Re: [Wien] updated email for -SP+SO case
For my comments, see below. I got the point but I still have some doubts. My system is half metallic. For -up it is metallic while for -dn channel it is a semiconductor. So in case of semiconductor, the value of tetra in case.in2/c could be same (TETRA 0.000 (GAUSS,ROOT,TEMP,TETRA,ALL eval) while for a metallic case, it should be TETRA 101.0 (GAUSS,ROOT,TEMP,TETRA,ALL eval). This is for -sp without -so as you mention below, right? If so, that is probably okay. Or to be "safe", use the 101 [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg14472.html ]. steps followd: runsp_lapw -p -ec 0.0001 -so -cc 0.0001 -i 300 -NI x kgen -so : increase k-points. Query: 1: For optical properties (assuming that for -up state it is metallic and for other it is semiconductor): # x_lapw lapw1 -p -up (with TETRA 101.0) x_lapw lapw1 -p -dn (with TETRA 0.000) x_lapw lapwso -up -p (with TETRA 101.0) x_lapw lapw2 -p -fermi -up -so (with TETRA 101.0) x_lapw lapw2 -p -fermi -dn -so (with TETRA 0.000) x_lapw optic -so -up -p (with TETRA 101.0) x_lapw optic -so -dn -p (with TETRA 0.000) x_lapw joint -up -p (with TETRA 101.0) x_lapw joint -dn -p (with TETRA 0.000) addjoint_updn_lapw x_lapw kram Is the above steps are right? I am in doubt as there is not lapwso -dn used. No, see next comment below. I got confused from the thread provided Prof. Peter. https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16552.html where he suggested to follow: runsp -so # spin-polarized + SO x optic -up -so x joint -up x kram -up (n*o addjoint-updn*) provided that I use "joint.f" provided in above thread. As he also did not mention lapw1, lapwso, lapw2, etc. Please correct me in above steps. Yes, I would use that provided "joint.f" with WIEN2k 17.1 and utilize those steps in that thread. Select OPTIC from the menu in w2web, there you should see that the kgen, lapw1, and lapw2 steps are "optional" unless you are changing the k-mesh. For changing the k-mesh, it looks like that would need [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg12365.html ]: x kgen -so x lapw1 -up x lapwso -up For brevity, it seems that the lapw2 step was omitted in that post. So you could add it: x lapw2 -fermi -up -so Next query is for band structure: Query: 2 For bands I followed below steps to calculate the dispersion curve: x_lapw lapw1 -band -p -up x_lapw lapw1 -band -p -dn x_lapw lapwso -p -up >>> I did not use lapwso -dn as it is not supported. You don't use lapwso -dn, because lapwso mixes the spin up and dn together [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg03239.html , https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg00607.html ]. The -up switch on lapwso just tells the program it needs to run for a spin-polarized case [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg02807.html ]. x_lapw lapw2 -so -band -qtl -p -up x_lapw lapw2 -so -band -qtl -p -dn x_lapw spaghetti -so -p -up x_lapw spaghetti -so -p -dn (results are same when I use -c with lapw1, lapw2 and lapwso; tested as it used case.in2c file) If you using the x script with a program such as lapw2 like above (i.e., the "x program" lines seen above), it is recommended not use the "-c" flag. This is because WIEN2k will automatically detect what it should be and adds it internally if it is needed [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16066.html ]. It is able to this by checking if the case directory has the case.in1 and case.in2 files (non-complex) or a case.in1c and case.in2c files (complex). So be careful not to create those files by hand yourself (unless you know what your doing), you should let the init_lapw script (or possibly other scripts) do that. The system is PbCrO3. For -sp (without -so) case it is semiconductor for -up state while metallic for -dn state. In case of -sp+-so (with above steps), it is metallic for both stated (-up and -dn). Where have I mistaken? I'm not an expert on half metallics with optic, so I could be wrong. However, it looks like you answered your own question. For the -sp without -so, it looks like you can do the -up state for the semiconductor and -dn state for the metallic. This you could do for the Imag(epsilon) as it can be plotted separately, but the Real(epsilon) would need to be combined [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16536.html ]. For the -sp with -so, the spin orbit couples the spin up and dn together. So it seems that you can only get a single mixed metallic-semiconductor case as you have mentioned above. ___ 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.c
Re: [Wien] updated email for -SP+SO case
Thank you very much Gavin for more clarification. Now I have few more queries, please see : For the -sp with -so, the spin orbit couples the spin up and dn together. So it seems that you can only get a single mixed metallic-semiconductor case as you have mentioned above. 1. As the system is halfmetal, should I follow the steps to be done for metals (calculations of plasma frequency)? I am not sure for this as we are changing TETRA to 101.0. 2. Need to recompile optic package? I am 100 % sure just for confirmation. 3. As mnetioned in one of the thread of prof. Prof. peter (and your above statment): https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg00607.html " X spaghetti -so -up or -dn gives the "same" bandstructure, only the "fat-band" representation will be different, since up and dn-spin projections are different. So how we correlate density of states with band structure as we are able to get DOS for -up and -dn ? Kind regards Bhamu ___ 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/wi e...@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
[Wien] why we need to divide plasma frequency by root(2)?
