Re: [Wien] Fw: Optical property calculation when spin orbit added
"x lapwso -up" creates uplapwso.def with WIEN2k 17.1. If your using WIEN2k 14.1 as mentioned at http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16027.html it seems that version creates lapwso.def instead. Using "./" is not a requirement, I just suggested trying it. That is because if I recall correctly, there were many reports in the mailing list of users having frequent success using "./", but more failures when using something else. Your post below seems to be a mix of unorganized information. In the error message you previously reported [ http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16062.html ], there was: /scratch/10820461.yak.local/SCFsp-U-SO.vectorsoup It sounds like this is a result of "export SCRATCH=$PBS_O_WORKDIR" in your job script. The lapwso.def should have the same path above, such that a non-empty SCFsp-U-SO.vectorsoup should exist in /scratch/10820461.yak.local/ if it was generated by lapwso without any errors. However, the lapwso.def in your post below has instead: /global/scratch/WIEN2k/NiO-14/NiO-14.vectorsoup My guess is this is a result of "export SCRATCH=./" in your job script. It sounds like this may have resolved the previous "can't open unit" error with SCFsp-U-SO.vectorsoup. However, you say that the case.vectorsoup and case.vectorsodn files now in your working directory are empty. This sounds like some other error has occurred in the running of "x lapwso -up". Did you check the *.error files again to see if a new error appeared? The error when using $PBS_O_WORKDIR, did you check what $PBS_O_WORKDIR is (perhaps using echo or a similar technique) each time you run a job script? Is it static and always the same or does itchange every time you run a new job (e.g., the 10820461 in "10820461.yak.local" stay the same or change to something else like say "10820465")? If it is static, you should be able to run separate job scripts for the scf and optic. However, if it is dynamically set, I expect that both the scf cycles and optic would need to happen in the same job script. If the path changed for each new job and you ran scf and optic calculation separately, this might possibly be what caused the error. The updates page on the WIEN2k website [ http://susi.theochem.tuwien.ac.at/reg_user/updates/ ] under VERSION_17.1: 30.6.2017 has: SRC_optic: opmain.f (SO calculations are labeled "SO" in case.symmat/mat_diag files to fix normalization) So with the optic SO calculation that you trying to do [ http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16062.html ], this seems to indicate that the older WIEN2k versions (like 14.1) may have a normalization issue. Therefore, it is up to you if want to take on the risk and challenges yourself of using an old version instead of using 17.1. I mentioned before about some of the headaches of trying to maintain an old version of the code by yourself [ http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg11071.html ]. On 7/12/2017 11:10 AM, shaymlal dayananda wrote: Dear Dr. Abo And Dr. Bhamu Thank you very much for your comments and advices. I was trying to follow your instructions and waited to see their results before to reply. After some efforts i could get case.vectorsoup, case.vectorsodn files are now in my working directory. But still they are empty. I am using remote computers and thus I do not have much freedom to do my own. Our supporter in the supercomputers adviced me not to add "export SCRATCH=./" as (according to him) it send the files to logging nodes. Instead he added "export SCRATCH=$PBS_O_WORKDIR" in my jobscript. This created the empty case.vectorsoup, case.vectorsodn in the working directory. Could you please tell me ; Is the requirement of "export SCRATCH=./" only to get those files to working directory or does it has any issue with those generated files. ?? Further I do not see uplapwso.def in the directory!! Part of lapwso.def is below. 4 ,'NiO-14.in1', 'old','formatted',0 5 ,'NiO-14.inso', 'old','formatted',0 6 ,'NiO-14.outputso', 'unknown','formatted',0 8 ,'NiO-14.scfso', 'unknown','formatted',0 9 ,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectordn', 'old', 'unformatted',9000 10 ,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectorup', 'unknown', 'unformatted',9000 18,'NiO-14.vspdn', 'old','formatted',0 19,'NiO-14.vspup', 'unknown','formatted',0 20 ,'NiO-14.struct','old','formatted',0 22,'NiO-14.vnsdn', 'old','formatted',0 23,'NiO-14.vnsup', 'unknown','formatted',0 41,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectorsodn', 'unknown','unformatted',9000 42,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectorsoup', 'unknown','unformatted',9000 44,'NiO-14.vect1', 'unknown','unformatted',9000 45,'NiO-14.normsodn', 'unknown','formatted',0 Thank you in advance. Chami ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at
