[Pw_forum] where is the end of an unconverged NEB calculation
lfhuang
lfhuang at theory.issp.ac.cn
Fri Nov 6 12:32:06 CET 2009
Dear vega lew:
I see you use normal NEB method, why don't use the Climbing-Image NEB? CI-NEB will let you find the saddle-point state. And I see the path length is about 5.0 anstrong, which is a little long path that 7 images in no-CI maybe not many enough if the potential energy surface is not smooth enough. Also, increasing the spring constants k_min and k_max and decreasing the atomic masses in the input file may help some. The meanings of these parameters can be found in the original papers on Dimer and NEB methods, which can be found in the bibliography in the Q_E homepage.
In addition, It seems I have not made you understand me in the previous email, and I meant the convergence of the electronic-structure calculations for every image can be checked only when you open the PW.out files in the directory ./tmp/**_i (i=1,2...7). The potential barrier is just the energy difference between the saddle-point state (not well difined in the no-CI method) and the initial state, so when there is something wrong with the convergence of E_ac, there must be something wrong with the energy convergence of the "saddle-point state" and the initital state, which can be modified by changing some parameters e.g. k-points numbers, smearing, nband. Or there may
be some defects in the pseudopotential and you should better change for another one. Probably the no-CI method also can result in this kind of convergence problem due to that the structure with highest energy can not stablized during your calculations.
One important note: the NEB is cpu time consuming, and can eaily cost over 100 iterations to converge if your parameters are not optimized, so you'd better take some tests on smaller systems to find the optimum parameters before more complicated calculations.
May these be helpful for you!
Best Wishes!
Yours Sincerely
L.F.Huang
> My convergence of every single NEB iteration is very well. My convergence
> problem is about the calculated "activation energy" (I call it as energy
> barrier in last mail.), not for the energy convergence of every image.
>
> here is a few lines from my output file,
>
> Program PWSCF v.4.0.3 starts ...
> Today is 8Oct2009 at 23:53:20
>
> Parallel version (MPI)
>
> Number of processors in use: 5
> path-images division: nimage = 5
>
> For Norm-Conserving or Ultrasoft (Vanderbilt) Pseudopotentials or PAW
>
> Current dimensions of program pwscf are:
> Max number of different atomic species (ntypx) = 10
> Max number of k-points (npk) = 40000
> Max angular momentum in pseudopotentials (lmaxx) = 3
>
> gamma-point specific algorithms are used
>
>
> initial path length = 7.8290 bohr
> initial inter-image distance = 1.3048 bohr
>
> calculation = neb
> restart_mode = from_scratch
> opt_scheme = quick-min
> num_of_images = 7
> nstep = 9999
> CI_scheme = no-CI
> first_last_opt = F
> coarse-grained phase-space = F
> use_freezing = F
> ds = 1.0000 a.u.
> k_max = 0.1000 a.u.
> k_min = 0.1000 a.u.
> suggested k_max = 0.6169 a.u.
> suggested k_min = 0.6169 a.u.
> path_thr = 0.0500 eV / A
>
> ------------------------------ iteration 1
> ------------------------------
>
> cpu = 0 tcpu = 0.2 self-consistency for image 1
> cpu = 3 tcpu = 0.2 self-consistency for image 5
> cpu = 4 tcpu = 0.2 self-consistency for image 6
> cpu = 2 tcpu = 0.2 self-consistency for image 4
> cpu = 1 tcpu = 0.2 self-consistency for image 7
> cpu = 0 tcpu = 23616.8 self-consistency for image 2
> cpu = 1 tcpu = 25603.1 self-consistency for image 3
>
> activation energy (->) = 3.272591 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119084.9563724 2.640583 F
> 3 -119083.3975295 2.385012 F
> 4 -119082.3894835 2.896402 F
> 5 -119083.2192599 2.998904 F
> 6 -119084.7252314 2.403066 F
> 7 -119085.3804446 0.025165 T
>
> path length = 7.829 bohr
> inter-image distance = 1.305 bohr
>
> ------------------------------ iteration 2
> ------------------------------
>
> cpu = 2 tcpu = 29648.3 self-consistency for image 4
> cpu = 0 tcpu = 42567.1 self-consistency for image 2
> cpu = 1 tcpu = 44477.8 self-consistency for image 3
> cpu = 3 tcpu = 30626.8 self-consistency for image 5
> cpu = 4 tcpu = 32197.2 self-consistency for image 6
>
> activation energy (->) = 3.093816 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119085.0987352 2.105271 F
> 3 -119083.6078363 2.170459 F
> 4 -119082.5682586 2.050101 F
> 5 -119083.5224117 1.596776 F
> 6 -119084.8769055 1.343862 F
> 7 -119085.3804446 0.025165 T
>
> path length = 7.843 bohr
> inter-image distance = 1.307 bohr
> ......
