rs,
Fabrizio
From: Mauro Francesco Sgroi
Sent: Friday, August 2, 2024 2:35 PM
To: Fabrizio Ferrari Ruffino
Cc: Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Help for compilation with Nvidia HPC SDK
Dear Fabrizio,
thanks a lot for the explanation.
I was unsure about how t
Hi,
there are a few minor FFTXlib calls somewhere in QE which are still on CPU,
therefore it is better to have a CPU fft backend enabled too. Whether to use
the internal one or FFTW3 should not make much difference, since all the main
stuff runs on gpu (therefore calling cuFFT).
In a CPU run t
Hello,
'VDW1' is a Van-der-Waals contribution only, in order to get a consistent dft
you should also add the LDA/GGA terms, for example 'sla+pw+revx+vdw1'.
Regarding the energy graph, I would try a smaller convergence threshold
(conv_thr) first.
Cheers,
Fabrizio
CNR-IOM
_
in terms of results, because the pw runs
are correct anyway, and the dos run does not compute xc functionals.
Best,
Fabrizio
From: Jing Lian Ng
Sent: Tuesday, September 19, 2023 9:16 PM
To: Fabrizio Ferrari Ruffino ;
users@lists.quantum-espresso.org
Subject:
2023 11:52 PM
To: Fabrizio Ferrari Ruffino ;
users@lists.quantum-espresso.org
Subject: Re: [QE-users] SCAN Functional not recognized in DOS.x
Hello Dr Ruffino,
Thank you so much for the advice! After applying changes to the funct.f90, do I
need to compile everything with 'make all' aga
Hello,
this is a combination of dfts rarely used with QE and whose choice has been
enabled fairly recently. However it should be possible to use it in PP runs too.
For some reasons in PP runs such as dos.x, a functional with a VV10 term, when
detected, is converted in its standard form (the one y
Hello,
the file CMakeLists.txt is related to the CMake builder, which you are not
using (you are using autotools), so it does not help. Try to change the
configure option like this:
--with-libxc-prefix=/usr/local/libxc
so no /lib/
Also, if you need to use a libxc gga_x functional be aware of thi
Hello,
there is a bug in QE version 7.2 for exchange_only functionals in libxc, that
has been fixed in the develop version.
If you do not want to download the develop branch, tou just have to add the two
following lines:
* xcoef=1.d0
* if (ishybrid .and. exx_started .and. exx_fraction>0.
Hello,
are you using dftd3? If so, which dftd3_version (input file)?
Only some selected functionals are allowed with dftd3, scan is usable with
version 4 only.
Cheers,
Fabrizio
From: users on behalf of Monteiro
Campos de Melo, P.M. (Pedro) via users
Sent: Monda
Hi Michele,
which version of Libxc did you use?
I suspect you are using the develop one, for which compatibility with QE cannot
be ensured, for obvious reasons.
I recommend you tu use the latest release, e.g. 6.1.0
Best,
Fabrizio
CNR-IOM
From: users on behalf of
Hello,
I am not sure what you mean with self-consistent calculated parameters.
The ones you see in xc-infos.x are the default values of the libxc-functional
external parameters, as you said. Maybe you mean that some of these parameters
can be 'iteration dependent'? Can you point the equations in
Hello,
are you sure Libxc was correctly linked when compiling QE?
In case, details on how to link it are present in the user_guide (Doc folder).
Cheers,
Fabrizio
From: users on behalf of Jibiao Li
Sent: Wednesday, December 21, 2022 6:40 AM
To: users
Subject: [Q
Hi Ricardo,
I don't know well the rVV10 functional, however, as it is implemented now in
QE, the b_parameter is set to 15.7 whenever a MGGA term is present. However, if
you want to change the b_parameter, just go to line 933 of Modules/funct.f90
and set the value you want. Then of course recompi
So you start from the scratch of the index notation:
XC-000i-000i-000i-000i-000i-000i
The first slot is for LDA-exchange, the second LDA correlation, the third GGA
exchange, then GGA correlation, MGGA exchange and MGGA correlation.
Now, you want to use the MGGA_C_M06 (with no exchange?) which has
Dear Jibiao,
I stress the following lines from the user guide:
The complete list of libxc functionals (and their IDs) is available at:
https://www.tddft.org/programs/libxc/functionals/
So you can find there the IDs to put in the index notation.
