Activity for Elk

Miroslav Iliaš posted a comment on discussion Elk Developers

Miroslav Iliaš — Sun, 18 Oct 2026 07:33:32 -0000

Thanks for the reply. Yes, we want to study superheavy elements (up to Z=120,but also higher Z) with periodic DFT codes, and the all-electron Elk seems to be the good computational tool. The f90 files you provided are from the latest Elk 11.2.3, right ? Would be good to have more verbose output when generating the species files for new SHEs.

Peter Elliott posted a comment on discussion Elk Developers

Peter Elliott — Sun, 18 Oct 2026 00:18:28 -0000

From your output it doesn't look like it crashed, it just gave a warning. This warning only comes from the first pass, which uses a bad guess for the occupations, but after that it finds the right occupations and behaves as it should. So the generated species file is more or less fine. There is a subtlety that it since it didn't occupy the n=7 shell in the first pass, it fixes the radial mesh smaller than it ideally would. Probably it would be safer to use the attached genfspecies.f90 file, that...

Miroslav Iliaš posted a comment on discussion Elk Developers

Miroslav Iliaš — Sat, 17 Oct 2026 21:44:34 -0000

Greetings, dear Elk developers, I am trying to generate the Cn (Z=112) species file, but it crashes - see below. Input, output files are attached. Elk is installed within the conda channel. Best, Miro (elk_env) miroi@MIRO :~/work/projects/open-collection/theoretical_chemistry/software/elk-ase/runs/Cn/simple_run/generate_Cn_species_file/.run_elk.bash ┌─────────────────────────────────┐ │ Elk code version 10.2.4 started │ └─────────────────────────────────┘ Warning(atom): maximum iterations exceeded...

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Wed, 14 Oct 2026 12:28:29 -0000

Hi, The GW calculation in Elk is one-shot G0W0 in the Matsubara frequency representation. It is very important to converge all the relevant parameters for the GW calculation. Here is an example for Ni which is properly converged: tasks 0 10 600 610 630 spinpol .true. bfieldc 0.0 0.0 0.1 reducebf 0.85 highq .true. ! Matsubara frequency cut-off wmaxgw 15.0 ! response function G-vector cut-off gmaxrf 3.5 ! temperature of many-body system (determines Matsubara frequency spacing) tempk 1000.0 wplot 3000...

Marks posted a comment on discussion Elk Developers

Marks — Mon, 12 Oct 2026 09:03:43 -0000

Dear Elk developers, I am interested in performing a GW calculation using the Elk code, and I would like to clarify how the GW implementation works and how to perform a one-shot (G0W0) calculation. I have been going through the Elk user guide and found the following GW-related tasks: 600 - Output the GW self-energy matrix elements. 610 - Generates the GW spectral function. 620 - Generates the GW band structure, i.e. the k-dependent spectral function. 630 - Writes the GW Fermi energy to GWEFERMI.OUT....

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Mon, 14 Sep 2026 12:47:58 -0000

Further update: I've contacted Susi Lehtola and he requests that you file a ticket on the Libxc Gitlab tracker with a reproducer.

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Mon, 14 Sep 2026 11:23:08 -0000

Update: I've now reproduced the issue by using Libxc version 7.1.2. This version does not fail gracefully with negative sigma for r2SCAN like version 7.0.0. This should be fixed in Libxc but we may have make a work-around in Elk for the time being. Thanks and regards, Kay.

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Mon, 14 Sep 2026 09:35:12 -0000

Hi Ricardo, I ran your example and it works fine for me. I've attached my INFO.OUT file. Perhaps your stack size was insufficient. Try using the script elk/elk.sh to run the code: #!/bin/bash export OMP_NUM_THREADS=32 export OMP_PROC_BIND=false export OMP_STACKSIZE=512M ulimit -Ss 524288 ~/elk/src/elk Best wishes, Kay.

Ricardo Ribeiro posted a comment on discussion Elk Developers

Ricardo Ribeiro — Mon, 14 Sep 2026 09:21:41 -0000

Elk version: 11.2.3 (stock, unmodified) Libxc version: 7.1.2 Platform: Linux x86-64, OpenMPI 5 + OpenMP 1. Summary With any Libxc meta-GGA correlation functional, a symmetrised collinear antiferromagnet whose lattice vectors are orthogonal produces NaN in the exchange-correlation potential from SCF loop 1, and hence a NaN total energy. The cause is a catastrophic cancellation in the total-density gradient invariant that Libxc reconstructs from the three spin-resolved invariants Elk passes to it....

