[QE-users] NEB: Atoms do not get relaxed

Ari P Seitsonen Ari-Paavo.Seitsonen at cnrs.fr
Thu Apr 27 20:42:21 CEST 2023


I think that I found it: Switching the 'restart_mode' from 'restart' to 
'from_scratch' in the name list '&control' "released" the atoms to relax 
properly.

   I have no idea why this should have an effect on the relaxation: I 
always keep it as 'restart' in the calculations with the NEB, and the code 
'pw.x' performs the restart if the files needed for a restart are present, 
otherwise the calculation starts from the given configuration and with the 
same initialisation of the electronic structure as with 'from_scratch', so 
I was expecting the same to happen with the NEB. My mistake, "never expect 
the logic of the developer to be the same as your own". Now happily 
computing all kinds of barriers and paths! :)

     Greetings from Paris and Grazie a Giuseppe again,

        apsi

-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
                             Слава Україні


On Thu, 27 Apr 2023, Ari P Seitsonen wrote:

>
> Dear Giuseppe,
>
>  Thank you for the comment. One of my concerns was that the energy in the 
> images does not really change:
>
> grep -w '  4  ' neb.out
>         4     -78481.6171806           29.739170            F
>         4     -78481.6171811           29.738137            F
>         4     -78481.6171811           29.736993            F
>         4     -78481.6171811           29.734436            F
>         4     -78481.6171812           29.727381            F
>         4     -78481.6171809           29.718296            F
>         4     -78481.6171808           29.712880            F
>         4     -78481.6171810           29.713046            F
>         4     -78481.6171810           29.719086            F
>         4     -78481.6171811           29.729326            F
>         4     -78481.6171812           29.740800            F
>         4     -78481.6171805           29.752151            F
>         4     -78481.6171812           29.761356            F
>         4     -78481.6171812           29.769083            F
>         4     -78481.6171805           29.774024            F
>         4     -78481.6171812           29.776518            F
>         4     -78481.6171800           29.776688            F
>         4     -78481.6171811           29.773882            F
>
> There are forces on the atoms, here again on one of them on the fourth image:
>
>     atom   84 type  2   force =    -0.00407872    0.00078289   -0.00018719
>     atom   84 type  2   force =    -0.00407494    0.00078224   -0.00018711
>     atom   84 type  2   force =    -0.00407554    0.00078237   -0.00018713
>     atom   84 type  2   force =    -0.00407560    0.00078248   -0.00018709
>     atom   84 type  2   force =    -0.00407511    0.00078229   -0.00018712
>     atom   84 type  2   force =    -0.00407619    0.00078255   -0.00018726
>     atom   84 type  2   force =    -0.00407623    0.00078272   -0.00018718
>     atom   84 type  2   force =    -0.00407528    0.00078225   -0.00018721
>     atom   84 type  2   force =    -0.00407564    0.00078257   -0.00018705
>     atom   84 type  2   force =    -0.00407500    0.00078222   -0.00018713
>     atom   84 type  2   force =    -0.00407577    0.00078249   -0.00018711
>     atom   84 type  2   force =    -0.00406905    0.00078120   -0.00018643
>     atom   84 type  2   force =    -0.00407551    0.00078243   -0.00018722
>     atom   84 type  2   force =    -0.00407271    0.00078203   -0.00018711
>     atom   84 type  2   force =    -0.00407796    0.00078294   -0.00018727
>     atom   84 type  2   force =    -0.00407607    0.00078266   -0.00018716
>     atom   84 type  2   force =    -0.00407437    0.00078207   -0.00018715
>     atom   84 type  2   force =    -0.00407642    0.00078274   -0.00018719
>
> The 'path length' indeed keeps on increasing - what is this, if the energies 
> of the images remain the same?
