[QE-users] using charge density from file or not ; bad behavior

[Paolo Giannozzi]

Please note that restarting from a previous incomplete calculation works
only if the previous calculation was cleanly stopped, that is: maximum cpu,
specified in input with the max_seconds variable, reached; or, the presence
of a file "prefix".EXIT in the output directory is detected by the running
code. In any other case, there is no guarantee that the code will be able
to properly restart.

Paolo

On Tue, Feb 23, 2021 at 12:27 AM Sergey Lisenkov <proffess at yandex.ru> wrote:

> Dear all,
>
> I noticed the following thing. When calculations abort (6.7 version) due
> to some reason (queue's runlimit has been reached or node crash), I cannot
> really use "charge density " file as "startingpot". This happens when I
> deal with magnetic systems. This is what happens:
>
>     starting charge  556.10148, renormalised to  556.00000
>
>      negative rho (up, down):  5.518E+01 5.758E+01
>      Starting wfcs are  448 randomized atomic wfcs
>
>      total cpu time spent up to now is       24.8 secs
>
>      Self-consistent Calculation
>
>      iteration #  1     ecut=    90.00 Ry     beta= 0.20
>      Davidson diagonalization with overlap
>      c_bands:  3 eigenvalues not converged
>      c_bands:  1 eigenvalues not converged
>      c_bands:  3 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      c_bands:  1 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      ethr =  1.00E-05,  avg # of iterations = 56.9
>
>      negative rho (up, down):  3.409E+01 3.552E+01
>
>      total cpu time spent up to now is      582.1 secs
>
>      total energy              =   -1746.57738261 Ry
>      estimated scf accuracy    <    6330.40806007 Ry
>
>      total magnetization       =    -0.63 Bohr mag/cell
>      absolute magnetization    =    12.88 Bohr mag/cell
>
> ....
> and calculations go south. If I remove all prefix files, calculations go
> in the different direction:
>
> ...
> Cannot read rho : file not found
>
>      Initial potential from superposition of free atoms
>
>      starting charge  555.99286, renormalised to  556.00000
>      Starting wfcs are  448 randomized atomic wfcs
>
>      total cpu time spent up to now is       27.2 secs
>
>      Self-consistent Calculation
>
>      iteration #  1     ecut=    90.00 Ry     beta= 0.20
>      Davidson diagonalization with overlap
>      c_bands:  2 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      c_bands:  4 eigenvalues not converged
>      c_bands:  3 eigenvalues not converged
>      c_bands:  3 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      c_bands:  2 eigenvalues not converged
>      c_bands:  3 eigenvalues not converged
>      ethr =  1.00E-05,  avg # of iterations = 40.0
>
>      total cpu time spent up to now is      251.8 secs
>
>      total energy              =   -2138.05525957 Ry
>      estimated scf accuracy    <      11.57578252 Ry
>
>      total magnetization       =    -0.00 Bohr mag/cell
>      absolute magnetization    =    21.13 Bohr mag/cell
>
> In on other words, even if my calculation do converge (SCF) nicely, from
> each ionic step to another, if something happens to the job, I cannot
> restart.
>
> Is it an expected behavior?
>
> Thanks,
>  Sergey
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 206, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20210224/98f05f80/attachment.html>