Cheers,
Fabrizio
orno 6 dic 2022, alle ore 09:58, Fabrizio Ferrari Ruffino
ha scritto:
Hello,
I run sometimes QE on my laptop with the gtx 1050 (mainly for debugging). You
just need to download the hpc_sdk toolkit. Of course you are limited to small
calculations and I often find troubles when I try to run 2+ MPI o
Hello,
I run sometimes QE on my laptop with the gtx 1050 (mainly for debugging). You
just need to download the hpc_sdk toolkit. Of course you are limited to small
calculations and I often find troubles when I try to run 2+ MPI on the same
GPU. But you can certainly give it a try.
Cheers,
Fabriz
Hello,
I guess the error is coming from the fact that you have two different xc
functionals coming from the two pseudos (PBE and PBESOL), so the program
doesn't know which one to choose.
Cheers,
Fabrizio
From: users on behalf of Levin
Rojas, Natalia
Sent: Tues
Hi Riccardo
LD_LIBRARY_PATH is an environmental variable which contains the paths where to
look for during execution when dynamical libraries are needed.
So after compilation (which you did smoothly I guess), just locate the file
libmkl_scalapack_lp64.so. Type from shell:
locate libmkl_scalapa
Hello,
exporting to LD_LIBRARY_PATH the path to that file should work:
export LD_LIBRARY_PATH=path_to_libmkl_scalapack.
Cheers,
Fabrizio
CNR IOM
From: users on behalf of Riccardo
Piombo uniroma1 via users
Sent: Wednesday, November 16, 2022 2:59 PM
To: user
where it is multiplied by exx_fraction?
All the best,
Dr. Roman Voronkov
P.N. Lebedev Physical Institute of RAS
P.S. Forgot to specify letter subject, sorry.
26.07.2022, 14:36, "Fabrizio Ferrari Ruffino" :
Dear Roman,
I suggest you to switch to a more recent version of QE (7.1 is
Dear Roman,
I suggest you to switch to a more recent version of QE (7.1 is the latest)
since there have been major changes in the XC part of QE starting from v6.8.
Now all the XC related routines (except vdW) are in a separate library, XClib.
The hybrid functionals, as all the others, are compute
Dear Banhi,
this is the file you were looking for. However, I strongly suggest you to get a
newer (ideally the latest) QE release, since there have been a lot of changes
in the dft part since v6.7, including the management of the libxc external
parameters which now should be simpler (you can hav
A small correction, in the two input_dft options I mentioned before it is
better to put 'vv10', not 'rvv10', since the correct name for that single
nonlocal term in qe is the former, even if it should work anyway.
Cheers,
Fabrizio
On Tue, Dec 21, 2021 at 12:10
Dear Christoph,
I am having some hard times in reproducing your errors. A few things:
- I get the segmentation fault with Libxc-6.0.0 (libxc-master), never with
v5.1.7. By the way, since v6.0.0 is actually the develop branch, it changes
from day to day (or week to week), so, if the library you are
Hello,
the first issue can be circumvented as you said, however in the new
imminent release there will be only one allowed notation for Libxc
functionals based on the indices and that should eliminate at the root all
these kinds of issues (explanations in the user_guide).
Regarding the second issue
Hello,
I suggest you to have a look at section 2.6.1 of the user_guide (Doc
folder) and try the options suggested. If you still have troubles let me
know.
Cheers,
Fabrizio
On Wed, Dec 8, 2021 at 3:57 PM Michal Husak wrote:
> Hi
>
> I had successfully compiled libxc and than recompiled QE 6.8 w
ive me the advice how I should compile libxc.5.1.5?
>
>
>
> Best,
>
> Hong
>
>
>
>
>
>
>
> Sent from Mail <https://go.microsoft.com/fwlink/?LinkId=550986> for
> Windows 10
>
>
>
> *From: *tanghong...@outlook.com
> *Sent: *Monday, Ju
Hello,
libxc-5.1.5 version is very recent, while QE 6.4.1 and 6.6 versions are
not. Backward compatibility of QE with libxc is ensured (at least since
v4.3.4), but the opposite obviously is not (but 6.6 should work, maybe with
some limitations).
In QE 6.6 and later you should use -lxcf03, not -lxcf
Dear Jibiao,
I (hopefully) fixed the problem in the develop branch of QE. You can try
it. It should be pretty stable since the next release is imminent.