>
>     initial path length           =  5.0000 bohr
>     path length          =  5.000 bohr
>     path length          =  5.000 bohr
>     path length          =  5.003 bohr
>     path length          =  5.013 bohr
>     path length          =  5.094 bohr
>     path length          =  5.337 bohr
>     path length          =  5.702 bohr
>     path length          =  6.163 bohr
>     path length          =  6.695 bohr
>     path length          =  7.277 bohr
>     path length          =  7.897 bohr
>     path length          =  8.546 bohr
>     path length          =  9.216 bohr
>     path length          =  9.904 bohr
>     path length          = 10.606 bohr
>     path length          = 11.321 bohr
>     path length          = 12.047 bohr
>     path length          = 12.783 bohr
>
> Probably I am doing something really stupid somewhere... I will try to look 
> at the source code next and try to understand somewhat better for what is 
> done in the code, the meaning of the variables and so on.
>
>    Greetings,
>
>       apsi
>
> -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
>  Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
>                            Слава Україні
>
>
> On Thu, 27 Apr 2023, Giuseppe Mattioli wrote:
>
>> 
>> Dear Ari
>> The NEB is not exactly "stuck"
>>
>>>      path length          =  8.546 bohr
>>>      path length          =  9.216 bohr
>>>      path length          =  9.904 bohr
>> 
>> The MEP length keeps increasing. Maybe a look into the trajectory files can 
>> tell you what's happening. If any other source of misbehavior can be 
>> excluded, try to go ahead (maybe fixing the position of "useless" atoms). 
>> I've experienced NEB in large systems converged after several hundreds 
>> steps.
>> HTH
>> Giuseppe
>> 
>> Quoting Ari P Seitsonen <Ari-Paavo.Seitsonen at cnrs.fr>:
>> 
>>> Dear Users,
>>>
>>>   I have a problem in the calculations with the method NEB: The atoms do 
>>> not get relaxed even if there are clearly forces on the atoms. The 
>>> electronic structure converges well, and the relaxation of the ions 
>>> follows nicely when the NEB is not used.
>>>
>>>   I have in the input
>>> 
>>> BEGIN
>>> BEGIN_PATH_INPUT
>>> &PATH
>>>   restart_mode      = 'restart'
>>>   string_method     = 'neb'
>>>   nstep_path        =  30
>>>   ds                = 0.2
>>>   opt_scheme        = "broyden"
>>>   num_of_images     = 8
>>>   k_max             = 0.3
>>>   k_min             = 0.2
>>>   CI_scheme         = "auto"
>>> !  minimum_image     = .true.
>>>   first_last_opt    = .true.
>>>   path_thr          = 0.1
>>> /
>>> END_PATH_INPUT
>>> 
>>> and
>>>
>>>  &ions
>>>  /
>>> 
>>> I have tried with different values of 'num_of_images', 'k_min' and 
>>> 'k_max', with the climbing image and without. One calculation - a 
>>> diffusion path of one ad-atom on a surface - worked well, all other become 
>>> stuck with the output like:
>>> 
>>>
>>>      ------------------------------ iteration  12 
>>> ------------------------------
>>>
>>>      cpu =  0   tcpu =   9589.5    self-consistency for image   1
>>>      cpu =  0   tcpu =   9656.9    self-consistency for image   3
>>>      cpu =  0   tcpu =   9823.9    self-consistency for image   5
>>>      cpu =  0   tcpu =  10043.8    self-consistency for image   7
>>>
>>>      activation energy (->) =  10.012055 eV
>>>      activation energy (<-) =   6.833374 eV
>>>
>>>      image        energy (eV)        error (eV/A)        frozen
>>>
>>>          1     -78491.5920678            0.000987            F
>>>          2     -78490.3334035            7.018378            F
>>>          3     -78486.4101831           18.058881            F
>>>          4     -78481.6171805           29.752151            F
>>>          5     -78481.5800131           29.996389            F
>>>          6     -78485.0169815           13.489042            F
>>>          7     -78487.7020649            7.675245            F