J. K. Dewhurst created a blog post

J. K. Dewhurst — Fri, 28 Aug 2026 10:29:48 -0000

Elk version 11.2.3 released

Elk released /elk-11.2.3.tgz

Elk — Fri, 28 Aug 2026 10:27:02 -0000

Long Yu modified a comment on discussion Elk Users

Long Yu — Fri, 21 Aug 2026 15:34:58 -0000

Dear Kay, A more sophisitcated way would be to use TDDFT with a constantly increasing vector potential, E=-1/c dA(t)/dt. Has this function been implemented now? Thank you. Regards, Long

Long Yu posted a comment on discussion Elk Users

Long Yu — Fri, 21 Aug 2026 15:33:49 -0000

Dear Kay, A more sophisitcated way would be to use TDDFT with a constantly increasing vector potential, E=-1/c dA(t)/dt. Has this function been implemented now? Regards, Long

Zixuan Li posted a comment on discussion Elk Users

Zixuan Li — Sat, 15 Aug 2026 01:36:20 -0000

Dear all I've tested the SCF and TDDFT successive two-step calculation with or without setting the reducek tag in the SCF part and the results of them are different. Then I've checked the INFO.OUT produced by the two calculation, it shows that the kpoint reduction seems to be closed automatically in the TDDFT process regardless the reducek tag is set or not. So, reasonably, I prefer that setting 'reducek = 0' in the SCF part may be the right one. By the way, I've search the forum, and discovery that...

Jerry Li modified a comment on discussion Elk Users

Jerry Li — Sun, 09 Aug 2026 04:21:00 -0000

Dear All, Cupric oxide (CuO) is a p-type semiconductor with an indirect band gap reported to be in the range of 1.2–1.9 eV. It has strong electron correlations, and antiferromagnetic ground state. I have tried both DFT+U and hybrid-functional approaches, but neither method is able to reproduce the experimental band gap correctly, as shown in the attached screenshot. The ELK input files used for these calculations are also attached for reference. I would greatly appreciate any suggestions or insights...

Jerry Li modified a comment on discussion Elk Users

Jerry Li — Sun, 09 Aug 2026 03:52:29 -0000

Dear All, Cupric oxide (CuO) is a p-type semiconductor with an indirect band gap reported to be in the range of 1.2–1.9 eV. I have tried both DFT+U and hybrid-functional approaches, but neither method is able to reproduce the experimental band gap correctly, as shown in the attached screenshot. The ELK input files used for these calculations are also attached for reference. I would greatly appreciate any suggestions or insights regarding the possible reasons for this discrepancy and recommendations...

Jerry Li posted a comment on discussion Elk Users

Jerry Li — Sun, 09 Aug 2026 03:50:41 -0000

marcindulak modified a comment on discussion Elk Developers

marcindulak — Sun, 19 Jul 2026 05:20:26 -0000

I'm sorry for the noise. I noticed my elk was not latest. elk-11.0.2 with LIB_LIBXC = -L%{_libdir} -lxcf03 -lxc and SRC_LIBXC = libxcifc.f90 builds with f03 libxc-7.1.0.

marcindulak posted a comment on discussion Elk Developers

marcindulak — Sun, 19 Jul 2026 05:05:29 -0000

I'm sorry for the noise. I noticed my elk was not latest. elk-11.0.2 with LIB_LIBXC = -L%{_libdir} -lxc and SRC_LIBXC = libxcifc.f90 builds with f03 libxc-7.1.0.

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Sat, 18 Jul 2026 19:42:16 -0000

The f03 interface.

marcindulak posted a comment on discussion Elk Developers

marcindulak — Sat, 18 Jul 2026 19:36:46 -0000

OK, so it's related to packaging libxc, as hinted above. I don't find libxc f90 modules packaged on Fedora. Are you using f90 or f03 interface? rpm -ql libxc-devel | grep modules /usr/lib64/gfortran/modules/xc_f03_funcs_m.mod /usr/lib64/gfortran/modules/xc_f03_lib_m.mod

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Sat, 18 Jul 2026 19:14:16 -0000

Hi Marcin, Libxc 7.0.0 has worked fine with Elk for over a year now (since version 10.5.16). Regards, Kay.

marcindulak posted a comment on discussion Elk Developers

marcindulak — Sat, 18 Jul 2026 18:32:34 -0000

Hi, What's the status of the libxc7 support in elk? I see there is a new https://gitlab.com/libxc/libxc/-/tags/7.1.0, but I see no mention of xc_f03 in the elk source: grep xc_f03 -r elk-10.1.12 The patch to switch to f03 https://src.fedoraproject.org/rpms/elk/blob/962f050def3391e3b2ba943d9148f262afc64736/f/elk-9.2.12-libxc7.patch is rather large, and does not apply anymore.

CY ZHUNG modified a comment on discussion Elk Users

CY ZHUNG — Tue, 07 Jul 2026 04:07:20 -0000

Yes. but that is had done. i do learn to use three DFT tool suits like QE,CASTEP. Read the make.inc and do some modification is an easy work. Any more suggestion?

CY ZHUNG posted a comment on discussion Elk Users

CY ZHUNG — Mon, 06 Jul 2026 20:48:35 -0000

Yes. but that is had done. i do learn to use three DFT tools suit like QE,CASTEP. Read the make.inc and do some modification is an easy work. Any more suggestion?

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Mon, 06 Jul 2026 08:44:36 -0000

OK. We notice that sometimes users overwrite the latest make.inc file with one from an older version. This can cause problems. In your make.inc file you should have the following lines uncommented: LIB_LIBXC = libxcf03.a libxc.a SRC_LIBXC = libxcf03.f90 libxcifc.f90 (Note the f03 convention rather than f90) Is this correct?