Re: [Wien] Fw: Optical property calculation when spin orbit added
Dear Dr. Abo And Dr. Bhamu Thank you very much for your comments and advices. I was trying to follow your instructions and waited to see their results before to reply. After some efforts i could get case.vectorsoup, case.vectorsodn files are now in my working directory. But still they are empty. I am using remote computers and thus I do not have much freedom to do my own. Our supporter in the supercomputers adviced me not to add "export SCRATCH=./" as (according to him) it send the files to logging nodes. Instead he added "export SCRATCH=$PBS_O_WORKDIR" in my jobscript. This created the empty case.vectorsoup, case.vectorsodn in the working directory. Could you please tell me ; Is the requirement of "export SCRATCH=./" only to get those files to working directory or does it has any issue with those generated files. ??Further I do not see uplapwso.def in the directory!! Part of lapwso.def is below. 4 ,'NiO-14.in1', 'old', 'formatted',0 5 ,'NiO-14.inso', 'old', 'formatted',0 6 ,'NiO-14.outputso', 'unknown','formatted',0 8 ,'NiO-14.scfso', 'unknown','formatted',0 9 ,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectordn', 'old', 'unformatted',9000 10 ,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectorup', 'unknown', 'unformatted',9000 18,'NiO-14.vspdn', 'old','formatted',0 19,'NiO-14.vspup', 'unknown','formatted',0 20 ,'NiO-14.struct', 'old', 'formatted',0 22,'NiO-14.vnsdn', 'old','formatted',0 23,'NiO-14.vnsup', 'unknown','formatted',0 41,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectorsodn', 'unknown','unformatted',9000 42,'/global/scratch/WIEN2k/NiO-14/NiO-14.vectorsoup', 'unknown','unformatted',9000 44,'NiO-14.vect1', 'unknown','unformatted',9000 45,'NiO-14.normsodn', 'unknown','formatted',0 Thank you in advance. Chami On Sunday, July 9, 2017 10:10 AM, Gavin Abowrote: You should be able to use "x optic -so [-up]" without "-c", because the x script part of the command should automatically detect and add it if it is needed [1]. If you are interested in the details of how that is done, I think it is lines 377 to 381 in the x_lapw script file of WIEN2k 17.1. Also, there should be no difference between "opticc" and "optic -c". As seen on line 139 in x_lapw, "-c" sets "cmplx = c". set cmplx = c On lines 422 to 423, "opticc" also sets "cmplx = c": else if ($command == opticc) then set cmplx = c So both "opticc" and "optic -c" will run the executable "opticc" as seen by lines 1027 to 1028: case optic: set exe = $command$cmplx$para as $command = optic and $cmplx = c such that $command$cmplx = opticc. [1]http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg12330.html On 7/9/2017 9:15 AM, Dr. K. C. Bhamu wrote: A correction on "opticc": On the page number 176 of UG it is mentioned that: "In systems without inversion symmetry, the complex version opticc must be executed." On the page number 177 under the section 8.17.1, we see run optic: x opticc -so [-up]. So what have followed by Chami was okay. But, in many threads on the mailing list it is mentioned that for the complex system we do not need to add "-c: and recent versions of the Wien2k will take care of it. Also, when we run init_lapw for a simple case and a complex case, we see for "x dstart" that for simple systems wien2k consider only "x dstart" and for complex cases it includes "x dstart -c". Could you please tell me what is the difference between "opticc" and "optic -c". Do we really need to add "-c" with optic or the latest version of Wien2k will automatically take care about "-c" switch? Regards Bhamu Dr. K. C. Bhamu (UGC-Dr. D. S. Kothari Postdoc Fellow) Department of Physics Goa University, Goa-403 206 India Mob. No. +91-9975238952 On Sun, Jul 9, 2017 at 8:24 PM, Gavin Abo wrote: But now when "x opticc -so -orb -up " it keep crashing. I do not have case.vectorsoup file really. How I create it.? ERROR: 'OPTIC' - can't open unit: 10 'OPTIC' - filename: /scratch/10820461.yak.local/ SCFsp-U-SO.vectorsoup 'OPTIC' - status: OLD form: UNFORMATTED In "Table 4.3: Input and output files of main programs in an SCF cycle" on page 36 of the WIEN2k 17.1 usersguide [1] under the generates column for LAPWSO, you should see "case.vectorso" for the non-spin polarized case. From that, it could be inferred that "case.vectorsoup" is expected to be created by it for the spin-polarized case. So, it should have been created when you did "x lapwso -up". You might check SCRATCH in your .bashrc: username@computername:~/ wiendata/test$ grep "export SCRATCH" ~/.bashrc export SCRATCH=./ <- This is likely set to "/scratch/10820461.yak.local/" in your system instead of "./". Sometimes the SCRATCH may need to
Re: [Wien] Charge convergence
