[Pw_forum] vc-relax and LDA+U: Segmentation fault

Matteo Cococcioni matteo at umn.edu
Thu Sep 25 06:30:43 CEST 2008


Dear Marino,

I have asked people in R. Wentzcovitch's group and they said that damp-w 
vc-relax had no problem with LDA+U using
3.2.3. I think that something like what Lorenzo was explaining happened 
in between. So I guess the problem can be tracked
and solved just looking at the list of last madifications.
don't take this as an encouragement to use previous versions of the 
code.....

regards,

Matteo



Marino Vetuschi Zuccolini wrote:
> Dear Matteo,
>
> from documentation appears that U_projection_type = 'atomic' is the  
> default and then I never declared it in the SYSTEM. When tprnfor and  
> tstress were declared .true. the segmentation fault happened.  
> If .false. happens again.
>
> Then it stop here at
> ...
>       0.2500   0.2500   0.0028   0.0027  -0.0044  -0.0004  -0.0004   
> -0.0002
>       0.0000   0.0000   0.0000   0.0000   0.0000   0.0000   0.0000    
> 0.0000
>       0.0000   0.0000   0.0000   0.0000   0.0000   0.0000   0.0000    
> 0.0000
>       0.0000
>
>       the Fermi energy is     9.3699 ev
>
> !    total energy              = -2166.89264188 Ry
>       Harris-Foulkes estimate   = -2166.89264402 Ry
>       estimated scf accuracy    <     0.00000264 Ry
>
>       total magnetization       =     0.00 Bohr mag/cell
>       absolute magnetization    =    22.98 Bohr mag/cell
>
> I used damp-w as cell-dynamic and default for ion-dynamic. I will try  
> bfgs next.
>
> Thanks,
>
> m.
>
> On 24/set/08, at 16:59, Matteo Cococcioni wrote:
>
>   
>> Dear Marino,
>>
>> I have never run a vc-relax simulation with LDA+U but I know of  
>> people who have done that successfully.
>> if U_projection_type = 'atomic' the code should compute forces and
>> stresses with LDA+U no problem. Do you obtain forces and stresses?
>> Also I'm not sure I have understood where the code stops. Does the  
>> segmentation fault appear after the first scf cycle (and after the  
>> first calculation
>> of stresses and forces)?
>> Also what kind of dynamics have you picked?
>>
>> regards,
>>
>> Matteo
>>
>>
>> Marino Vetuschi Zuccolini wrote:
>>     
>>> Hello to all,
>>> from a very recent email posted on the forum it seem to me that   
>>> relaxation (maybe vc-relax) is possible using LDA+U, and that   
>>> experimental implementation of phonons with LDA+U is going on (for   
>>> this I hope to be a beta-tester if my structure converge..).
>>> My question is: does anyone of you have experience a Segmentation   
>>> fault using vc-relax and LDA+U just after the wfc convergence has  
>>> been  achieved, starting 'from_scratch'?
>>> I'm using 4.0CVS in serial mode, but it also in 4.0.2 version it   
>>> happens.
>>> Any help is welcome.
>>>
>>> Marino
>>>
>>>
>>> *******************************************************
>>> Marino Vetuschi Zuccolini
>>> zucco at dipteris.unige.it
>>> Researcher / Geochemist
>>> Laboratory of Geochemistry
>>>
>>> DIPartimento per lo studio della TErra e delle sue RISorse -   
>>> Università  di Genova
>>> Tel. ++39 010 3538136 Fax. ++39 010 352169
>>> Corso Europa 26, 16132 - Genova - Italy
>>>
>>>
>>> _______________________________________________
>>> Pw_forum mailing list
>>> Pw_forum at pwscf.org
>>> http://www.democritos.it/mailman/listinfo/pw_forum
>>>
>>>       
>> <matteo.vcf>_______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
>>     
>
>
>
> *******************************************************
> Marino Vetuschi Zuccolini
> zucco at dipteris.unige.it
> Researcher / Geochemist
> Laboratory of Geochemistry
>
> DIPartimento per lo studio della TErra e delle sue RISorse -  
> Università  di Genova
> Tel. ++39 010 3538136 Fax. ++39 010 352169
> Corso Europa 26, 16132 - Genova - Italy
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>   




More information about the users mailing list