Cheers,
Fabrizio
On Tue, Jul 13, 2021 at 11:47 AM Jibiao Li wrote:
> Dear Dr. Lorenzo Paulatto,
>
> I followed your instruction, and add the li
Hello,
up to now not all the possible combinations of single dft terms are allowed
for nonscf calculations, but only the ones identified by the 'short names'.
They are listed in the comment block in Modules/funct.f90. This might
change in the near future.
Cheers,
Fabrizio
On Wed, Jun 30, 2021 at 2
Hello,
in QE6.7 go to Modules/funct.f90 and at line 911 replace 600 with 900.
There has been a pretty big increase in the number of functionals in the
recent versions of libxc so the maximum id has to be increased.
In general most of the choices in libxc are usable in QE, however many
functionals d
Hello,
as far as I know, the GPU version should work with metaGGA functionals as
the non-GPU one.
The specific routines for metaGGA functionals are not accelerated (they
will probably be in the next release), but they should not make a big
difference in terms of performance.
Same thing for libxc, i
Hello,
compatibility with the recent libxc version 5.1.3 still hasn't been fully
checked in QE, however, I managed to run a few tests with QE6.7 with no
problems, both with libxc 5.1.2 and 5.1.3 (in QE6.7 there should be no need
to change xc_f90 to xc_f03 anywhere). Has your QE code been modified w
Hello,
by having a look at the dft-d3 code (subroutine 'setfuncpar' in 'core.f90')
I see that the functional you enforced, which corresponds to the 'optbk88'
dft shortname is not available. I don't know if it can be added
straightforwardly or not. I will have a look, but I hope that someone that
kn
Hello,
it seems it does not find the math libraries. You need to locate the path
with libcublas.so and libcufft.so too.
Cheers,
Fabrizio
On Thu, Mar 11, 2021 at 9:03 PM Zeeshan Ahmad wrote:
> Hi,
>
> I am trying to install quantum espresso 6.7-gpu for V100 gpu using the
> configure command:
>
>
Hello,
the message simply means that the chosen functional depends on a number of
parameters that can be 'manually' set by the user, otherwise their default
value is kept. As far as I could see the external dependence on parameters
in SCAN has been introduced in the last releases of libxc (v5.0.0 a
Correct, too much in a hurry. But I still recommend to use v4.3.4 or older.
Cheers
Fabrizio
Il ven 29 gen 2021, 18:38 Bidault, Xavier ha scritto:
> Hello,
>
> Actually, the message error is quite self-explanatory: This module file
> was not generated by any release of this compiler.
> If you co
Hello,
actually the compatibility of QE has been tested starting from libxc 4.3.4.
Compatibility with older versions is not guaranteed.
In the user_guide of QE (doc folder) you will find more informations on the
use of Libxc in QE.
Cheers,
Fabrizio
On Fri, Jan 29, 2021 at 5:32 PM Hien Vo wrote:
> found sometime ago:
>
> - run configure without any libxc flag
> - go to the make.inc file and, by following the comments inside, add the
> needed flags to DFLAGS, IFLAGS and LD_LIBS.
>
> Do you think that it is better not to use libxc at all?
>
> José C. Conesa
> E
Hello,
are you compiling with libxc? Which version?
Fabrizio
On Tue, Jan 26, 2021 at 6:39 PM José Carlos Conesa Cegarra <
jccon...@icp.csic.es> wrote:
> Dear all,
>
> I tried compiling qe-6.7. The compilation fails with a number of
> messages like these:
>
> funct.f90(40): error #7002: Error in
brizio
On Wed, Dec 30, 2020 at 2:46 PM mkondrin wrote:
> Hi, Fabrizio,
>
> Thank you for your reply, I have follow your suggestion but with no
> avail. The compilation stopped with a message :
> cannot find -lxcf03
>
> Sincerely yours,
> Mikhail Kondrin
>
> On
Hello,
did you check the notes in section 2.5.1 of the user_guide (Doc folder)?
WIth version 4.3.4 is better to use the flag -lxcf03 for the headers in
make.inc.
Cheers,
Fabrizio
On Wed, Dec 30, 2020 at 1:01 PM mkondrin wrote:
> When I have tried another way (editing make.inc file after simple
>
Hello,
f03 works for libxc 4.3.4 and should work for the upcoming releases, but
not for v5.0.0.
Please see section 2.5.1 of the 'user_guide' in Doc folder on how to
compile QE with v5.0.0 (especially the note for v5.0.0).
