[Q-e-developers] Re: Next Quantum ESPRESSO release -- June 20th

[nisihara225 at gmail.com]

Dear Dr. Paolo Giannozzi,




I am sorry for the delay in my reply.





> My suggestion for the next release is to align 3D-RISM-SCF to the QE version that will be released (this should be easy); 

> to release 3D-RISM-SCF as a modified QE (not yet sure how in practice, if as a patch, or as a bunch of modified routines,

> or as new PW/src and Modules/ directories); then, to start seriously to think at how to merge it into QE. 

> Unfortunately, merging takes a lot of time and nobody is really in charge of it (in the future this may change, hopefully).




As you point out, the 3D-RISM-SCF works inside the self-consistent loop.

And it may require careful consideration and enough time to merge the code.


I accept your suggestion to avoid the integration of the 3D-RISM-SCF for the next release of Q.E.

I hope it will be integrated in the future.






> About "custom" FFTs: these are used only in hybrid calculations, to speed up the computation of (V_x\psi)(r) using FFTs

> with a grid corresponding to a reduced cutoff (this is 1-2 times ecutwfc, the cutoff for plane waves used 

> in the expansion of \psi(r)=\sum_G\psi(G) e^{iGr}, instead of 4*ecutwfc, needed in principle to perform the exact computation).

> I am surprised to hear that this is used in 3D-RISM-SCF (or are you planning to use them in the future)?


I know well the calculation of exact-exchange with reduced FFT’s mesh,

because I have implemented Hybrid-GGA to another first-principles software.

In the 3D-RISM-SCF, the purpose to use “Custom” FFTs is to speed up the computation  (same as exact-exchange).

Solvent’s distributions have to be calculated by “Dense” FFTs, strictly.

However, “Custom” FFTs with small cutoff yield little decreasing of precision and a lot of speed-up.



Thus, I have used “Custom” FFTs, and I will use them in the future.


If you change or repeal the implementation of “Custom” FFTs,

I would like for you to remain their versatility or 

to prepare another generic modules of FFT with arbitrary cutoff.




Regards,

Satomichi Nishihara








差出人: Paolo Giannozzi
送信日時: ‎2015‎年‎6‎月‎9‎日 ‎火曜日 ‎0‎:‎56
宛先: nisihara225 at gmail.com
CC: General discussion list for Quantum ESPRESSO developers, Filippo S







This is a rather complex issue and I am not sure it can be solved in the next few days. If I understand correctly, 
3D-RISM-SCF modifies the way the self-consistent code works internally, by adding computation of additional 
quantities inside the self-consistent loop. Until now this kind of changes has been added inside QE directly by 
the author, but after many uncoordinated additions the result has become a mess. One idea that has been around
for a while is to locate some "hooks" in the code where additional computations can be confined; then release the
implementation of additional methods (like 3D-RISM-SCF) as "add-ons" that modify QE in those specific points only


(notice the plugins_* subroutines). Unfortunately this is still ongoing (or maybe "notgoing") work. 

My suggestion for the next release is to align 3D-RISM-SCF to the QE version that will be released (this should be easy); 
to release 3D-RISM-SCF as a modified QE (not yet sure how in practice, if as a patch, or as a bunch of modified routines,


or as new PW/src and Modules/ directories); then, to start seriously to think at how to merge it into QE. 
Unfortunately, merging takes a lot of time and nobody is really in charge of it (in the future this may change, hopefully).

About "custom" FFTs: these are used only in hybrid calculations, to speed up the computation of (V_x\psi)(r) using FFTs

with a grid corresponding to a reduced cutoff (this is 1-2 times ecutwfc, the cutoff for plane waves used in the expansion of \psi(r)=\sum_G\psi(G) e^{iGr}, instead of 4*ecutwfc, needed in principle to perform the exact computation). I am surprised

to hear that this is used in 3D-RISM-SCF (or are you planning to use them in the future)?



Paolo




On Mon, Jun 8, 2015 at 3:38 PM, Filippo Spiga <spiga.filippo at gmail.com> wrote:

Dear Satomichi,

thank you very much for your quick feedback. I will attempt myself to merge your branch into a local copy of current QE codebase and run the tests. If I get something weird, I will contact you directly. I suggest to sync by email again beginning next week, I will get in touch privately. From my perspective, once the code compiles and is aligned and test are provided then it can be released. I let senior QE developers and PIs comment further.

Thank you to contributing to Quantum ESPRESSO, your effort is really appreciated!

Regards,
Filippo



On Jun 8, 2015, at 1:02 PM, nisihara225 at gmail.com wrote:
> Dear Dr. Filippo Spiga,
>
> the answers to your questions are as follows.
>
> > Which QE version is compatible with it?
> The branch of 3D-RISM-SCF is compatible with the Q.E. 5.1.2.
> In more detail, it has branched from the 11516th revision of SVN.
> The URL of the branch is http://qeforge.qe-forge.org/svn/q-e/branches/espresso-3drismscf .
>
> > Did you develop 3D-RISM-SCF as external plug-in or directly inside the QE code?
> The 3D-RISM-SCF was implemented directly inside the Q.E. code.
> Source files in Modules/ and PW/ were mainly modified.
>
> > Do you have extensive tests to validate your implementation?
> Yes, I prepare some tests or examples:
> /branches/espresso-3drismscf/espresso/PW/examples/RISM_example .
> And I confirm that these examples provide reasonable results.
>
> > Do you have one or more reference papers that explains the method utilized?
> Yes, main reference papers are
> H.Sato et al., J. Chem. Phys. 2000, 112, 9463 , and
> A.Kovalenko, F.Hirata, Chem. Phys. Lett. 1998, 290, 237 .
> Some formulas are utilized in the 3D-RISM-SCF, and
> their references are written in comments of corresponding source codes or manuals.
>
> > Is your code suitable for general public release under GPL license?
> Yes, my code is based on GPL license (http://www.gnu.org/copyleft/gpl.txt).
> I have wrriten so, in header comments of the source codes.
>
> > Are you willingness to maintain and take care of bug fixes and improvements of this particular piece of code for at least one year starting from the release date?
> Of course, I will maintain the code.
>
> Regards,
> Satomichi Nishihara




--
Mr. Filippo SPIGA, M.Sc.
Quantum ESPRESSO Foundation
http://fspiga.github.io ~ skype: filippo.spiga

«Nobody will drive us out of Cantor's paradise.» ~ David Hilbert

*****
Disclaimer: "Please note this message and any attachments are CONFIDENTIAL and may be privileged or otherwise protected from disclosure. The contents are not to be disclosed to anyone other than the addressee. Unauthorized recipients are requested to preserve this confidentiality and to advise the sender immediately of any error in transmission."





-- 


Paolo Giannozzi, Dept. Chemistry&Physics&Environment,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
hone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20150612/011ece51/attachment.html>