CY ZHUNG modified a comment on discussion Elk Users

CY ZHUNG — Mon, 06 Jul 2026 01:03:15 -0000

"nm libxc.a | grep libxc_free" gets the flags but "nm libxcf03.a | grep libxc_free" gets nothing on Ubuntu 26.04. U libxc_free_flags U libxc_free_flags 0000000000000690 T libxc_free_flags The flags are almost the same on my archlinux laptop except the last digtal number in "0000……*** T libxc_free_flages"

CY ZHUNG posted a comment on discussion Elk Users

CY ZHUNG — Mon, 06 Jul 2026 01:02:55 -0000

"nm libxc.a | grep libxc_free" get the flags but "nm libxcf03.a | grep libxc_free" get nothing on Ubuntu 26.04. U libxc_free_flags U libxc_free_flags 0000000000000690 T libxc_free_flags The flags are almost the same on my archlinux laptop except the last digtal number in "0000……*** T libxc_free_flages"

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Sun, 05 Jul 2026 21:07:03 -0000

Could you try: nm libxc.a | grep libxc_free and nm libxcf03.a | grep libxc_free in the elk/src directory and see if libxc_free exists.

CY ZHUNG modified a comment on discussion Elk Users

CY ZHUNG — Sun, 05 Jul 2026 09:24:16 -0000

I have checked. gnu libxc has not been installed as the attachment shows. Libxc was compiled with cmake and the lib files have been cp to folder elk-./src. Do you have any other suggestion？ attachments: sudo apt remove libxc-dev [sudo: authenticate] Password: Package 'libxc-dev' is not installed, so not removed

CY ZHUNG posted a comment on discussion Elk Users

CY ZHUNG — Sun, 05 Jul 2026 09:23:15 -0000

I have checked. gnu libxc has not been installed as the attachment shows. Libxc was compiled with cmake and the lib files have been cp to folder elk-./src. Do you have any other suggestion？

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Sun, 05 Jul 2026 08:44:10 -0000

Hi, I think it's because Ubuntu 26.04 has an old version of Libxc (5.2.3). You'll have to download the latest Libxc source from here and compile it yourself. Regards, Kay.

CY ZHUNG posted a comment on discussion Elk Users

CY ZHUNG — Sat, 04 Jul 2026 11:44:10 -0000

Dears. ELK-11.02 seems could not read libxcfc03.o on Ubuntu 26.04 LTS. Both intel oneapi and gnu compiler+mkl have the problem when compiling. Could you any help? Attachment from gnu compiler+mkl f03.a libxc.a libwannier.a /usr/bin/x86_64-linux-gnu-ld.bfd: libxcf03.o: in function __xc_f03_lib_m_MOD_xc_f03_functional_get_name': libxcf03.f90:(.text+0xa1e): undefined reference tolibxc_free' /usr/bin/x86_64-linux-gnu-ld.bfd: libxcf03.f90:(.text+0xa9f): undefined reference to `libxc_free' collect2: error:...

J. K. Dewhurst posted a comment on discussion Elk Developers

J. K. Dewhurst — Tue, 30 Jun 2026 19:17:57 -0000

Dear Miroslav, That's quite easy to do, just use the fractional species block fspecies. Here is an example which finds the optimal unit cell volume of flerovium: tasks 2 ! structural optimisation ! fractional species fspecies -114.0 'Fl' ! high-quality calculation highq .true. ! no conduction band local-orbitals lorbcnd .false. ! use LAPW instead of APW+l.o. nxoapwlo 1 ! optimise the lattice parameters latvopt 1 ! no optimisation of atomic positions atpopt 0 avec 1.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 1.0...

Miroslav Iliaš posted a comment on discussion Elk Developers

Miroslav Iliaš — Tue, 30 Jun 2026 17:09:24 -0000

Dear Elk-developers, I see elk-11.0.2/species/* are up to Z=104 (element Rf). Would it be please possible to have constructed species up to Z=120 ? The pdf manual has a paragraph called "4.4.2 Species files", but without detailed instructions... Any help is appreciated.

Jerry Li posted a comment on discussion Elk Users

Jerry Li — Thu, 25 Jun 2026 16:21:19 -0000

128 threads

Kai Luo posted a comment on discussion Elk Users

Kai Luo — Sun, 14 Jun 2026 07:40:50 -0000

Thanks for testing. How many threads did you use? I tried 8 threads runs fine but could not run beyond 20 outer macro iterations.

Jerry Li modified a comment on discussion Elk Users

Jerry Li — Fri, 12 Jun 2026 15:05:03 -0000

I tested the RDMDFT/NiO example with Latest version: 11.0.2. No problem at all!

Kai Luo posted a comment on discussion Elk Users

Kai Luo — Fri, 12 Jun 2026 11:37:16 -0000

I haven't tried it, but I guess rdmft part is not changed at all.