Dear Prof. Marks, Blaha and Dobysheva, First of all, thank you very much for your comments. *Dear Prof. Marks:* 1) The warning message is related to WARNING: RKmax reduced due to NMATMAX but I am aware of this. No reasons to worry about. 2) My system is a molecule (as pointed by Prof. Blaha) with 60 C atoms, 5 N atoms and 35 H atoms. The cell has vacuum along two of the three directions. I believe that the large changes in atomic positions are related to rotations of the H bonds (is there a way to check it ?). It may explain the "significant" changes in :DIS without disturbing the energy. Do you believe that this can also explain the small GREED value ? MSR1a certainly found soft modes. *Dear Prof. Blaha:* 1) I thought that once the -cc and -ec criteria are fulfilled, the optimization process stops changing the structure. Now I know that the stop criterium is more complex than this (obviously, the forces have to be taken into account in a minimization process) *Dear Prof. Dobysheva:* 1) Thank you for your clarification. It will be useful. All the best, Luis 2017-07-12 7:35 GMT-03:00 Lyudmila Dobysheva: > 11.07.2017 23:31, Luis Ogando wrote: > >> :DIS : CHARGE DISTANCE ( 0.0014560 for atom 33 spin 1) >> :DIS : CHARGE DISTANCE ( 0.0014130 for atom 33 spin 1) >> >> On Tue, Jul 11, 2017 at 9:40 AM, Luis Ogando wrote: >> >>> The energy is finally converged, but the charge is not despite the > fact that its change is below the limit in the last 10 iterations. > ... > >> > CarbazoleLDAopt.dayfile::CHARGE convergence: 0 0.001 .0004560 > CarbazoleLDAopt.dayfile::CHARGE convergence: 0 0.001 .0004130 > If I am not wrong, the convergence is achieved after 3 consecutive > iterations with changes below the limit, but in my case this is > happening for more than 10 iterations without convergence. > A very small addition: dayfile prints here not :DIS, but :DIS minus > 0.001 > So, these lines show that the criterium ":DIS smaller than 0.001" is not > fulfilled yet, the convergence is achieved when negative values appear here > three times. > > Best wishes > Lyudmila Dobysheva > -- > Phys.-Techn. Institute of Ural Br. of Russian Ac. of Sci. > 426001 Izhevsk, ul.Kirova 132 > RUSSIA > -- > Tel.:7(3412) 432459(office), 722529(Fax) > E-mail: l...@ftiudm.ru, lyuk...@mail.ru (office) > lyuk...@gmail.com (home) > Skype: lyuka17 (home), lyuka18 (office) > http://ftiudm.ru/content/view/25/103/lang,english/ > -- > ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] A confirmation is needed about mbj+so
As I said in my previous email, you can not do minimization with SO (except if SO is switched off on atoms which are moving in the unit cell). So, do not use "-so" with "min". On Wednesday 2017-07-12 15:22, fatima DFT wrote: Date: Wed, 12 Jul 2017 15:22:33 From: fatima DFTReply-To: A Mailing list for WIEN2k users To: A Mailing list for WIEN2k users Subject: Re: [Wien] A confirmation is needed about mbj+so Thank you Sir Dr. Tran for reply, I still have some doubts. started min -j "run_lapw -ec xxx -cc xxx -fc x -i x" My first question: Is the above process is okay, can we do that? >From above method, my band gap is in the reasonable agreement with Experimental band gap (still optimization is going on). Why are you executing initso_lapw since the option -so is not used? I forgot to mention -so with above command. The best is to do first a geometry optimization with a GGA like PBE or PBEsol and without SO. Then, at this optimized geometry (or at the experimental one if available) do the calculations that you want (GGA+SO, mBJ, mBJ+SO). Okay, so my question is still need one more statment: I have done geometry optimization with PBE then applied mBJ. Then I have completed -SO calculation in the same directrity where I have done mBJ. SO with min -j "run_lapw -so" gave me accurate band gap while -SO without "min -j" switch gave me overestimated band gap. So my test says: SO with min -j "run_lapw -so *" on a [PBE(opt geometry) + mBJ] calculation is the right choice and this is mBJ+SO. In this case we can do geometry optimization by fixing positions of some atoms. In the second way where we do a geometry optimization with PBE/LDA and then do mBJ+SO following below process: save_lapw initso_lapw init_mbj_lapw run_lapw -so -i 1 -NI (we may avoide -so here) init_mbj_lapw 0/1/2/3 min -j "run_lapw -so -i 1" Insted of this step we need to do only run_lapw -so .. as mBJ ans -SO are about run in a single scf step and we can not apply -fc on mBJ. Sincerely Ms. Fatima . ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] A confirmation is needed about mbj+so