> ......
> ------------------------------ iteration 120
> ------------------------------
>
> cpu = 2 tcpu = ******** self-consistency for image 4
> cpu = 0 tcpu = ******** self-consistency for image 2
> cpu = 3 tcpu = ******** self-consistency for image 5
> cpu = 4 tcpu = ******** self-consistency for image 6
> cpu = 1 tcpu = ******** self-consistency for image 3
>
> activation energy (->) = 1.786596 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119085.4786718 0.223126 F
> 3 -119083.9049529 0.170985 F
> 4 -119083.8754787 0.076161 F
> 5 -119085.1083173 0.075946 F
> 6 -119085.3050720 0.015511 F
> 7 -119085.3804446 0.025165 T
>
> path length = 9.729 bohr
> inter-image distance = 1.621 bohr
>
> ------------------------------ iteration 121
> ------------------------------
>
> cpu = 2 tcpu = ******** self-consistency for image 4
> cpu = 3 tcpu = ******** self-consistency for image 5
> cpu = 0 tcpu = ******** self-consistency for image 2
> cpu = 4 tcpu = ******** self-consistency for image 6
> cpu = 1 tcpu = ******** self-consistency for image 3
>
> activation energy (->) = 1.775789 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119085.4781477 0.130868 F
> 3 -119083.8862860 0.275250 F
> 4 -119083.8905252 0.089256 F
> 5 -119085.1088347 0.047586 F
> 6 -119085.3051231 0.015512 F
> 7 -119085.3804446 0.025165 T
>
> path length = 9.725 bohr
> inter-image distance = 1.621 bohr
>
> ------------------------------ iteration 122
> ------------------------------
>
> cpu = 2 tcpu = ******** self-consistency for image 4
> cpu = 3 tcpu = ******** self-consistency for image 5
> cpu = 0 tcpu = ******** self-consistency for image 2
> cpu = 1 tcpu = ******** self-consistency for image 3
> cpu = 4 tcpu = ******** self-consistency for image 6
>
> activation energy (->) = 1.791716 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119085.4766388 0.150020 F
> 3 -119083.8703582 0.360096 F
> 4 -119083.9053588 0.136487 F
> 5 -119085.1096015 0.054464 F
> 6 -119085.3051878 0.017319 F
> 7 -119085.3804446 0.025165 T
> ......
> ......
> ......
> ------------------------------ iteration 160
> ------------------------------
>
> cpu = 2 tcpu = ******** self-consistency for image 4
> cpu = 0 tcpu = ******** self-consistency for image 2
> cpu = 3 tcpu = ******** self-consistency for image 5
> cpu = 1 tcpu = ******** self-consistency for image 3
> cpu = 4 tcpu = ******** self-consistency for image 6
>
> activation energy (->) = 2.015566 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119085.4084826 0.122618 F
> 3 -119083.6465085 0.151901 F
> 4 -119084.1207907 0.114239 F
> 5 -119085.1378794 0.024820 F
> 6 -119085.3093981 0.014540 F
> 7 -119085.3804446 0.025165 T
>
> path length = 9.760 bohr
> inter-image distance = 1.627 bohr
>
> ------------------------------ iteration 161
> ------------------------------
>
> cpu = 2 tcpu = ******** self-consistency for image 4
> cpu = 3 tcpu = ******** self-consistency for image 5
> cpu = 0 tcpu = ******** self-consistency for image 2
> cpu = 4 tcpu = ******** self-consistency for image 6
> cpu = 1 tcpu = ******** self-consistency for image 3
>
> activation energy (->) = 2.019415 eV
> activation energy (
> image energy (eV) error (eV/A) frozen
>
> 1 -119085.6620746 0.024462 T
> 2 -119085.4066890 0.190727 F
> 3 -119083.6426594 0.205492 F
> 4 -119084.1216091 0.058992 F
> 5 -119085.1383781 0.022947 F
> 6 -119085.3094981 0.024415 F
> 7 -119085.3804446 0.025165 T
>
> path length = 9.765 bohr
> inter-image distance = 1.627 bohr
> ......
> ......
> ......
>
> as I said last mail, the activation energy goes to 1.49 eV from the intial
> 2.99 eV in about 120 iterations. Then the energy barrier rise to 1.73 eV bit
> by bit in following 40 iterations. And the calculation is still running
> while this value is still going up.