This is in continuation to my previous email about -sp+-so (final answer on that email is still needed). meanwhile, I have calculated plasma frequency for a metal case and I got below (case is -sp+-so): Plasma frequencies: !!! WARNING: w_pl = sqrt( w_pl^2(up-spin) + w_pl^2(dn-spin) )!!! w_p_xx w_p_yy w_p_zz[eV] 2.0740 2.0740 1.2640 So the format of my case.inkram should be -case.injoint xx xx xx 2.0740 2.0740 1.2640 xx --- were xx means no change in lines of case.inkram. At page numer 164 under section "8.11 KRAM (Kramers-Kronig transformation) in UG" *"For a metal, the Plasma-frequencies (intraband transitions) for up and dn should be added, but then divided by√2, before using x kram." * Does it mean that we need to add all three components and then divide by root(2) and put a single number [(2.074+2.074+1.264)/root(2)=3.88] instead of all three in case.injoint? kind regards Bhamu ___ 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] updated email for -SP+SO case
1. Another user had to calculate plasma frequency for a half metallic [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg07695.html ]. So it is likely that you need to do it to. 2. Yes, you must recompile the optic package after replacing joint.f because it is a Fortran source code file and not a script, . 3. Sorry, I don't know. Instead of plotting the spin up/dn band structure next to the DOS https://www.researchgate.net/figure/257076112_fig8_FIG-8-Total-DOS-and-Band-Structure-of-Co-2-TiSb-using-the-LSDAU it may be that you need to plot the fat band structure and PDOS (partial DOS) next to each other. I don't know of WIEN2k program or script to do that. So you would likely have to plot it by hand in software like Origin [ http://www.originlab.com/ ]. Or if you're good a programming, maybe you could make a script similar to what someone had did for VASP: http://gvallver.perso.univ-pau.fr/?p=587 Spaghetti-primavera on the unsupported page [ http://susi.theochem.tuwien.ac.at/reg_user/unsupported/ ] might be helpful for creating the fat band (or band-character) plot. COHP or COOP might also help with this: http://schmeling.ac.rwth-aachen.de/cohp/index.php?menuID=2 Unfortunately, that is not currently implement for WIEN2k: https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16542.html https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16543.html On 10/21/2017 3:06 PM, Dr. K. C. Bhamu wrote: Thank you very much Gavin for more clarification. Now I have few more queries, please see : For the -sp with -so, the spin orbit couples the spin up and dn together. So it seems that you can only get a single mixed metallic-semiconductor case as you have mentioned above. 1. As the system is halfmetal, should I follow the steps to be done for metals (calculations of plasma frequency)? I am not sure for this as we are changing TETRA to 101.0. 2. Need to recompile optic package? I am 100 % sure just for confirmation. 3. As mnetioned in one of the thread of prof. Prof. peter (and your above statment): https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg00607.html " X spaghetti -so -up or -dn gives the "same" bandstructure, only the "fat-band" representation will be different, since up and dn-spin projections are different. So how we correlate density of states with band structure as we are able to get DOS for -up and -dn ? Kind regards Bhamu ___ 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] why we need to divide plasma frequency by root(2)?