Cheers,
Fabrizio
On Tue, Nov 10, 2020 at 10:03 AM Christoph Wolf
wrote:
>
Hello,
you can enforce the b3lyp functional from input just by adding
'input_dft=b3lyp' in &system in the input file. There are also some
variants of b3lyp in libxc that can be found here:
https://www.tddft.org/programs/libxc/functionals/
You can use them too (see the user_guide in Doc folder of QE
Hi Michal,
at present libbeef is used by QE only for GGA. The libxc functional
'mgga_x_mbeefvdw' formally appears as an ordinary mGGA functional, with no
dependence on extra parameters, so I would say that, at least formally, it
is possible to use it as an input_dft. That said, I don't know much ab
( 0 0 3 4 0 0 0)
> Any further DFT definition will be discarded
> Please, verify this is what you really want
>
> Subspace diagonalization in iterative solution of the eigenvalue
> problem:
> one sub-group per band group will be used
&g
Hello,
I tried a couple of runs with your input dft and libxc4.3.4 and I did not
face the error. Which libxc version are you using?
The fact that it works by substituting am05 with pbe could just mean that
the program just scans the 'pbe' word inside the string and uses the
ordinary internal (in QE
Dear Yuvam,
I am not an expert of vdw interactions. If you don't know it already, there
are some examples of vdw calculations with QE in folder
'test-suite/pw_vdw'. Not sure if you'll find what you are looking for, but
perhaps you could start by having a look there.
Best regards,
Fabrizio
On Fri,
Hi Jibiao,
please do this:
1-configure without any libxc flag;
2-go to the make.inc file in the main folder and add the various flags to
DFLAGS, IFLAGS and LD_LIBS by following the comments inside;
3-compile as usual.
This should answer to both your emails
Best regards,
Fabrizio
On Fri, Jul 3, 2
input ?
>
> Regards
>
> Jibiao Li
>
> ------ Original --
> *From:* "Fabrizio Ferrari";
> *Date:* Mon, Jun 29, 2020 07:25 PM
> *To:* "Quantum ESPRESSO users Forum";
> *Subject:* Re: [QE-users] Unrecognized dft when specifying BEEF-vdW
> functional
'BEEF_VDW', and 'BEEF-VDW', but all these cases gave me the same
> error: unrecognized dft. I don't know what should I do to resolve this
> problem. Further help is need.
>
> Best
>
>
>
> -- Original --
> *From:* "
.3.4?
> Best
> Jibiao Li
>
>
> -- Original ------
> *From:* "Fabrizio Ferrari";
> *Date:* Mon, Jun 29, 2020 05:37 PM
> *To:* "Quantum ESPRESSO users Forum";
> *Subject:* Re: [QE-users] Unrecognized dft when specifying BEEF-vdW
> functional
&g
Hello,
I don't know much about BEEF functionals, but the corresponding library in
QE (LIBBEEF, which is independent from libxc) was added quite recently.
Perhaps you should try with the develop version of QE.
Hope it helps, cheers
Fabrizio
On Mon, Jun 29, 2020 at 10:33 AM Jibiao Li wrote:
> Dea
Hello,
the compatibility check of QE with the latest version v5.0.0 of libxc is
still in progress. It will be set in the near future. For now I suggest you
to keep using v4.3.4.
I usually link libxc in QE by adding the flags in DFLAGS, IFLAGS and
LD_LIBS in the make.inc file (you can follow the com
dn't declare input_dft option. And not declaring would
> > result in error (as I reported).
> > Thanks,
> > Sergey Lisenkov
> > 26.05.2020, 10:36, "Fabrizio Ferrari" :
> >
> > Hello Sergey,
> > that is strange. I have tried some simple tests too
#x27;scan') without any problem - I searched archive and saw several examples.
> Also, I used SCAN pseudopotentials from QE website, so in principle I
> shouldn't declare input_dft option. And not declaring would result in error
> (as I reported).
>
> Thanks,
> Sergey Lisenk
y (or something
> like that).
>
>
> 25.05.2020, 17:21, "Fabrizio Ferrari" :
>
> Hello,
> in your case (scan) that error should come out only if libxc has not been
> properly linked during compilation. Have you tried runs with other libxc
> functionals (not necessarily mG
Hello,
in your case (scan) that error should come out only if libxc has not been
properly linked during compilation. Have you tried runs with other libxc
functionals (not necessarily mGGA)? Do they work?