Jerry Li posted a comment on discussion Elk Users

Jerry Li — Thu, 11 Jun 2026 20:42:29 -0000

No problem with Latest version: 11.0.2

Kai Luo posted a comment on discussion Elk Users

Kai Luo — Wed, 10 Jun 2026 02:35:04 -0000

Dear All, I have recently tried to use RDMFT calculation with version 10.9.5. Using the example input for NiO, I encountered the following error. I used 1 rank with 16 threads and it finishes the Kohn-Sham task and gives two RDMC_ENERGY.OUT then crashed. Could anyone give suggestions on where it goes wrong? Thanks! 1 -1540.783177 2 -1538.946974 forrtl: severe (174): SIGSEGV, segmentation fault occurred Image PC Routine Line Source libpthread-2.28.s 00001502A6AC5990 Unknown Unknown Unknown elk 0000000001701221...

J. K. Dewhurst created a blog post

J. K. Dewhurst — Fri, 05 Jun 2026 16:53:41 -0000

Elk version 11.0.2 released

Elk released /elk-11.0.2.tgz

Elk — Fri, 05 Jun 2026 14:21:03 -0000

Abderrahmane Reggad posted a comment on discussion Elk Developers

Abderrahmane Reggad — Wed, 03 Jun 2026 18:39:23 -0000

Dear Elk Developers, I am a teacher and researcher and I have created two modified versions of Elk 3.3.17: elkpdg - a pedagogical copy designed to teach ab initio calculations step by step, using 3 simplified input files. elkader - a research-oriented modification with 3 input files and two Python scripts: one that merges the 3 input files into a standard elk.in, and another that converts all numerical parameters into human-readable form. The algorithms and input file structure are my own design....

MUHTASIM MINHAZ posted a comment on discussion Elk Users

MUHTASIM MINHAZ — Sun, 17 May 2026 08:04:31 -0000

Dear ELK developers and users, I am calculating the spontaneous electric polarization of NiI₂, a spin-spiral multiferroic, using ELK version 10.9.5. My setup uses spinsprl .true. with a spin-spiral q-vector, DFT+U on Ni (FLL, U=4.2 eV, J=0.8 eV), GGA, spin-orbit coupling, and a 1×1×1 primitive cell. After reading the full task list and grepping the source code, I found that ELK does not appear to have a standalone task for computing spontaneous Berry phase polarization. The subroutine polar() in...

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Mon, 04 May 2026 20:57:07 -0000

Hi, If reducebf < 1 then bfieldc and bfcmt are automatically reduced to zero for all tasks except for 0,2,28,200,208,350,380,420,421,440. In other words, if reducebf < 1 then you don't need to do anything if you want zero fields (i.e. when they are just for symmetry breaking). If the fields are supposed to be finite (like an external magnetic field) then reducebf should be set to exactly 1 and the RT-TDDFT calculation will include them. Regards, Kay.

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Mon, 04 May 2026 20:48:35 -0000

Hi, You're quite correct: the formula on page 75 is for an earlier version, and also doesn't have the chirp part. I'll fix that for the next release. Thanks and regards, Kay.

inho lee posted a comment on discussion Elk Users

inho lee — Sun, 03 May 2026 23:17:00 -0000

Dear developers, In the manual the Elk code manual, the formulas of vectorpotential in[5.99 pulse] (page 34) and in [7.48 genafieldt] (page 75) looks inconsistent. I am not sure whether the A_0 is divided by (sigma)(sqrt(2pi)) or not. When I change FWHM of pulse, [AFIELD.OUT] did not changed the amplitude, therefore, I think A-0 is not divided by (sigma)(sqrt(2pi)). Would you check the definition of A(t) is consistent in the code?, Also, would you check the peak laser pulse intensity written in [TD-INFO.OUT]...

inho lee posted a comment on discussion Elk Users

inho lee — Sun, 03 May 2026 21:31:12 -0000

Dear All, I would like to ask about [bfieldc] and [bfcmt] variables. I am investigating RT-TDDFT calculation of ferromagnetic CrI3 monolayer, with spin orbit coupling I used small bfieldc value for ground state calculation [tasks=0]. My questions is, Are bfieldc or bfcmt included in RT-TDDFT calculation? (using tasks=460) (I think that the equation1 in https://pubs.acs.org/doi/pdf/10.1021/acs.nanolett.7b05118 means external B-field is included) If so, do I have to remove bfieldc or bfcmt variables...

Chamara Somarathna modified a comment on discussion Elk Users

Chamara Somarathna — Thu, 30 Apr 2026 15:16:37 -0000

Hi Kay, I really appreciate you helping me out a second time. I’m incredibly grateful for your time and expertise With your suggestions of setting lorbcnd=.false. and reduced rgkmax, I was able to get rid of, Warning(rhonorm): total charge density incorrect and Warning(occupy): minimum eigenvalue less than minimum linearisation energy But I still get the warning, Warning(linengy): could not find 1 linearisation energies in all of the sc loops. Did you also see this and which version of Elk do you...

Chamara Somarathna posted a comment on discussion Elk Users

Chamara Somarathna — Thu, 30 Apr 2026 15:15:40 -0000

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Thu, 30 Apr 2026 07:45:48 -0000

Hi Chamara, It's best not to enable both nxoapwlo=1 and lorbcnd=.true. as this can lead to instability in the first-variational eigenvalue problem. Also having rgkmax too high can cause similar instabilities. Here's my input file that works for the structure at high compression: tasks 0 xctype 3 vhighq .true. ngridk 18 18 18 stype 0 swidth 0.001 rgkmax 8.5 gmaxvr 32.0 lmaxi 2 nxoapwlo 1 lorbcnd .false. avec 5.058464396264 0.000000000000 0.000000000000 -0.000000000000 5.058464396264 0.000000000000...