>
> How could I deal with this?
>
> thank you for reading and reply
>
> best,
>
> vega
> On Fri, Nov 6, 2009 at 4:50 PM, lfhuang wrote:
>
> > Dear vega lew:
> > Have you gone into the ./tmp directory to see the energy convergence of
> > every image? The calculational statuses of all the images can be found
> > there, and I guess your convergence problem may be due to the same reasons
> > as ordinary electronic structrue calculations, which may be modified by
> > change some parameters e.g. k-points numbers, smearing, nband. Or there may
> > be some defects in the pseudopotential and you should better change for
> > another one. May these be helpful.
> >
> > Best Wishes!
> > Yours Sincerely
> > L.F.Huang
> >
> > > Date: Fri, 6 Nov 2009 11:35:39 +0800
> > > From: vega lew
> > > Subject: [Pw_forum] where is the end of an unconverged NEB calculation
> > > To: PWSCF Forum
> > > Message-ID:
> > >
> > > Content-Type: text/plain; charset="iso-8859-1"
> > >
> > > Dear all,
> > >
> > > During my NEB calculations, I found there are two possible situation.
> > First,
> > > the calculation could be converged after about 100 iterations. Second,
> > the
> > > calculation never converged in more than 300 iterations. And for the
> > second
> > > case, the energy barrier went down and down to a certain value, and then
> > the
> > > reported energy barrier went up gradually. For example in my resent job,
> > the
> > > energy barrier goes to 1.49 eV from the intial 2.99 eV in about 120
> > > iterations. Then the energy barrier rise to 1.73 eV bit by bit in
> > following
> > > 40 iterations. And the calculation is still running while this value is
> > > still going up. How could I deal with this situation? Could I take the
> > > lowest value as the energy barrier of the reaction? How could I know
> > whether
> > > there will be another smaller value in the longer calculation?
> > >
> > > thank you for reading. Ang hints will be deeply appreciated.
> > >
> > > best,
> > >
> > > vega
> > >
> > > --
> > >
> > ==================================================================================
> >
> > > Vega Lew ( weijia liu)
> > > Graduate student
> > > State Key Laboratory of Materials-oriented Chemical Engineering
> > > College of Chemistry and Chemical Engineering
> > > Nanjing University of Technology, 210009, Nanjing, Jiangsu, China
> > >
> > ******************************************************************************************************************
> >
> > > Email: vegalew at gmail.com
> > > Office: Room A705, Technical Innovation Building, Xinmofan Road 5#,
> > Nanjing,
> > > Jiangsu, China
> > >
> > ******************************************************************************************************************
> >
> >
> > ------
> > ======================================================================
> > L.F.Huang(???) ph.D candidate, DFT and phonon physics
> > ======================================================================
> > Add: Research Laboratory for Computational Materials Sciences,
> > Instutue of Solid State Physics,the Chinese Academy of Sciences,
> > P.O.Box 1129, Hefei 230031, P.R.China
> > Tel: 86-551-5591464-328(office)
> > Fax: 86-551-5591434
> > Web: http://theory.issp.ac.cn> (website of our theory group)
> > http://www.issp.ac.cn> (website of our institute)
> > ======================================================================
> >
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://www.democritos.it/mailman/listinfo/pw_forum
> >
> >
>
>
> --
> ==================================================================================
> Vega Lew ( weijia liu)
> Graduate student
> State Key Laboratory of Materials-oriented Chemical Engineering
> College of Chemistry and Chemical Engineering
> Nanjing University of Technology, 210009, Nanjing, Jiangsu, China
> ******************************************************************************************************************
> Email: vegalew at gmail.com
> Office: Room A705, Technical Innovation Building, Xinmofan Road 5#, Nanjing,
> Jiangsu, China
> ******************************************************************************************************************
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: http://www.democritos.it/pipermail/pw_forum/attachments/20091106/4210da57/attachment.htm
>
> ------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
> End of Pw_forum Digest, Vol 29, Issue 15
> ****************************************
>
------
======================================================================
L.F.Huang(黄良锋) ph.D candidate, DFT and phonon physics
======================================================================
Add: Research Laboratory for Computational Materials Sciences,
Instutue of Solid State Physics,the Chinese Academy of Sciences,
P.O.Box 1129, Hefei 230031, P.R.China
Tel: 86-551-5591464-328(office)
Fax: 86-551-5591434
Web: http://theory.issp.ac.cn (website of our theory group)
http://www.issp.ac.cn (website of our institute)
======================================================================
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20091106/2320d006/attachment.html>
More information about the users
mailing list