Just in case you want to check, to properly compile with libxc, after the
qe configuration, you
Hello, this bug should have been fixed a few months ago:
https://lists.quantum-espresso.org/pipermail/developers/2020-January/002159.html
Please try your job with the develop version of qe available on gitlab.
Best regards
On Tue, Apr 28, 2020 at 1:07 PM wrote:
> Dear experts,
>
> I have encount
Hi,
I noticed a mismatch between the length of the dft string and the variable
assigned to it in one of the involved routines. This causes error in some
cases. It seems an easy fix and I will commit it on the develop branch
soon. I'll inform you when it's ready.
Cheers
On Mon, Apr 27, 2020 at 3:03
Hello,
in general yes, but how depends a bit on the version you are using. In 6.5
it is sufficient to enforce it from input by putting
input_dft='GGA_XC_HLE16' in &system in the input file.
However I recommend you to use the develop version of QE, since there is a
quite recent factor 2 correction
Hello,
I am not sure if libxc is fully usable with qe v6.1. If you want to use
libxc SCAN with v6.1 you should add the scan routines and their calls in
metagga.f90 and funct.f90 respectively (they are in Modules folder), as it
was done in v6.2. But I do not guarantee it will work straightforwardly
Dear Arini,
apparently the lxcf03 flag is missing (it is needed in v6.5). You should
add it to LD_LIBS in the make.inc file. If it still does not compile, I
suggest you to link 'maually' libxc only after the configuration:
- run configure without any libxc flag
- go to the make.inc file and, by fol
Hello,
at present the scan0 functional is usable only by CP code. For some reasons
the call to 'errore()' which is present in 6.3 and 6.4 has been removed in
6.5, but still the EXX calculation is not performed by PW.
Some changes will be introduced in the near future to facilitate the use of
libxc
Dear Valentina,
in order to use SCAN functionals you need to link libxc during the
compilation, since they are not present in the qe internal library of xc
functionals.
To do this you have to add the libxc flags in the make.inc file in the main
folder (it is generated after the configuration). Add
Dear Muhammad,
the first of the two issues should have been fixed in one of the recent
commits in the develop branch of QE on GitLab repository, together with a
factor 2 correction on the libxc-gga functionals.
Please try your DOS calculation by downloading the latest develop version.
Best regards,
Hello, that message comes out only if libxc has not been linked during
compilation. Maybe you should try a make clean and recompile everything by
checking that libxc stuff is properly set in the make.inc file from the
beginning.
Cheers,
Fabrizio
On Thu, Dec 5, 2019 at 11:09 PM Abhirup Patra
wrote
me, so my guess is that maybe I misunderstood how to
> change it.
>
> Sorry for the trouble, and thank you for your patience
> Fábio Costa
> --
> *De:* users em nome de
> Fabrizio Ferrari
> *Enviado:* sexta-feira, 8 de novembro de 2019 11:51
Dear Fabio,
I added some lines in the attached file 'Modules/xc_mgga_drivers.f90' so
that, by replacing it with the original one, you should be able to set the
c parameter in tb09. You can do it directly into that module by setting the
initial value of cc_param or in any part of the program by call
Dear Fabio,
there has been quite recently a 'more consistent' incorporation of libxc
functionals into the QE code (look at 'develop' branch in GitLab).
In order to use them, as you already did, you have to put their names into
the input_dft string, but it is necessary to write their complete name,
Dear José,
the current version of qe (develop branch in GitLab) allows you to use
libxc functionals for LDA, GGA and mGGA. No kinetic functionals at the
moment.
To use libxc functionals you must enforce them from input (use 'input_dft'
in &system) and write their names in the input string. The orde
locbuff(ir,ibnd,ikq)=0.0_DP
> ENDDO
> ENDDO
> ENDDO
> END IF
> ELSE
>
> ?
>
> On Tue, 9 Jul 2019 at 11:37, Fabrizio Ferrari
> wrote:
>
>> Hello,
>> if you put the loop at line 467 of PW/src/exx.f90 inside
Hello,
if you put the loop at line 467 of PW/src/exx.f90 inside an
'IF(gamma_only)', the segfault should disappear. I'm just checking if that
is the only fix needed.
Fabrizio
On Tue, Jul 9, 2019 at 12:27 PM Michal Krompiec
wrote:
> I got a similar segfault using a fresh installation of QE 6.4.1
74 matches
Mail list logo