Chamara Somarathna posted a comment on discussion Elk Users

Chamara Somarathna — Sun, 26 Apr 2026 15:10:05 -0000

Hi all, Elk version used: 10.8.16 I am a new Elk user and your help is very much appriciated. I try to do a ground-state calculation (task 0) for the B1 crystal stucture of MgO at very high compression (~75% compression). But I keep get the following errores in all the sc loops, Warning(rhonorm): total charge density incorrect for s.c. loop 13 Calculated : 79.86817685 Required : 80.00000000 and Warning(linengy): could not find 1 linearisation energies in s.c. loop 11 Warning(occupy): minimum eigenvalue...

Long Zhang modified a comment on discussion Elk Users

Long Zhang — Fri, 24 Apr 2026 20:17:08 -0000

Dear Michael and Kay, Thank you both for all of your inputs! I think I have been brought to the point where the rest of the job is to fine tune the calculation. I'll bring up any other issue, if any. Apologies for the delayed response. Best, Long

Long Zhang posted a comment on discussion Elk Users

Long Zhang — Fri, 24 Apr 2026 16:49:51 -0000

Dear Michael and Kay, Thank you both for all of your inputs! I think I have been brought to the point where the rest of my job is to fine tune the calculation. I'll bring up any other issue, if any. Sorry for my delayed response. Best, Long

Chamara Somarathna posted a comment on discussion Elk Users

Chamara Somarathna — Wed, 22 Apr 2026 15:46:21 -0000

Hi Kay, Thank you very much for your very detailed reply with supportive calculations. This is helping me to go farward solving an issue I had for a week. I will do the batch calculation as you pointed out. From the task 440, I was calculating the pressure in the wrong way. As you pointed out, the stress value given by Elk is ∂E/∂t, but I thought it was the ∂E/∂V. So again, thank you verymuch for your tremendous help. Best, Chamara

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Wed, 22 Apr 2026 12:52:27 -0000

(continued) You can also determine the pressure with task=440 although this is not quite as accurate as fitting to an EOS. Here is the input file I used: tasks 440 wrtdisk .false. deltast 0.001 xctype 3 vhighq .true. lorbcnd .false. nxoapwlo 1 rgkmax 10.0 gmaxvr 32.0 lmaxi 2 ngridk 24 24 24 avec 7.818401918226 0.000000000000 0.000000000000 0.000000000000 7.818401918226 0.000000000000 0.000000000000 0.000000000000 7.818401918226 atoms 2 : nspecies 'Mg.in' : spfname 4 : natoms 0.0 0.0 0.0 0.0 0.5 0.5...

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Wed, 22 Apr 2026 12:29:06 -0000

Hi Chamara, This type of calculation has to be converged very well to get reliable pressures. The first and most reliable way of performing the calculation is to calculate an energy-vs-volume curve and fit it to an equation of state (EOS). Here is the input file I used for this: tasks 0 1 1 1 1 1 1 1 1 1 1 batch .true. wrtdisk .false. trmt0 .false. xctype 3 vhighq .true. lorbcnd .false. nxoapwlo 1 rgkmax 10.0 gmaxvr 32.0 lmaxi 2 ngridk 24 24 24 avec 7.818401918226 0.000000000000 0.000000000000 0.01...

Chamara Somarathna posted a comment on discussion Elk Users

Chamara Somarathna — Tue, 21 Apr 2026 14:59:02 -0000

Hi, I am new to Elk and trying to compare the VASP pseudopotentials pressure vs Elk all-electron pressures. I started with a simple B1 unit-cell of MgO which gives me about 1 GPa pressure in VASP. However, when I use Elk for the same geometry (converted to Bohr), the STRESS.OUT gives me about 1900 GPa which is not correct. I converted the Elk pressure using 1 Hartree/Bohr^3 = 29421.03 GPa. I would greatly appreciate it if someone could help me to understand this inconsistency. Here is my elk input...

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Mon, 13 Apr 2026 20:32:09 -0000

Dear Long and Michael, This is quite difficult to converge. I think Michael is correct using mixtype=1 and rgkmax=7.5. I use task=28 to calculate the MAE. This is because it also rotates the spherical covering used for the spherical harmonic transform which eliminates some error. Here is my input file: tasks 28 npmae -1 mixsave .true. wrtdisk .false. epsengy 1.e-6 epspot 1.e-8 spinorb .true. bfieldc 0.0 0.0 0.5 reducebf 0.85 ngridk 24 24 24 rgkmax 7.5 gmaxvr 18.0 lmaxapw 9 lmaxo 8 xctype 20 mixtype...

mfechner posted a comment on discussion Elk Users

mfechner — Fri, 10 Apr 2026 14:11:33 -0000

excuse the delay here is my input see the reducebf ! tasks 0 maxscl 200 nwrite 50 epsengy 1.e-6 epspot 1.e-8 spinpol .true. spinorb .true. sppath '' ngridk 8 8 8 rgkmax 7.5 gmaxvr 18.0 lmaxapw 9 lmaxo 8 xctype 20 0 0 mixtype 1 reducebf 0.95 dft+u 1 1 : dftu, inpdftu 1 3 0.294 0.000 : is, l, U, J !! U on Tb-f, 0.294Ha = 8eV !rotavec ! 0 1 0 270 !! rotation of 270 degrees around +y makes a1 on +z avec -0.1716939763E-14 0.000000000 9.346584740 0.8584698813E-15 8.094379823 -4.673292370 -7.540006694 0.000000000...

