[Wien] ghost bands

[Jorissen Kevin]

Dear Sahu,
 
the value of k-max (=rkmax / RMT) is quite different in these two calculations, and it gets rather low (8/4 = 2).  I know that rkmax ,and not kmax, is considered as the crucial number in wien, but kmax=2 seems very small, and I would not be very surprised to get different results for calculations with so different kmax.
 
Some 'general' advice for QTL-B's :
When a calculation produces QTL-B warnings, it is useful to find out to which states the QTL-B is related.  A warning may be printed in case.output2 or case.scf2, or you can check the help-files.
You may be able to find out which kind of states causes the QTL-B (i.e., which atom, which l-value, which energy).  This should be of help in solving the riddle (as pointed out in a recent e-mail by P. Blaha, high QTL-B values may be okay for certain states - see a recent topic 'different questions' started by Gabriela Fernanda Cabeza) / the problem (i.e., the cases where QTL-B really means 'ghostbands').
It may also be interesting to check case.scf1 and see which linearisation energies were chosen in each case.  Also read chapter 12 of the UG, which gives some very good advice.
 
 
Kind regards,
 
Kevin Jorissen.
 

	-----Original Message----- 
	From: Dr. B.R.Sahu [mailto:sahu at matter3.ph.utexas.edu] 
	Sent: Sun 8/3/2003 11:35 PM 
	To: wien at zeus.theochem.tuwien.ac.at 
	Cc: 
	Subject: [Wien] ghost bands
	
	


	Dear W2k users
	
	      I had sent a query regarding ghost bands in supercell calculation
	few days ago. If any user have some suggestions, It would be helpful.If
	you need any more information than provided pl. let me know.
	
	 I have a supercell of K host + one Ti
	 with distance of the Ti from first K shell , d(K-Ti)=8.01 au.  In this
	 run, I had set
	 d(K-Ti)=5.0 au  to begin with (ie RMT=2.5 for K and Ti each) 
	 and the calculation converged without any ghost bands(u-dot
	 contribution smaller than u).
	
	 I wanted to increase the RMT of K and Ti to 4.0 a.u each so that they
	 almost touch each other ie d(K-Ti)=8.0 au. I started the calculation all
	 over from scratch. This time the calculation is faster than the previous
	 run(due to small number of plane waves in the interstitials) but I get the
	 QTL-B value very large ie
	
	 QTL-B VALUE .EQ.   32.30382   !!!!!!
	
	 I understand that such a large value for QTL-B is indeed ghost band
	 signature in the scf calculation.
	
	 The parameters used for both runs are SAME:
	
	 RKMAX= 8.0
	
	 Number of k-points in IBZ = 120( a dense mesh is not needed for such a
	 large cell but I believe it is not one of the reason for appearance of
	 ghost band).
	
	 Unit cell size = about 20 au.
	
	 RMT(Ti)=2.5 and RMT(K)=2.5 in the first case(NO GHOST BANDS).
	
	 RMT(Ti)=4.0 and RMT(K)=4.0 in the second case(GHOST BANDS).
	
	 Is RKMAX=8.0 is large enough to cause ghost bands?
	
	 Here is the case.in1 file:
	
	 _______________________________________________________________________________
	 WFFIL        (WFPRI, SUPWF)
	   8.00       10    4 (R-MT*K-MAX; MAX L IN WF, V-NMT
	   0.30    4  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
	 APW/LAPW)
	  0   -2.29      0.010 CONT 1
	  0    0.30      0.000 CONT 1
	  1   -1.07      0.010 CONT 1
	  1    0.30      0.000 CONT 1
	   0.30    4  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
	 APW/LAPW)
	  0   -2.29      0.010 CONT 1
	  0    0.30      0.000 CONT 1
	  1   -1.07      0.010 CONT 1
	  1    0.30      0.000 CONT 1
	   0.30    4  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
	 APW/LAPW)
	  0   -2.29      0.010 CONT 1
	  0    0.30      0.000 CONT 1
	  1   -1.07      0.010 CONT 1
	  1    0.30      0.000 CONT 1
	   0.30    4  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
	 APW/LAPW)
	  0   -2.29      0.010 CONT 1
	  0    0.30      0.000 CONT 1
	  1   -1.07      0.010 CONT 1
	  1    0.30      0.000 CONT 1
	   0.30    5  0      (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
	 APW/LAPW)
	  0    0.30      0.000 CONT 1
	  0   -4.35      0.005 STOP 1
	  1   -2.58      0.010 CONT 1
	  1    0.30      0.000 CONT 1
	  2    0.30      0.010 CONT 1
	 K-VECTORS FROM UNIT:4   -9.0       2.5      emin/emax window
	 _______________________________________________________
	
	
	 where I am going wrong?pl. suggest.
	
	 best regards
	 sahu
	
	
	_______________________________________________
	Wien mailing list
	Wien at zeus.theochem.tuwien.ac.at
	http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
	

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 8942 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20030804/04d58445/attachment.bin