[QE-developers] Interfacing MBD library to QEspresso

[Hsin-Yu Ko]

Dear Paolo and Daniele,

Thank you so much for your prompt response! Regarding Paolo's question,
libmbd is currently built using CMake; however, Szabolcs and I should be
able to write an alternative makefile without too much issue (which seems
to be the right way to proceed at least for now).

Daniele, thank you for kindly offering help. We really appreciate it! We
will give it a try and let you know if you should encounter any issues.

Best Regards,
Hsin-Yu


On Tue, Nov 17, 2020 at 8:59 AM Daniele Cesarini <d.cesarini at cineca.it>
wrote:

> Dear Hsin-Yu,
>
> yes, we are working to move on CMake build system, this is a
> work-in-progress activity but we are in a good shape.
>
> To add an external library to QE using CMake is quite straightforward (I
> am sorry, it currently lacks of documentation), just follow the example of
> FoX library. You can add a git submodule to the project and call it from
> the CMakeLists.txt located in the "external" directory of the project. You
> can copy the CMake code of 'external/CMakeLists.txt' between line 4 and 27
> and adapt it to MBD library. You just need to define the appropriate CMake
> targets and create a new CMake target library, which you can link to QE
> libraries/executables as well as other targets.
>
> Let me know you if you have questions/problems, I would happy to help you!
>
> BTW, if you need help just point me to your QE fork, I can help/assist you.
>
> Best,
> Daniele
> On 17/11/20 14:10, Hsin-Yu Ko wrote:
>
> Dear All,
>
> I just wanted to add to this thread some more details: We forked the
> official QEF/q-e repo and followed closely the currently implemented
> `ts_vdw` (Tkatchenko-Scheffler van der Waals) procedure.
>
> The issue that we could really use your help is in the `configure` and the
> new `cmake` building mechanism. What we currently have is the below (following
> FoX library mechanism):
> 1. We (manually) download the most recent release of the `libmbd` to
> `archive/` as a tarball
> 2. We add `libmbd` build procedures (added as a dependency of `mods`)  as
> `libfox` in `Makefile` and `install/extlibs_makefile`
> 3. We append the include/library paths to `IFLAGS`/`LD_LIBS` in `install/
> make.inc.in
> <https://urldefense.com/v3/__http://make.inc.in__;!!P1tgJ-3e!Wb0faV2y6SftYRg3KV3mgtAdi-0a_69XJor6LylnF5qXkcaf3rYa-jaYudERPn6btA$>
> `.
>
> By doing so, it seems to build no problem from our local machines.
> However, there is an issue when we use it in the GitLab pipeline. The issue
> is that `libmbd` depends on `cmake` which is not present in the docker
> image. As such, we had to include the cmake installation in the
> `.gitlab-cl.yml` file. We also noticed that there is a new
> `build:cmake-gnu` pipeline. Since there seems to be some tendency of moving
> to `cmake`, we were wondering if we could ask for some advice on the how to
> proceed. Thank you!
>
> Best Regards,
> Hsin-Yu
>
> --
> Hsin-Yu Ko
> Postdoctoral Research Fellow
> Department of Chemistry and Chemical Biology
> Cornell University
>
> On Tue, Nov 17, 2020 at 7:31 AM Szabolcs GOGER <szabolcs.goger at uni.lu>
> wrote:
>
>> Dear Developers,
>>
>>
>> We are currently working on writing a module for Many-Body Dispersion
>> correction (MBD) for Quantum Espresso here in Luxembourg in Alexandre
>> Tkatchenko's group. We are doing this in collaboration with Cornell
>> University, in particular, Hsin-Yu Ko from the group of Robert DiStasio is
>> closely invested in the project. With his technical assistance we are
>> getting to a state when our personal implementation might be ready for
>> becoming public.
>>
>>
>> The main idea behind the implementation is to use a portable library
>> (called libMBD) which can be readily called from a Fortran code. The
>> library has already been linked to other DFT codes so using it ensures
>> reproducibility and transferability. Currently our code can be used in the
>> SCF energy calculations, while using the MBD forces for structure
>> optimization is still an ongoing project. Since most of the development is
>> done in the portable library itself, this new feature would only need a
>> minimal amount of modification in the main code. We are also working on
>> preparing a well-rounded test suite.
>>
>>
>> We would be grateful if someone, who has experience in building projects
>> using an external library, could assist us in finding an efficient way of
>> preparing our code to be compatible with the main release.
>>
>>
>> Kind Regards,
>> Szabolcs Goger
>> Szabolcs Goger
>> Ph.D. student
>> Department of Physics and Materials Science
>>
>> UNIVERSITÉ DU LUXEMBOURG
>>
>> CAMPUS LIMPERTSBERG
>> 162a, avenue de la Faïencerie
>> L-1511 Luxembourg
>> T +352 46 66 44 6694
>> szabolcs.goger at uni.lu /www.uni.lu
>> <https://urldefense.com/v3/__http://www.uni.lu/__;!!P1tgJ-3e!Wb0faV2y6SftYRg3KV3mgtAdi-0a_69XJor6LylnF5qXkcaf3rYa-jaYudHuJ6vdLg$>
>> / www.tcpunilu.com
>> <https://urldefense.com/v3/__http://www.tcpunilu.com/__;!!P1tgJ-3e!Wb0faV2y6SftYRg3KV3mgtAdi-0a_69XJor6LylnF5qXkcaf3rYa-jaYudG0jk8qWA$>
>>
>>
> _______________________________________________
> developers mailing listdevelopers at lists.quantum-espresso.orghttps://urldefense.com/v3/__https://lists.quantum-espresso.org/mailman/listinfo/developers__;!!P1tgJ-3e!Wb0faV2y6SftYRg3KV3mgtAdi-0a_69XJor6LylnF5qXkcaf3rYa-jaYudHo49vfeA$
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20201117/38ca89a7/attachment-0001.html>