 2018 15:49
in discussion Forum / Support Questions » LDA-1/2

I also think that a script for determining the optimum rCut for atoms would be useful. If I understand correctly, Fig. 2 of CPC 220, 263 (2017) does not account for SIC, it only shows _uncorrected_ LDA results for epsilon_HOMO; so, that paper did not show any concrete atomic IP values (well) agreeing with experiment.

Ioan

by Ioan (guest), 16 Mar 2018 15:49
Dmitrii (guest) 16 Mar 2018 08:40
in discussion Forum / Support Questions » exciting carbon hanging after G_0W_0

Hmm, I have never experienced this problem before. I'd say that your GW run was successful in case you have EVALQP.DAT file printed and it contains reasonable numbers inside.
Could you probably provide a bit more details: what are your compiler, mpi version, and input file?

Dmitrii
(the exciting code team)

by Dmitrii (guest), 16 Mar 2018 08:40

I have performed a gw cycle in the parallel version of exciting, and the last thing written to disk (two days ago) is the timing information in GW_INFO.OUT. The program is not terminating. 2 out of 20 processes are sleeping, and the other two are running. I would speculate that the two are involved in some sort of communication conflict. Can I assume that since the timing information is already printed, the computation has succeeded? Is there a detailed description of all the GW output somewhere? Let me know if you would like any additional diagnostic information.

exciting carbon hanging after G_0W_0 by Levi keller (guest), 15 Mar 2018 14:12
ticu (guest) 14 Mar 2018 21:32
in discussion Forum / Support Questions » LDA-1/2

Dear Ronaldo,

Could you please post here in forum a script or, better, add an extra tutorial to LDA-1/2 (similar to those for bulk Si or GaN) for an atom taken from Fig. 2 of your CPC 220, 263 (2017) showing how, at the rCut value maximizing the hl gap, the HOMO energy is in good agreement with experiment?

I am convinced that this would be useful for the community.

Many thanks in advance for taking your time,

ticu

PS. In calculations with exciting, I sometimes/often get " Warning(gndstate): Reached self-consistent loops maximum"
in WARNINGS.OUT. Browsing the documentation, I could not find ways to enforce SCF convergence (not even by increasing the number of cycles, although this rarely helps). Can you help please?

by ticu (guest), 14 Mar 2018 21:32
Ioan (guest) 14 Mar 2018 17:17
in discussion Forum / Support Questions » LDA-1/2

Dear Ronaldo,

I obtained rCut (values for Si and N as indicated my posting) by _maximizing_ the hl gap; calculations did yield a gap maximum.
If I increase rCut beyond the value that maximize hl (ignoring thereby the basic philosophy of LDA-1/2), the HOMO energy of the N atom remains nearly constant: -11.98 eV at rCut = 6 a.u. -12.03 eV at rCut = 8 a.u (largest rCut explored); so, the difference from 14.81 eV (PRB 2008) and 14.53 eV (exp.) is still substantial. (The N atom is one example of many I have examined.)

I understand that there are some differences between exciting and the method of PRB 2008, but I can hardly believe that they are responsible for the differences in IPs. I also checked that core freezing does not matter; so, I think that the usage of pseudopotentials (SIESTA) is not the issue here.

Rather, for reasons that I can give in a later posting, I am inclined to believe that this is a problem of implementation.

Best regards,

Ioan

by Ioan (guest), 14 Mar 2018 17:17
ronaldorpelaronaldorpela 14 Mar 2018 13:13
in discussion Forum / Support Questions » LDA-1/2

Dear Ticu, dear Ioan.

Actually the IPs of atoms published in this paper (PRB 78, 125116, 2008) have been obtained with spin-orbit coupling and with the code "atom", provided with siesta. Another point is that the IPs published in this paper were not obtained with LDA-1/2, but with the half-ionization technique of Slater. Probably, you did not get the difference between the methods, but they are not strictly the same.
In this paper (Computer Physics Communications 220, 263 (2017)), we employ LDA-1/2 and are able to obtain the same IPs with exciting and Quantum Espresso. If you want to calculate IPs of atoms using a code which employs periodic boundary conditions, then you should apply rcut as largest as you can (you can try to vary it, and you will see that your IPs will converge with a sufficiently large rcut). I would recomend to set rcut to L/2, where L is the size of the box you are using.

Best regards,
Ronaldo R. Pela

by ronaldorpelaronaldorpela, 14 Mar 2018 13:13
Ioan (guest) 13 Mar 2018 23:36
in discussion Forum / Support Questions » LDA-1/2

Dear Ticu,

I also failed to reproduce the values of the atomic IPs of the PRB paper you cited.

