[Pw_forum] ld1 & search of ghost states

[Latévi Max LAWSON DAKU]

Dear Q.-E. users,

I have a question about the use of the ld1 code and the identification
of ghost states. I am trying to generate a norm-conserving TM PP for
iron (xc=blyp), with the 3s and 3p semicores states treated as valence
states. The reference configuration used is: '[Ne] 3s2 3p6 3d6 4s0 4p0'.
I generated one with the fhi98pp package, but I would like to also do it 
with the LD1 code, using the blyp and also other more recent functionals.

According to my understanding of the LD1 manual (INPUT_LD1.html),
testing the generated PP with a basis set of Bessel functions allow the
identification of ghost states. In order to see how this shows up, I used
too large cutoff radii ('rcut') for generating the PP (pseudotype=1). This
one was stored in a UPF file.

*) I used lloc=0 and, during the test, the following warning was issued
     >>>
     Computing logarithmic derivative in   2.19388

       WARNING! Expected number of nodes:   0=     2-  1-  1, number of 
nodes found:  1.
       Setting wfc to zero for this iteration.
       (This warning will only be printed once per wavefunction)

     Computing logarithmic derivative in   2.19388
     Computing the partial wave expansion
       no projector for channel:   0
     <<<
I therefore guessed that a ghost state at least appears in the L=1
channel (according to my understanding of "ascheqps.f90").

*) The calculation of the logarithm was followed by a test of the PP:
     >>>
     
Fe                                                                        
     scalar relativistic calculation

     atomic number is 26.00   valence charge is 16.00
     dft =BLYP   lsd =0 sic =0 latt =0  beta=0.20 tr2=1.0E-14
     mesh =1191 r(mesh) = 101.18024 xmin = -7.00 dx = 0.01250

     n l     nl             e AE (Ry)        e PS (Ry)    De AE-PS (Ry)
     1 0     3S   1( 2.00)       -8.25458       -8.25457        0.00000
     2 1     3P   1( 6.00)       -5.76444       -5.76444        0.00000
     3 2     3D   1( 6.00)       -1.89183       -1.89183        0.00000
     2 0     4S   1( 0.00)       -1.50529       -1.74603        0.24073
     3 1     4P   1( 0.00)       -1.07194       -1.13184        0.05990

     eps = 7.2E-15  iter = 23
 
     Etot =   -2543.705260 Ry,   -1271.852630 Ha,  -34608.870131 eV
     Etotps =  -237.821799 Ry,    -118.910899 Ha,   -3235.730128 eV

     ...
     <<<

*) Then, there is the test with the basis set of Bessel functions. But
no warning or error message was printed during this test:
     >>>
     Box size (a.u.) :   30.0

     Cutoff (Ry) :   50.0
                           N = 1       N = 2       N = 3
     E(L=0) =        -8.2690 Ry   -1.7535 Ry   -0.6442 Ry
     E(L=1) =        -5.8198 Ry   -1.1415 Ry   -0.4622 Ry
     E(L=2) =        -1.9392 Ry   -0.5486 Ry   -0.2724 Ry

     Cutoff (Ry) :   55.0
                           N = 1       N = 2       N = 3
     E(L=0) =        -8.2692 Ry   -1.7535 Ry   -0.6442 Ry
     E(L=1) =        -5.8266 Ry   -1.1429 Ry   -0.4626 Ry
     E(L=2) =        -1.9629 Ry   -0.5508 Ry   -0.2732 Ry

     ..etc...
     <<<
By looking at the source (test_bessel.f90), I understood that no such
a warning or error message is to be printed. The reading of the above
eigenvalues does not help me tell whether there is a ghost state or not:
I'm clearly missing something !! Probably something very obvious

*) My guess is that I actually should use the semilocal form of the PP to
compute the eigenvalue spectrum and compare this spectrum with
the above one. Is this correct ? If so, how should I proceed ?

I thanks you in advance for your advice and/or help.


Best regards,
Max





-- 
***********************************************
Latevi Max LAWSON DAKU
Departement de chimie physique 
Universite de Geneve - Sciences II
30, quai Ernest-Ansermet  
CH-1211 Geneve 4 
Switzerland 

Tel: (41) 22/379 6548 ++ Fax: (41) 22/379 6103
***********************************************