J. K. Dewhurst created a blog post

J. K. Dewhurst — Thu, 09 Apr 2026 14:22:04 -0000

Elk version 10.9.5 released

Elk released /elk-10.9.5.tgz

Elk — Thu, 09 Apr 2026 14:20:04 -0000

Long Zhang modified a comment on discussion Elk Users

Long Zhang — Wed, 01 Apr 2026 11:20:33 -0000

Dear Michael, I did not get that Warning about linearisation energy when I used rotavec to rotate. But I failed to reproduce your success when I directly use the rotated structure copied from GEOMETRY.OUT. I also tried rgkmax 7 and mixtype 1, still no luck. Here is the elk.in I used, with structure copied from GEOMETRY.OUT. tasks 0 maxscl 200 nwrite 50 epsengy 1.e-6 epspot 1.e-8 spinpol .true. spinorb .true. reducebf 0.95 sppath './species/' ngridk 24 24 24 rgkmax 8.0 gmaxvr 18.0 lmaxapw 9 lmaxo...

Long Zhang posted a comment on discussion Elk Users

Long Zhang — Wed, 01 Apr 2026 11:19:23 -0000

Dear Michael, I did not get that Warning about linearisation energy when I used rotavec to rotate. But I failed to reproduce your success when I used the rotated structure copied from GEOMETRY.OUT. I also tried rgkmax 7 and mixtype 1, still no luck. Here is the elk.in I used, with structure copied from GEOMETRY.OUT. tasks 0 maxscl 200 nwrite 50 epsengy 1.e-6 epspot 1.e-8 spinpol .true. spinorb .true. reducebf 0.95 sppath './species/' ngridk 24 24 24 rgkmax 8.0 gmaxvr 18.0 lmaxapw 9 lmaxo 8 xctype...

mfechner posted a comment on discussion Elk Users

mfechner — Tue, 31 Mar 2026 10:24:11 -0000

two more small things to recommend: you should switch on reducebf, which decreases the applied magnetic field as you want to have at the end a computation without applied field. A value of 0.95 is good in most cases. In addition switching to linear mixing sometimes keep things also more stabe!

mfechner posted a comment on discussion Elk Users

mfechner — Tue, 31 Mar 2026 10:22:23 -0000

Dear Long, I tried your case and can indeed confirm the issue. However, could you check whether, when you rotate, there are any messages in the standard output files such as: "Warning(occupy): minimum eigenvalue less than minimum linearisation energy : -44.32280382 -4.383400000 for s.c. loop 137" This suggests that there is an issue when you rotate. Instead of rotating the structure using Elk, I tried directly using the rotated structure. (You can find it in GEOMETRY.OUT and directly copy it into...

Long Zhang posted a comment on discussion Elk Users

Long Zhang — Mon, 30 Mar 2026 20:34:12 -0000

Dear ELK users and developers, I am studying magnetic TbCo5 using dft+u, with the current version 10.8.16 of ELK. The elk.in is copied at the end of this message. I am basically doing the MAE thing for this 4f SOC system, but with bfcmt instead of bfieldc used in the given mae example, because I want to get the afm in unit cell state. The lattice vectors are like this: a1 and a2 in xy-plane and 120 degrees away, a3 on z-axis. |a1|=|a2|=a, |a3|=c. I managed to converge it when I set magnetic direction...

Markus modified a comment on discussion Elk Users

Markus — Fri, 27 Mar 2026 09:59:42 -0000

Yes, there is an example for CaO in the examples folder. You need the core states as valence, i.e. you have to model them with local orbitals. There is a special species file for Ca in the example which shows you how to do it for the 2p states. These calculations are extremely expensive and need lots of RAM. Carefully optimize the initial and final states to keep the matrix size as small as possible and start with a very small number of k-points. XMCD is also possible, compare with the XMCD example....

Markus posted a comment on discussion Elk Users

Markus — Fri, 27 Mar 2026 09:58:16 -0000

Krishnendu Mukherjee posted a comment on discussion Elk Users

Krishnendu Mukherjee — Thu, 26 Mar 2026 14:58:36 -0000

Dear All, Has anybody modelled the coherence length of superconductor using elk code. Any guidance will be greatly appreciated. Greetings, Krishnendu

Jerry Li posted a comment on discussion Elk Users

Jerry Li — Thu, 26 Mar 2026 04:04:29 -0000

Hello everyone, I am interested in computing X-ray Magnetic Circular Dichroism (XMCD) and X-ray Absorption Near Edge Structure (XANES) for magnetic semiconductors. Has anyone performed Elk calculations of X-ray absorption spectra (XAS) using the Bethe–Salpeter equation (BSE)? I would greatly appreciate it if you could share an example of your input files. Thank you in advance!

