[Pw_forum] calculation = "nscf" with CP

[Young-Su Lee]

Dear CP developers,

I've been using "calculation = nscf" option with CP version 2.0.
I found that this option is disabled in Espresso version 3.0 and managed 
to turn it on.
If I understand correctly, the only necessary change to make "nscf" 
option work is to read charge density in subroutine rhoofr.

What worries me is that the results from two versions are different.
The occupied 4 states are fine and eigenvalues are the same as those 
from self-consistent calculations in both cases.

Ver 2.0 ================================================

                total energy =       -4.06675 a.u.
              kinetic energy =        9.34798 a.u.
        electrostatic energy =       -7.34356 a.u.
                         esr =        0.29882 a.u.
                       eself =        7.97885 a.u.
      pseudopotential energy =       -3.09480 a.u.
  n-l pseudopotential energy =        0.19565 a.u.
 exchange-correlation energy =       -3.17201 a.u.
           average potential =       -0.02336 a.u.
                                                                                                                                    

                                                                                                                                    

                                                                                                                                    

                                                                                                                                    

 eigenvalues at k-point:  0.000 0.000 0.000
  -16.96   -9.38   -9.38   -9.38   -0.55    0.83    0.88    0.89    
0.89    1.04
    1.06    1.81    1.97    2.09    2.21    2.23    2.24    2.34    
2.38    2.44

Ver 3.0 =================================================

                total energy =       -4.07598 Hartree a.u.
              kinetic energy =        9.33894 Hartree a.u.
        electrostatic energy =       -7.34357 Hartree a.u.
                         esr =        0.29882 Hartree a.u.
                       eself =        7.97885 Hartree a.u.
      pseudopotential energy =       -3.09481 Hartree a.u.
  n-l pseudopotential energy =        0.19548 Hartree a.u.
 exchange-correlation energy =       -3.17202 Hartree a.u.
           average potential =        0.00000 Hartree a.u.
                                                                                                                                    

                                                                                                                                    

                                                                                                                                    

   Eigenvalues (eV), kp =   1 , spin =  1
                                                                                                                                    

  -16.96   -9.38   -9.38   -9.38   -0.55    0.80    0.88    0.89    
0.89    1.03
    1.05    1.81    1.97    2.09    2.18    2.21    2.22    2.31    
2.35    2.42

========================================================



Is there any significant change from 2.0 to 3.0 ?
I'd really appreciate if anyone can give me some advice about which 
subroutines I should look at.

Young-Su Lee


P.S.* Two minor bugs
1. subroutine read_rho_xml in Modules/xml_io_base.f90

The following two lines should go before IF( kowner(k) == mpime ) THEN 
statement

         IF ( kowner( k ) /= ionode_id ) &
            CALL mp_put( rho_plane, rho_plane, mpime, ionode_id, 
kowner(k), k )

2. subroutine vofrho in CPV/cplib.f90

The variable "vave" should be from module energies.
It's locally defined now and average potential is always zero in output.