Thank you Sir Dr. Tran for reply, I still have some doubts. > started min -j "run_lapw -ec xxx -cc xxx -fc x -i x" >> >> My first question: >> Is the above process is okay, can we do that? >> From above method, my band gap is in the reasonable agreement with >> Experimental band gap (still optimization is going on). >> > > Why are you executing initso_lapw since the option -so is not used? I forgot to mention -so with above command. The best is to do first a geometry optimization with a GGA like > PBE or PBEsol and without SO. Then, at this optimized geometry > (or at the experimental one if available) do the calculations that you > want (GGA+SO, mBJ, mBJ+SO). > Okay, so my question is still need one more statment: I have done geometry optimization with PBE then applied mBJ. Then I have completed -SO calculation in the same directrity where I have done mBJ. SO with min -j "run_lapw -so" gave me accurate band gap while -SO without "min -j" switch gave me overestimated band gap. So my test says: SO with min -j "run_lapw -so *" on a [PBE(opt geometry) + mBJ] calculation is the right choice and this is mBJ+SO. In this case we can do geometry optimization by fixing positions of some atoms. In the second way where we do a geometry optimization with PBE/LDA and then do mBJ+SO following below process: save_lapw initso_lapw init_mbj_lapw run_lapw -so -i 1 -NI (we may avoide -so here) init_mbj_lapw 0/1/2/3 min -j "run_lapw -so -i 1" Insted of this step we need to do only run_lapw -so .. as mBJ ans -SO are about run in a single scf step and we can not apply -fc on mBJ. Sincerely Ms. Fatima > . > ___ > Wien mailing list > Wien@zeus.theochem.tuwien.ac.at > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien > SEARCH the MAILING-LIST at: http://www.mail-archive.com/ > wien@zeus.theochem.tuwien.ac.at/index.html > > ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] Charge convergence
11.07.2017 23:31, Luis Ogando wrote: :DIS : CHARGE DISTANCE ( 0.0014560 for atom 33 spin 1) :DIS : CHARGE DISTANCE ( 0.0014130 for atom 33 spin 1) >> On Tue, Jul 11, 2017 at 9:40 AM, Luis Ogando wrote: The energy is finally converged, but the charge is not despite the fact that its change is below the limit in the last 10 iterations. ... > CarbazoleLDAopt.dayfile::CHARGE convergence: 0 0.001 .0004560 > CarbazoleLDAopt.dayfile::CHARGE convergence: 0 0.001 .0004130 If I am not wrong, the convergence is achieved after 3 consecutive iterations with changes below the limit, but in my case this is happening for more than 10 iterations without convergence. A very small addition: dayfile prints here not :DIS, but :DIS minus 0.001 So, these lines show that the criterium ":DIS smaller than 0.001" is not fulfilled yet, the convergence is achieved when negative values appear. Best wishes Lyudmila Dobysheva -- Phys.-Techn. Institute of Ural Br. of Russian Ac. of Sci. 426001 Izhevsk, ul.Kirova 132 RUSSIA -- Tel.:7(3412) 432459(office), 722529(Fax) E-mail: l...@ftiudm.ru, lyuk...@mail.ru (office) lyuk...@gmail.com (home) Skype: lyuka17 (home), lyuka18 (office) http://ftiudm.ru/content/view/25/103/lang,english/ -- ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] A confirmation is needed about mbj+so
I just finished some calculation with PBE/PBE+SO/mBJ+SO. I know that with mBJ we can not do force optimization. I know how to do force optimization with -SO Forces with SO are not available in WIEN2k. Thus, optimization (min) and SO can be used together only if SO is switched off on the atoms which require an optimization of their positions. I have done PBE+mBJ calculation save_lapw -d case initso_lapw and fixed position for some atoms and then started min -j "run_lapw -ec xxx -cc xxx -fc x -i x" My first question: Is the above process is okay, can we do that? From above method, my band gap is in the reasonable agreement with Experimental band gap (still optimization is going on). Why are you executing initso_lapw since the option -so is not used? As we know the standard process for the mBJ is: 1.init_mbj_lapw 2. run_lapw -i 1 -NI 3. save_lapw ** 4. init_mbj_lapw 5. 0/1/2/3 6. run_lapw ... So my second question: If we do a general mBJ and then do -SO calculation after save_lapw in mBJ directory then -so will be assumed from the step 6 only. What would be the effect of step 2 "run_lapw -i 1 -NI" (that have been done in the case of mBJ and the -SO assumed from step 6). From above two question, I am not able to find a proper way to follow (mBJ)+(SO) or (mBJ+SO) If your question is "Does it matter to use -so in step 2?", then the answer is no. We can do in two ways if I understood it correctly, first run an mBJ and then do scf with -SO is different then doing mBJ+SO in a single scf step. From one we can relax the structure while from two we can not. The best is to do first a geometry optimization with a GGA like PBE or PBEsol and without SO. Then, at this optimized geometry (or at the experimental one if available) do the calculations that you want (GGA+SO, mBJ, mBJ+SO).___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html