With the input parameters given below for the Si atom, the HOMO-LUMO gap reaches
the maximum value hl=5.582 eV at cut=5.60 a.u. The corresponding HOMO energy is -IP=-6.258 eV vs. -8.19 eV calculated in PRB 2008 and -8.15 eV from experiment.

For the N atoms, the values are: HOMO-LUMO gap maximum hl=11.059 eV at cut=4.269 a.u., the HOMO energy being -11.697 eV vs. -14.81 eV calculated in PRB 2008 and -14.53 eV from experiment.

A help clarifying this confusion is most desirable.

Input for Si:

<input>

<title>Si atom: LDA-1/2</title>

<structure speciespath="$EXCITINGROOT/species">
<crystal>
<basevect>14.0 0.0 0.0</basevect>
<basevect> 0.0 14.0 0.0</basevect>
<basevect> 0.0 0.0 14.0</basevect>
</crystal>
<species speciesfile="Si.xml">
<atom coord="0.0 0.0 0.0"/>
<dfthalfparam
cut="5.598"
ampl="1"
exponent="8">
<shell number="0" ionization="0.5" />
</dfthalfparam>
</species>
</structure>

<groundstate
do="fromscratch"
rgkmax="6.0"
gmaxvr="16"
ngridk="1 1 1"
outputlevel="high"
xctype="LDA_PZ">
<dfthalf printVSfile="false"/>
</groundstate>

</input

Input for N:

<input>

<title>N atom: LDA-1/2</title>

<structure speciespath="$EXCITINGROOT/species">
<crystal>
<basevect>14.0 0.0 0.0</basevect>
<basevect> 0.0 14.0 0.0</basevect>
<basevect> 0.0 0.0 14.0</basevect>
</crystal>
<species speciesfile="N.xml" rmt="1.05">
<atom coord="0.0 0.0 0.0"/>
<dfthalfparam
cut="4.269"
ampl="1"
exponent="8">
<shell number="0" ionization="0.5" />
</dfthalfparam>
</species>
</structure>

<groundstate
do="fromscratch"
rgkmax="7.0"
gmaxvr="20"
ngridk="1 1 1"
outputlevel="high"
xctype="LDA_PZ">
<dfthalf printVSfile="false"/>
</groundstate>

</input>

by Ioan (guest), 13 Mar 2018 23:36

I not find in this site about spin-orbit calculation with more details. So, I created one input with spin-orbit calculation and automatic radius (autormt) and maked the convergence test, without errors and files "warnings.out". But, when I calculated volume optimization, it have points off the curve. I changed the muffin-tin radius and some parameters, but the error continues. Please, what can the problem? Below is the original input:

<input>

<title>TlN - zinc-blende</title>

<structure speciespath="/home/user-fiscampus/exciting-code/exciting/species" autormt="true">

<crystal scale="9.975">
<basevect>0.0 0.5 0.5</basevect>
<basevect>0.5 0.0 0.5</basevect>
<basevect>0.5 0.5 0.0</basevect>
</crystal>

<species speciesfile="Tl.xml">
<atom coord="0.00 0.00 0.00"/>
</species>

<species speciesfile="N.xml">
<atom coord="0.25 0.25 0.25"/>
</species>

</structure>

<groundstate
do = "fromscratch"
stype="Methfessel-Paxton 1"
ngridk="4 4 4"
xctype="LDA_PW"
rgkmax="7.0"
swidth="0.0001"
nempty="40">

<spin
bfieldc="0.0 0.0 -1.0"
reducebf="0.5"
spinorb="true">
</spin>

</groundstate>

</input>

samad (guest) 07 Mar 2018 22:02
in discussion Forum / Support Questions » creat input of graphene in exciting code

If i want to creat a input in excinting code with a large supercel of graphene and one atome carbon (C
) how to do this ?
sincerely

by samad (guest), 07 Mar 2018 22:02
samad (guest) 07 Mar 2018 21:51
in discussion Forum / Support Questions » creat input of graphene in exciting code

Hi,
I have some problem with my input values of graphene if any one can help .
If anyone can can tell me how to find ibrav ,A,B,C cos AB,cosAC,cosBC
&SYSTEM
ibrav = 14 ,
A = 9.24049
B = 9.24049
C = 25,
cosAB = 0.5,
cosAC = 0,
cosBC = 0,
nat = 66,
ntyp = 1,
ecutwfc = 65 ,!!65
! ecutrho = 650 ,
nosym = .true. ,
occupations = 'smearing' ,
degauss = 0.005, !!!
smearing = 'marzari-vanderbilt' ,
! nspin = 1 ,
spline_ps = .true.

by samad (guest), 07 Mar 2018 21:51
page 1123...next »
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License