Hanley Coath posted a comment on discussion Elk Users

Hanley Coath — Tue, 24 Mar 2026 17:09:23 -0000

Hello Markus, Thanks for pointing me towards the code, it was very useful.

Markus posted a comment on discussion Elk Users

Markus — Tue, 24 Mar 2026 16:33:49 -0000

Hi Hanley, you can find it in the code, start from occupy.f90: https://elk.sourceforge.io/doxygen/html/occupy_8f90_source.html Then look for stheta.f90 and stheta_fd.f90: https://elk.sourceforge.io/doxygen/html/stheta_8f90_source.html and https://elk.sourceforge.io/doxygen/html/stheta__fd_8f90_source.html The argument of the standard Fermi-Dirac distribution is (efermi-e)/swidth, see line 64 of occupy.f90. This is the distribution, then the approximation to the smooth delta function is specified...

Hanley Coath posted a comment on discussion Elk Users

Hanley Coath — Tue, 24 Mar 2026 12:06:13 -0000

In the elk manual, it says that the default function used when calculating the smooth approximation to the Dirac-Delta function (swidth), uses a Fermi-Dirac distribution. What is the exact form of the distribution used in the elk code? It doesn't seem to be referenced in the manual. Is it the standard Fermi-Dirac distribution? ie. f(z) = 1 / (1 + exp z−μ/kBT) or is it some other approximation of this function (such as: https://link.springer.com/article/10.1007/s11075-010-9397-6). If anyone knows,...

J. K. Dewhurst created a blog post

J. K. Dewhurst — Mon, 23 Feb 2026 14:46:43 -0000

Elk version 10.8.16 released

Elk released /elk-10.8.16.tgz

Elk — Mon, 23 Feb 2026 14:45:03 -0000

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Wed, 11 Feb 2026 23:10:41 -0000

Hi All, Don't forget to use primcell=.true. after you've set up your magnetic fields. Elk will then find the smallest unit cell which produces the same crystal. For example tasks 0 primcell .true. spinpol .true. ngridk 2 2 2 avec 9.230000000 0.000000000 0.000000000 0.000000000 9.230000000 0.000000000 0.000000000 0.000000000 18.46000000 atoms 2 : nspecies 'U.in' : spfname 8 : natoms; atposl, bfcmt below 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.10000000 0.00000000 0.00000000 0.50000000...

Hanley Coath posted a comment on discussion Elk Users

Hanley Coath — Wed, 11 Feb 2026 15:40:20 -0000

Hi Markus, thanks for the response. I had literally just figured this out this morning when I opened my structure file with xcrysden, and was quite surprised to see a UN2 structure. Definitely a quick fix. Cheers

Markus posted a comment on discussion Elk Users

Markus — Wed, 11 Feb 2026 14:26:53 -0000

Hi, your input is wrong, that is not UN, you made UN2 (with the wrong lattice constant). Replace the N position with 0.5 0.0 0.0, that gives you the correct structure and stoichiometry. Cheers, Markus

Hanley Coath posted a comment on discussion Elk Users

Hanley Coath — Mon, 09 Feb 2026 17:13:25 -0000

Hi all, I am currently trying to use the spacegroup function to generate a lattice for a type-I, afm, fm-3m (fcc) structure (UN). I tried generating 2 unit cells in the z-direction, and then manually changing the b-fields on the individual U atoms so that they would be alternating with planes, but this didn't seem to work. I also tried before to just use the atoms block to do this, folowing the example given in the DFT+U examples for NiO, but I couldn't figure out how to translate this format to...

Markus posted a comment on discussion Elk Users

Markus — Fri, 06 Feb 2026 16:48:22 -0000

Hi, did you comment this line: #SRC_W90S = w90_stub.f90 above the lines you have to uncomment? Is Wannier90 compiled with the same compiler? Cheers, Markus

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Fri, 06 Feb 2026 09:59:41 -0000

Hi, I'm not sure what the problem is. Elk should work fine with Wannier90 3.1.0. You could try using GFortran instead. Regards, Kay.

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Fri, 06 Feb 2026 09:57:03 -0000

Dear Hong, You can use task=320 and specify the q-vector with ! q-vector in lattice coordinates vecql 0.5 0.5 0.5 See elk/examples/TDDFT-optics/diamond-ALDA/. It is also possible to calculate the magnetic response at finite q using task 330. See elk/examples/TDDFT-magnetic-response/Ni-magnetic-response/ . Regards, Kay.

Markus posted a comment on discussion Elk Users

Markus — Fri, 06 Feb 2026 09:06:47 -0000

Dear Hong Tang, 2D systems are generally quite demanding in terms of convergence with GGA and mGGA. Instead of playing with all the basis set and numerics parameters by hand, I would recommend to try highq=.true. and vhigh=.true. first. There can be discontinuities in the density at the MT boundaries, which can be resolved with tighter basis set parameters. These spoil the gradient calculation and lead to numerical instability. Also, try with smaller k-point grid first, so you get a feeling for the...