>>>          8     -78488.4133872            8.072137            F
>>>
>>>      climbing image =  5
>>>
>>>      path length          =  8.546 bohr
>>>      inter-image distance =  1.221 bohr
>>>
>>>      ------------------------------ iteration  13 
>>> ------------------------------
>>>
>>>      cpu =  0   tcpu =  10165.0    self-consistency for image   1
>>>      cpu =  0   tcpu =  10267.1    self-consistency for image   3
>>>      cpu =  0   tcpu =  10487.8    self-consistency for image   5
>>>      cpu =  0   tcpu =  10710.2    self-consistency for image   7
>>>
>>>      activation energy (->) =  10.011390 eV
>>>      activation energy (<-) =   6.833374 eV
>>>
>>>      image        energy (eV)        error (eV/A)        frozen
>>>
>>>          1     -78491.5914028            0.000493            F
>>>          2     -78490.3334036            7.099250            F
>>>          3     -78486.4101837           17.820470            F
>>>          4     -78481.6171812           29.761356            F
>>>          5     -78481.5800130           29.998875            F
>>>          6     -78485.0169810           13.636010            F
>>>          7     -78487.7020648            7.675132            F
>>>          8     -78488.4133870            8.071850            F
>>>
>>>      climbing image =  5
>>>
>>>      path length          =  9.216 bohr
>>>      inter-image distance =  1.317 bohr
>>>
>>>      ------------------------------ iteration  14 
>>> ------------------------------
>>>
>>>      cpu =  0   tcpu =  10843.3    self-consistency for image   1
>>>      cpu =  0   tcpu =  10962.6    self-consistency for image   3
>>>      cpu =  0   tcpu =  11183.6    self-consistency for image   5
>>>      cpu =  0   tcpu =  11420.4    self-consistency for image   7
>>>
>>>      activation energy (->) =  10.011390 eV
>>>      activation energy (<-) =   6.833374 eV
>>>
>>>      image        energy (eV)        error (eV/A)        frozen
>>>
>>>          1     -78491.5914032            0.000582            F
>>>          2     -78490.3334034            7.139704            F
>>>          3     -78486.4101836           17.548234            F
>>>          4     -78481.6171812           29.769083            F
>>>          5     -78481.5800132           29.998295            F
>>>          6     -78485.0169814           13.849319            F
>>>          7     -78487.7020642            7.669197            F
>>>          8     -78488.4133871            8.072137            F
>>>
>>>      climbing image =  5
>>>
>>>      path length          =  9.904 bohr
>>>      inter-image distance =  1.415 bohr
>>>
>>>      ------------------------------ iteration  15 
>>> ------------------------------
>>> 
>>> ...
>>>
>>>   So the atoms do not move, the electronic structure is converged in few 
>>> iterations, but the energies of individual images change in the micro-eVs.
>>>
>>>   Any idea would be appreciated; thank you in advance for any hints.
>>>
>>>     Greetings,
>>>
>>>        apsi
>>> 
>>>
>> 
>> -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
>>>   Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
>>>                             Слава Україні
>> 
>> 
>> 
>> GIUSEPPE MATTIOLI
>> CNR - ISTITUTO DI STRUTTURA DELLA MATERIA
>> Via Salaria Km 29,300 - C.P. 10
>> I-00015 - Monterotondo Scalo (RM)
>> Mob (*preferred*) +39 373 7305625
>> Tel + 39 06 90672342 - Fax +39 06 90672316
>> E-mail: <giuseppe.mattioli at ism.cnr.it>
>> 
>> _______________________________________________
>> The Quantum ESPRESSO community stands by the Ukrainian
>> people and expresses its concerns about the devastating
>> effects that the Russian military offensive has on their
>> country and on the free and peaceful scientific, cultural,
>> and economic cooperation amongst peoples
>> _______________________________________________
>> 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


More information about the users mailing list