[QE-users] differently complied QE

[Aleksandra Oranskaia]

Hello Nicola,

[hope I am correctly using the forum and my messages reach you]

I tried GNU and Intel compilations, both on two very different workstations, all for 6.2.1 and 6.3 versions, for the same inputs with the tightened parameters.
In all the cases lattice constant is the same up to the 4th decimal (i.e, they all were like 3.458). However, on our University Supercomputer (Cray XC40, admin installation) I am getting the result that differs (being like 3.453).
3.458 vs 3.453 is like 3.46 vs 3.45 that probably is not cool at all.

Could you please suggest any “benchmark” vc-relax input to check if cray compilation on the supercomputer is correct?



Thanks in advance,
Alex.
___
Aleksandra Oranskaia (M.Sc.)
ChemS PhD student, KAUST
Phone: +966 50 1335254






> On Jan 8, 2019, at 2:49 AM, Aleksandra Oranskaia <aleksandra.oranskaia at kaust.edu.sa> wrote:
> 
> Thank you for the respond!
> 
> I was confused with the difference, because if the 3rd digit is “arbitrary” (depends on libraries, compilers, hardware, etc.) than there is no sense in reporting lattice constants in papers with this accuracy; however, it is a common practice.
> 
> I used:
> conv_thr=1.d-8
> tot_conv_thr=1.d-6
> forc_conv_thr=1.d-6
> cell_dofree=‘volume’ (for the cubic cell)
> 
> I will try to increase conv_thr and press_conv_thr, as well as to compare gnu and intel compillations, and reply back.
> 
> 
> Best,
> Alex.
> 
> ___
> Aleksandra Oranskaia (M.Sc.)
> ChemS PhD student, KAUST
> Phone: +966 50 1335254
> 
> 
>> On Jan 8, 2019, at 2:39 AM, Nicola Marzari <nicola.marzari at epfl.ch <mailto:nicola.marzari at epfl.ch>> wrote:
>> 
>> 
>> Last - note that 3.458 and 3.453 are really close - so the accuracy you have now is good enough for research purposes.
>> 
>> 			nicola
>> 
>> 
>> 
>> On 08/01/2019 00:38, Nicola Marzari wrote:
>>> Well, the numbers you get are not infinitely precise - there is
>>> a threshold where you stop minimizing forces and cell geometry,
>>> using forces and stresses that are not perfect because each SCF cycle is
>>> not converged to infinite perfection.
>>> Bottom line - try to tighten all the convergencies (scf convergence
>>> at every cycle, and forces at every ionic relxations, and pressure...)
>>> conv_thr for electrons
>>> etot_conv_thr for ions
>>> forc_conv_thr for ions
>>> press_conv_thr for cell
>>> are the variables to play with. I would make sure conv_thr is very tight
>>> (10^-10 per atom, or even less) and then play with forc_ and press_
>>>         nicola
>>> On 07/01/2019 14:24, Aleksandra Oranskaia wrote:
>>>> Hello dear users and developers of QE,
>>>> 
>>>> I have recently noticed that one and the same QE version that was installed with different libraries, compilers, on different machines (with very different hardware) not only gives different total energies for the same inputs but also quite different optimized lattice constants (say, 3.458 versus 3.453 that does not seem no be acceptable).
>>>> 
>>>> So the question is: how to know which installation accumulates some small(?) numerical problems?
>>>> In standard QE pw examples there is no example with variable-cell relaxation done on “ideally” installed code, so there is no reference point to reproduce.
>>>> 
>>>> 
>>>> 
>>>> Best,
>>>> Alex.
>>>> ___
>>>> Aleksandra Oranskaia (M.Sc.)
>>>> ChemS PhD student, KAUST
>>>> Phone: +966 50 1335254
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> ------------------------------------------------------------------------
>>>> This message and its contents, including attachments are intended solely for the original recipient. If you are not the intended recipient or have received this message in error, please notify me immediately and delete this message from your computer system. Any unauthorized use or distribution is prohibited. Please consider the environment before printing this email.
>>>> 
>>>> _______________________________________________
>>>> users mailing list
>>>> users at lists.quantum-espresso.org <mailto:users at lists.quantum-espresso.org>
>>>> https://lists.quantum-espresso.org/mailman/listinfo/users
>>>> 
>> 
>> 
>> -- 
>> ----------------------------------------------------------------------
>> Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL
>> Director, National Centre for Competence in Research NCCR MARVEL, EPFL
>> http://theossrv1.epfl.ch/Main/Contact <http://theossrv1.epfl.ch/Main/Contact> http://nccr-marvel.ch/en/project <http://nccr-marvel.ch/en/project>
> 


-- 

This message and its contents, including attachments are intended solely 
for the original recipient. If you are not the intended recipient or have 
received this message in error, please notify me immediately and delete 
this message from your computer system. Any unauthorized use or 
distribution is prohibited. Please consider the environment before printing 
this email.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190112/8c588a77/attachment.html>