[Wien] QTL

[jan kunes]

Dear Doru,

I have tried to run qtl with the cf1 file you sent me and I do not have 
your problem, i.e. I get identity matrix in the unitarity test. I suggest 
you try to find first where your 60.504660 comes from. It should be fairly 
easy to put couple of extra 'write's to the source code. Have a look at 
the size of the matrix that is manipulated inside the code. 

Regards
Jan


On Mon, 14 Mar 2005, Doru Torumba wrote:

> Dear Jan,
> 
> We have been trying to get a nice output from qtl. We can provide now some
> other details that could help to solve the problem.
> First, if we look at the 14x14 matrix in the .outputq-file we see some very
> large numbers in the imaginary part:
> 
> -case.outputq:
> 
>  IMAG:
> ...........
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000 60.504660
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000
> 0.000000  0.000000-60.504660  0.000000  0.000000  0.000000  0.000000
> 
> Therefore we think there must be something else behind, not only an output
> problem.
> 
> Then we have used as input in the .cf1-file, the matrices provided in the
> Wien2k templates. Using the matrix from the 'case.cf_f_rel'-file we get into
> the same problem, while if we use 'case.cf_f_nonrel' or 'case.cf_f_mm2' we
> get a "normal" output (no '*****' entries).
> 
> I have attached the struct-file we use in our test case (bcc Ho) and the
> matrix we use for the f-splitting.
> 
> Best regards,
> Doru Torumba
> 
> 
> 
> ----- Original Message ----- 
> From: "Stefaan Cottenier" <stefaan.cottenier at fys.kuleuven.ac.be>
> To: <wien at zeus.theochem.tuwien.ac.at>
> Sent: Saturday, March 12, 2005 10:59 AM
> Subject: Re: [Wien] QTL
> 
> 
> >
> > Hello Jan,
> >
> > Thanks for your clarification. It looks like we're close to
> > understanding it now :
> >
> > > I am a bit confused what kind of calculation you want to do. Is it
> > > qtl -so -up? If you use SOC you run qtl only with -up option since both
> > > spin channels are treated at the same time.
> >
> > Yes, that's what we want to do.
> >
> > > Back to your case. Having 6 olumns in the output is o.k. 1-3 refer to up
> > > DOS 4-6 to down DOS. Is there still any problem after this
> clarification?
> >
> > Partly. Although we understand now that having 6 columns is the desired
> > behaviour, there are still '*****' entries in the output from qtl -so -up
> :
> >
> > >>- case.qtl1 -up:
> > >>
> > >> LATTICE CONST.= 5.91350 5.91350 5.91350   FERMI ENERGY=   1.08090
> > >>    0 < NMAT <    0   SPIN=2   NATO  1
> > >> JATOM  1  MULT= 1  ISPLIT= 2  projected density of states
> > >> BAND:   1
> > >>  -2.47596  1 1.00000    0.00000 0.00002 0.00001 0.00000****************
> > >>  -2.47596
> > >>  -2.46503  1 1.00000    0.00000 0.00003 0.00003 0.00000****************
> > >>  -2.46503
> > >>  -2.44952  1 1.00000    0.00001 0.00004 0.00004 0.00000****************
> > >>   ............
> > >>  -2.31443
> > >>  -2.31053  1 1.00000    0.00000 0.00000 0.00000 0.00034 0.00001 0.00035
> > >>  -2.31053
> > >>  -2.36510  1 1.00000    0.00000 0.00000 0.00000 0.00000 0.00010 0.00000
> > >>  -2.36510
> > >>  -2.32488  1 1.00000    0.00000 0.00000 0.00000 0.00018****************
> >
> >
> > This is either a harmless problem with the output format (f-dn DOS very
> > large), or garbage output that does not fit into the default format.
> > Let's hope for the first reason... OK, we will print the numbers with
> > different formatting, and check this also for other cases. We will let
> > you know the result.
> >
> > Thanks,
> > Stefaan
> 

-- 
***********************************************************************
Dr. Jan Kunes                     email: kunes at yammer.ucdavis.edu
Department of Physics             Phone: (530) 752-0627
University of California Davis    Fax:   (530) 752-4717
Davis CA 95616
                       Home page:   http://www.fzu.cz/~kunes
***********************************************************************