[Wien] Bandstructure and LDA +U, Strange Error in Spagetti

Peter Blaha pblaha at zeus.theochem.tuwien.ac.at
Sat Apr 2 20:25:07 CEST 2005


Just yesterday I plotted a bandstructure including LDA+U! No problems.

spaghetti reads case.output1/up/dn/so (depending on the options) and also
tries to read case.irrep/up/dn. If the latter is empty, it quits this part 
and continues without irrep (and must not come into comprel).

spaghetti does not take -orb (or ignores it), it just reads output1up (I 
suppose an LDA+U calculation is spinpolarized). It looks for "K-vector" 
und reads it and the corresponding eigenvalues.

Maybe your ene-file is some leftover ? Check your output1up file and 
compare this eventually with _ene, but also check the messages in 
outputspup nd compare number of bands and k-vectors and "lines",....
 
> Hi Peter,
> the insp was not the problem, unfortunately.
> 
> Indeed, if I do a regular calculation for some fcc metal like Al the .irrep 
> files with zero length do not cause any error with comprel and spaghetti 
> delivers a nice bandstructure plot.
> It appears only with LDA+U (but I did not check for other cases -so or -so 
> -orb up to now.)
> 
> If checking the _ene files one sees that the length of the k vectors is much 
> too long just like going several times through the brillouin zone.
> For a particular band it looks like:
>   bandindex:          20
>    0.00000   0.00000   0.00000   0.00000  -4.01783
>    0.58952   0.54236  -0.54236   1.83795  -4.01925
> here is the supposed end of the k vector (from the calculation whithout U), 
> but it continues
>    0.09432   0.00000   0.00000   2.24639  -3.99883
> .... etc..
>    0.56594   0.28297   0.00000  35.47796  -4.40326
>   bandindex:          21
>    0.00000   0.00000   0.00000   0.00000  -3.06008
> ....
> 
> I did not check yet if it is just a repetition or cyclic or whatever else, in 
> the first or second case I can shift it back just by myself using a short 
> program.
> 
> Did anyone else experience similar problems ?
> 
> Have a nice weekend
> Gerhard
> 
> Am Freitag, 1. April 2005 14:07 schrieb Peter Blaha:
> > Yes I know that empty irrep* files are created, but they do not matter. 
> > Only non-zero irrep files which correspond to another calculation cause 
> > problems.
> > 
> > Thus your plotting problems may have its origin also in some wrong insp 
> > file. I'd recommend to copy this file again from SRC_templates and redo 
> > the spaghetties.
> > 
> > > that was at least one of the problems, unfortunately it creates directly 
> > > another trouble.
> > > 
> > > It seems that running spaghetti creates new case.irrepup case.irrepdn 
> files 
> > > that are empty.
> > > It seems that spaghetti tries to open the files and if absent it creates 
> empty 
> > > ones.
> > > If one tries accidently (just forgotten to switch frum up to dn) or 
> > > intentionally to run spaghetti a second time it will find the empty 
> > > case.irrepxx file and the error occurs again.
> > > 
> > > Running spaghetti without irrep will have the result that the plot is 
> empty.
> > > However, the created postscript file is not very instructive, but at least 
> > > some part of the data in the created _ene files looks to be ok if using a 
> > > different plot programm. Some parts seem just to be repeated most probably 
> > > due to the missing irrep, or something missing that is usually done by 
> irrep.
> > > 
> > > Anyway, at least we no know where our gap is in the bands.
> > > 
> > > Best wishes from Mainz
> > > Gerhard
> > > 
> > > 
> > > Am Freitag, 1. April 2005 09:20 schrieb Peter Blaha:
> > > > This error comes because you still have a case.irrep* file from a 
> previous 
> > > > run. Please remove any "old" case.irrep* file before running spaghetti.
> > > > 
> > > > (I guess I should change the x script and add the irrep file to 
> > > > spaghetti.def only on demand (-irrep))
> > > > 
> > > > Regards
> > > > 
> > > > > I was running a LDA+U calculation, but was not able to create the 
> > > > > bandstructure. I used to run lapw1 -band -up (-dn) -orb from the 
> single 
> > > prog.
> > > > > 
> > > > > spagetti stopped with the message:
> > > > > 
> > > > > number of k-points read in case.vector=         455
> > > > > ERROR in comprel: cannot find com.rel
> > > > > 0.736u 0.559s 0:01.30 98.4%     0+0k 0+0io 0pf+0w
> > > > > 
> > > > > Similar error message occurs with older Versions, too.
> > > > > 
> > > > > This message seems to be a a little misleading suggesting a file was 
> > > missing, 
> > > > > as comprel does not need a file com.rel but did not find the ncm at 
> the 
> > > given  
> > > > > ig12.
> > > > > 
> > > > > The problem seems to be that comprel doesn't find some symmetry 
> element.
> > > > > 
> > > > > The values checked in comprel are:
> > > > > comprel: cannot find com.rel for           13  to            7
> > > > >                 ig12 =     1013007  
> > > > >                 ncm=           0           0           0           0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0           0           
> 0
> > > > >            0           0           0           0
> > > > > 
> > > > > Indeed there is not such an ncm combination, if it makes sense at all.
> > > > > 
> > > > > The case.spaghettiup_ene and case.spaghettidn_ene files are created, 
> > > however, 
> > > > > inspecting them shows that the bandstructure does not match the DOS. 
> (The 
> > > > > latter shows a clear gap being absent in the bands)
> > > > > 
> > > > > Any idea what is going wrong
> > > > > 
> > > > > Ciao Gerhard
> > > > > 
> > > > > _______________________________________________
> > > > > Wien mailing list
> > > > > Wien at zeus.theochem.tuwien.ac.at
> > > > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > > > > 
> > > > 
> > > > 
> > > >                                       P.Blaha
> > > > 
> --------------------------------------------------------------------------
> > > > Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
> > > > Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
> > > > Email: blaha at theochem.tuwien.ac.at    WWW: 
> > > http://info.tuwien.ac.at/theochem/
> > > > 
> --------------------------------------------------------------------------
> > > > 
> > > > _______________________________________________
> > > > Wien mailing list
> > > > Wien at zeus.theochem.tuwien.ac.at
> > > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > > > 
> > > 
> > > _______________________________________________
> > > Wien mailing list
> > > Wien at zeus.theochem.tuwien.ac.at
> > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > > 
> > 
> > 
> >                                       P.Blaha
> > --------------------------------------------------------------------------
> > Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
> > Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
> > Email: blaha at theochem.tuwien.ac.at    WWW: 
> http://info.tuwien.ac.at/theochem/
> > --------------------------------------------------------------------------
> > 
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > 
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------




More information about the Wien mailing list