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Wed, 04 Feb 2026 06:44:55 -0000

Dear ELK team, Could you please help em with this question? I posted it a while ago. Appreciate your help very much! Hong

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Wed, 04 Feb 2026 06:43:29 -0000

Dear ELK team, Could you please help me with this question? I posted it already a while ago. Appreciate your help very much! Best, Hong

Elk released /elk-10.8.12.tgz

Elk — Wed, 28 Jan 2026 17:37:04 -0000

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Fri, 23 Jan 2026 06:07:08 -0000

Dear ELK team, I just use ELK 10.8.1 for TDDFT calculation. I tried the Bulk LiF example. It worked. It seems to me the calculation is only for q=0 case, the optical case, the vertical excitation. I want to calculate EELS with different q, the non-zero momentum transfer cases. q is the difference of k between the valence state and the excited conducting state, Could you please let me know how to set the input for this kind of calculations in ELK? I couldn't find this information in the ELK manual....

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Fri, 23 Jan 2026 05:57:36 -0000

Yes, thank you so much, Markus. libxc 7.0.0 worked for elk 10.8.1.

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Fri, 23 Jan 2026 05:54:56 -0000

Dear ELK team, I am using ELK.10.8.1 to do a TDDFT calculation. For LSDA, my job can run smoothly and finish the Bootstrap kernel TDDFT calculation very quick 2-5 minutes on 1Node with 32 cores. However, when I use PBE, PBEsol, or r2SCAN (through Libxc.7.0.0, I find even for the ground state calculation "tasks =0", my job cannot converge even for over 17 hours. Please see my elk.in input below. Please also see my attached INFO.OUT file. You can see " RMS change in Kohn-Sham potential (target) Absolute...

jianyong posted a comment on discussion Elk Users

jianyong — Fri, 23 Jan 2026 05:38:00 -0000

Hi, I've installed elk-10.8.1 and elk-10.5.16 with libwannier.a copied in the src folder, and I uncomment the lines: SRC_W90S = LIB_W90 = libwannier.a My wannier90 version is 3.1.0 and it works with VASP. The Elk code was successfully compiled but the wannier90 interface cannot work. The out put of ldd elk is: linux-vdso.so.1 (0x00007ffc9b0e9000) libmkl_intel_lp64.so.2 => /opt/intel/oneapi/mkl/2023.2.0/lib/intel64/libmkl_intel_lp64.so.2 (0x00007f1516c00000) libmkl_intel_thread.so.2 => /opt/intel/oneapi/mkl/2023.2.0/lib/intel64/libmkl_intel_thread.so.2...

Markus posted a comment on discussion Elk Users

Markus — Mon, 19 Jan 2026 22:56:55 -0000

... if I remember correctly, you have to use libxc 7.0.0 for elk 10.8.1. Please try this. Remember to compile libxc according to your preferred linking type (static or dynamic).

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Mon, 19 Jan 2026 21:46:00 -0000

I re paste my make.inc file: MAKE = make AR = ar Elk requires BLAS/LAPACK and fast Fourier transform (FFT) libraries. We recommend using optimised, mutithreaded BLAS/LAPACK libaries such as those provided by the Intel Math Kernel Library (MKL), OpenBLAS or BLIS. Elk can use either the Fastest Fourier Transform in the West (FFTW) or the MKL FFT. Both single and double precision versions of the FFT are needed. ------------------------------------------------------------------------------ Libraries...

Hong Tang posted a comment on discussion Elk Users

Hong Tang — Mon, 19 Jan 2026 21:27:51 -0000

Dear ELK team, I am compiling ELK 10.8.1 with libxc.6.2.1 and wanner90v2. I use Intel compilers. It seems that my libxc was installed correctly. but ELK install report errors, see below. I attached my install script for ELK and Libxc. Please check them. Note, without libxc, I can compile ELK seemly correctly and can run the test examples. The errors occur when compile ELK with libxc. Please help me to find the problem. I appreciate your help very much! Best. Hong my make.inc for ELK: MAKE = make...

J. K. Dewhurst posted a comment on discussion Elk Users

J. K. Dewhurst — Wed, 07 Jan 2026 16:39:39 -0000

Hi Jerry, Happy new year! I tried your input file and the sum is correct: you have to add all the components in the files. You can also enable dosmsum .true. to sum over the m degrees of freedom and just leave l components. Regards, Kay.

Jerry Li modified a comment on discussion Elk Users

Jerry Li — Wed, 07 Jan 2026 15:54:53 -0000

Happy New Year, Kay! It seems to be a bug with DOS. I just tried the latest version 10.8.1 with NiTe2. SCF is converged! I added all PDOS components (including IDOS+Ni+Te1+Te2) but the sum is not equal to TDOS. See the attached input file and plot, please check. Many thanks Jerry

Jerry Li posted a comment on discussion Elk Users

Jerry Li — Wed, 07 Jan 2026 15:31:50 -0000

J. K. Dewhurst created a blog post

J. K. Dewhurst — Wed, 07 Jan 2026 10:29:42 -0000

Elk version 10.8.1 released

Elk released /elk-10.8.1.tgz

Elk — Wed, 07 Jan 2026 10:28:03 -0000