> > > -case.injoint > xx > xx > 1 >> correction > 2.0740 2.0740 1.2640 > xx > --- > > were xx means no change in lines of case.inkram. > > > > At page numer 164 under section "8.11 KRAM (Kramers-Kronig > transformation) in UG" > > > *"For a metal, the Plasma-frequencies (intraband transitions) for up and > dn should be added, but then divided by√2, before using x kram." * > > Does it mean that we need to add all three components and then divide by > root(2) and put a single number [(2.074+2.074+1.264)/root(2)=3.88] > instead of all three in case.injoint? > > > kind regards > > Bhamu > > ___ 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] updated email for -SP+SO case
Thank you very much Gavin for detailed help. It is very helpful for us. Bhamu On Sun, Oct 22, 2017 at 3:59 AM, Gavin Abo wrote: > 1. Another user had to calculate plasma frequency for a half metallic [ > https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg07695.html > ]. So it is likely that you need to do it to. > > 2. Yes, you must recompile the optic package after replacing joint.f > because it is a Fortran source code file and not a script, . > > 3. Sorry, I don't know. > > Instead of plotting the spin up/dn band structure next to the DOS > > https://www.researchgate.net/figure/257076112_fig8_FIG-8- > Total-DOS-and-Band-Structure-of-Co-2-TiSb-using-the-LSDAU > > it may be that you need to plot the fat band structure and PDOS (partial > DOS) next to each other. > > I don't know of WIEN2k program or script to do that. So you would likely > have to plot it by hand in software like Origin [ > http://www.originlab.com/ ]. Or if you're good a programming, maybe you > could make a script similar to what someone had did for VASP: > > http://gvallver.perso.univ-pau.fr/?p=587 > > Spaghetti-primavera on the unsupported page [ http://susi.theochem.tuwien. > ac.at/reg_user/unsupported/ ] might be helpful for creating the fat band > (or band-character) plot. > > COHP or COOP might also help with this: > > http://schmeling.ac.rwth-aachen.de/cohp/index.php?menuID=2 > > Unfortunately, that is not currently implement for WIEN2k: > > https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16542.html > > https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16543.html > > On 10/21/2017 3:06 PM, Dr. K. C. Bhamu wrote: > > Thank you very much Gavin for more clarification. > > Now I have few more queries, please see : > > > For the -sp with -so, the spin orbit couples the spin up and dn together. > So it seems that you can only get a single mixed metallic-semiconductor > case as you have mentioned above. > > 1. As the system is halfmetal, should I follow the steps to be done for > metals (calculations of plasma frequency)? I am not sure for this as we are > changing TETRA to 101.0. > 2. Need to recompile optic package? I am 100 % sure just for confirmation. > > 3. As mnetioned in one of the thread of prof. Prof. peter (and your above > statment): > https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg00607.html > > > " X spaghetti -so -up or -dn gives the "same" bandstructure, only the > "fat-band" representation will be different, since up and dn-spin > projections are different. > > So how we correlate density of states with band structure as we are able > to get DOS for -up and -dn ? > > > > Kind regards > > Bhamu > > > > ___ > 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] why we need to divide plasma frequency by root(2)?
I could be wrong, but I think you just need to take the sqrt(2) of each plasma frequency: 2.074/sqrt(2) 2.074/sqrt(2) 1.264/sqrt(2) then put them in case.inkram (not case.injoint). You should be able to use multiple plasma frequencies in case.inkram but don't forget to add a Gamma for drude value for each of the plasma frequencies [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg14495.html ]. On 10/21/2017 4:29 PM, Dr. K. C. Bhamu wrote: -case.injoint xx xx 1 >> correction 2.0740 2.0740 1.2640 xx --- were xx means no change in lines of case.inkram. At page numer 164 under section "8.11 KRAM (Kramers-Kronig transformation) in UG" *"For a metal, the Plasma-frequencies (intraband transitions) for up and dn should be added, but then divided by√2, before using x kram." * Does it mean that we need to add all three components and then divide by root(2) and put a single number [(2.074+2.074+1.264)/root(2)=3.88] instead of all three in case.injoint? kind regards Bhamu ___ 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] why we need to divide plasma frequency by root(2)?
Thank you Gavin for a quick reply. I will follow all of your advice and will report if I face any difficulty. Bhamu On Sun, Oct 22, 2017 at 4:12 AM, Gavin Abo wrote: > I could be wrong, but I think you just need to take the sqrt(2) of each > plasma frequency: > > 2.074/sqrt(2) 2.074/sqrt(2) 1.264/sqrt(2) > > then put them in case.inkram (not case.injoint). > You should be able to use multiple plasma frequencies in case.inkram but > don't forget to add a Gamma for drude value for each of the plasma > frequencies [ https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac. > at/msg14495.html ]. > > On 10/21/2017 4:29 PM, Dr. K. C. Bhamu wrote: > > >> -case.injoint >> xx >> xx >> 1 >> correction >> 2.0740 2.0740 1.2640 >> xx >> --- >> >> were xx means no change in lines of case.inkram. >> >> >> >> At page numer 164 under section "8.11 KRAM (Kramers-Kronig >> transformation) in UG" >> >> >> *"For a metal, the Plasma-frequencies (intraband transitions) for up and >> dn should be added, but then divided by√2, before using x kram." * >> >> Does it mean that we need to add all three components and then divide by >> root(2) and put a single number [(2.074+2.074+1.264)/root(2)=3.88] >> instead of all three in case.injoint? >> >> >> kind regards >> >> Bhamu >> > > ___ > 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