[Wien] problem with SPAGHETTI

[Peter Blaha]

I don't know, what system you are calculating.
It must be very large ?

It looks the matrix is truncated at 14000 APWs, what is the "effective" 
RKMAX ?
Did your scf-cycle converge properly ? No warnings in :ENE ???

You are looking into case.output1_*  ??? This comes from a k-point 
parallel calculation.    x lapw1 -c -band does NOT produce such a file, 
you would need to add a -p switch. (Then of course also for x spaghetti -p)

It seems you have only very few (negative) eigenvalues. Do you see more 
eigenvalues in case.scf ??? Is this small molecule in a very big box ??

Is NUME very small ?? (siteconfig, dimensions )



katrusiat schrieb:
> Dear Dr. Blaha,
> Thanks for answering!
> I will try x lapw1 -c -band, it will take some time. I am only wandering why x lapw1c -band is not a good choice.
> The eigenvalues lists are ok. The case.output1_* do not look ok, but they too seem to be cut at the end like:
>        ********************************************************
>  NVEC2 too small.          25         -25   3.48482196426312   3.05670523153738 Matrix size        13999 Time for al,bl:    1.29000000000087 nlo         997 Time for los:    19.8900000000031 Time for LOOP 260:         2466.77000000000     HNS: Loop 140:      0.1 Loop 90:      0.4 matrix update:     15.3loop 30:     0.0(wall time)        1000  Eigenvalues computed Seclr4(Cholesky complete (CPU)) :             918.880     3980.806 Mflops Seclr4(Transform to eig.problem (CPU)) :     2734.360     4013.242 Mflops Seclr4(Compute eigenvalues (CPU)) :          8136.700     1798.214 Mflops Seclr4(Backtransform (CPU)) :                 167.330     1171.171 Mflops       TIME HAMILT (CPU)  =  2488.0, HNS =  1302.3, DIAG = 11960.9
>      K=   0.01724   0.00000   0.00000      MATRIX SIZE13999  WEIGHT= 1.00  PGR:     EIGENVALUES ARE:     -6.6025254   -6.6024794   -6.6024610   -6.6024316   -6.6020797     -6.6020404   -6.6019836   -6.6019380   -6.6019040   -6.6018471     -6.6018031   -6.6017642   -6.6017390   -6.6017036   -6.6016403     -6.6016001   -6.5911194   -6.5910823   -6.5910553   -6.5909545     -6.5908527   -6.5907548   -6.5906808   -6.5906041   -6.5746481     -6.5742373   -6.5719077   -6.5690439   -6.5657785   -6.5628050     -6.5627753   -6.5610205   -6.5559183   -6.5556711   -6.5554220     -6.5546469   -6.5545580   -6.5514129   -6.5512526   -6.5455203
>      -6.5454331   -6.5412339   -6.5401636   -6.5399373   -6.5366359     -6.5178022   -6.5151775   -6.5074289   -4.9987553   -4.9984120     -4.9978648   -4.9977797   -4.9977128   -4.9976294   -4.9976040     -4.9975722   -4.9975004   -4.9974661                           <============== end of .output1_1
> 
> with no error message.
> The Fermi energy was inserted into .insp. And it looks like that unit -5 is .insp since in spaghetti.def there is no "-5" file:5, 'slab.insp',         'old',    'formatted',06, 'slab.outputsp',     'unknown','formatted',09, 'slab.qtl',          'unknown','formatted',010,'slab.spaghetti_ene','unknown','formatted',011,'slab.spaghetti_ps', 'unknown','formatted',020,'slab.struct',       'old',    'formatted',030,'slab.irrep',          'unknown','formatted',040,'slab.bands.agr',       'unknown','formatted',07, 'slab.output1',  'old','formatted',0
> I also must say that, strangely enough, both x lapw1 -band and x lapw1c -p -band can initiate the eigenvalue calculation.
> Regards,Kateryna Foyevtsova
> 
> 
> 
> -- Жизнь без спама на Яндекс.Почте http://mail.yandex.ru/nospam _______________________________________________Wien mailing listWien at zeus.theochem.tuwien.ac.athttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 
-----------------------------------------
Peter Blaha
Inst. Materials Chemistry, TU Vienna
Getreidemarkt 9, A-1060 Vienna, Austria
Tel: +43-1-5880115671
Fax: +43-1-5880115698
email: pblaha at theochem.tuwien.ac.at
-----------------------------------------