ABINIT, developper input variables:

• 0 => Use standard Fortran IO routines
• 1 => Use MPI/IO routines
• 2 => Directly use NetCDF routines (this option is not available)
• 3 => Use ETSF_IO routines, creating NetCDF files according to the ETSF specification.

• "Specification of an extensible and portable file format for electronic structure and crystallographic data", X. Gonze, C.-O. Almbladh, A. Cucca, D. Caliste, C. Freysoldt, M. Marques, V. Olevano, Y. Pouillon, M.J. Verstraete, Comput. Mat. Science 43, 1056 (2008)
• "Sharing electronic structure and crystallographic data with ETSF_IO", D. Caliste, Y. Pouillon, M.J. Verstraete, V. Olevano, X. Gonze, Comput. Physics Communications 179, 748 (2008)
• see also http://www.etsf.eu/fileformats.

• 1 => band-per-band algorithm
• n => The minimization is performed using nband/n blocks of n bands.

• 1 => The minimization is performed using nband/npband blocks of npband bands.
• n => The minimization is performed using nband/(npband*n) blocks of npband*n bands.

• 1=> use FFT routines written by S. Goedecker.
• 2=> use machine-dependent FFT algorithm, taken from the vendor library, if it exists and if it has been implemented. The bare fftalg=200 has little chance to be faster than fftalg=112, but it might be tried. Implementing library subroutines with fftalg/=200 has not yet been done. Currently implemented library subroutines (fftalg=200) are:
  • on HP, z3dfft from Veclib;
  • on DEC Alpha, zfft_3d from DXML;
  • on NEC, ZFC3FB from ASL lib;
  • on SGI, zfft3d from complib.sgimath
• 3=> use serial or multi-threaded FFTW fortran routines (http://www.fftw.org). Currently implemented with fftalg=300.
• 4=> use FFT routines written by S. Goedecker, 2002 version, that will be suited for MPI and OpenMP parallelism.

• 0=> only use Complex-to-complex FFT
• 1=> real-to-complex is also allowed (only coded for A==1)

• 0=> no use of zero padding
• 1=> use of zero padding (only coded for A==1 and A==4)
• 2=> use of zero padding, and also combines actual FFT operations (using 2 routines from S. Goedecker) with important pre- and post-processing operations, in order to maximize cache data reuse. This is very efficient for cache architectures. (coded for A==1 and A==4, but A==4 is not yet sufficiently tested)

• idyson=1 : Solve the Dyson equation by direct matrix inversion
• idyson=2 : Solve the Dyson equation as a first-order differential equation with respect to the coupling constant lambda - only implemented for the RPA at the present stage (see header of dyson_de.f for details)
• idyson=3 : Calculate only the diagonal of the interacting susceptibility matrix by self-consistently computing the linear density change in response to a set of perturbations. Only implemented for the RPA at the present stage, and entirely experimental (see dyson_sc.f for details).

• 0 => RPA for the TDDFT but no kernel for the ACFD (testing purposes).
• 1 => RPA for the TDDFT and ACFD.
• 2 => ALDA (PW92) for the ACFD
• 3 => PGG for the ACFD [M. Petersilka, U.J. Gossmann and E.K.U. Gross, PRL 76,1212 (1996)]
• 4 => BPG for the ACFD. This amounts to half the PGG kernel plus half the ALDA kernel for spin-compensated systems [K. Burke, M. Petersilka and E.K.U. Gross, in "Recent Advances in Density Functional Methods", Vol. III, edited by P. Fantucci and A. Bencini (World Scientific, Singapore, 2002)]
• 5 => Linear energy optimized kernel [J. Dobson and J. Wang, PRB 62, 10038 (2000)]
• 6 => Non-linear energy optimized kernel [J. Dobson and J. Wang, PRB 62, 10038 (2000)]

• 0=> do "usual" xc quadrature on fft grid
• 1=> do higher accuracy xc quadrature using fft grid and additional points at the centers of each cube (doubles number of grid points)--the high accuracy version is only valid for boxcut>=2. If boxcut < 2, the code stops.

NOTE : NOT USED AT PRESENT (v5.3.0)

This variable is a bit-wise combination of what will be written into / read from a special WFK/DEN/POT file. The contents of the file follow the Nanoquanta/ETSF file format specifications.

Please check the "etsf_io" module of the ETSF I/O library for possible values.

NOTE : NOT USED AT PRESENT (v5.3.0)

This variable tells what will be written into / read from a special WFK/DEN/POT file. The contents of the file follow the Nanoquanta/ETSF file format specifications.

Please check the "etsf_io" module of the ETSF I/O library for possible values.

• 0 => density not changed (fixed charge), forces not corrected
• 1 => density not changed, forces corrected with rigid ion hypothesis (atomic charge moved with atom)
• 2 => density changed and forces corrected with rigid ion hypothesis (atomic charge moves with atom)
• 3 => density changed and forces corrected with a different implementation of the rigid ion hypothesis
• 4 => density not changed, forces corrected with the use of Harris functional formula (*)
• 5 => density changed using D. Alfe 2nd-order algorithm (**), forces not corrected
• 6 => density changed using D. Alfe 2nd-order algorithm (**) and forces corrected with the use of Harris functional formula (*)

• 0 => hessian is the identity matrix
• 1 => hessian is 0.5 times the identity matrix
• 2 => hessian is 0.25 times the identity matrix
• -1=> hessian is twice the identity matrix
• ... (simply corresponding power of 2 times the identity matrix)

• 1=> do NOT take advantage of the time-reversal symmetry
• 2=> use time-reversal symmetry for k=( 0 0 0 )
• 3=> use time-reversal symmetry for k=(1/2 0 0 )
• 4=> use time-reversal symmetry for k=( 0 0 1/2)
• 5=> use time-reversal symmetry for k=(1/2 0 1/2)
• 6=> use time-reversal symmetry for k=( 0 1/2 0 )
• 7=> use time-reversal symmetry for k=(1/2 1/2 0 )
• 8=> use time-reversal symmetry for k=( 0 1/2 1/2)
• 9=> use time-reversal symmetry for k=(1/2 1/2 1/2)
• 0=> (preprocessed) for each k point, choose automatically the appropriate time-reversal option when it is allowed, and chose istwfk=1 for all the other k points.

Sets proper input values for the determination of U and J i.e. for pawujat (first atom treated with PAW+U), irdwfk (=1), tolvrs (=10^(-8)), nstep (=255), diemix (=0.45), atvshift (pawujat) pawujv). Do not overwrite these variables manually unless you know what you do.

• macro_uj=1 (and nsppol=2) Standard procedure to determine U on atom pawujat through a shift of the potential on both spin channels.
• macro_uj=1 (and nsppol=1) Non standand procedure to determine U from potential shift on atom pawujat (experimental).
• macro_uj=2 (and nsppol=2) Non standand procedure to determine U from potential shift on atom pawujat through a shift on spin channel 1 on this atom and the response on this channel (experimental).
• macro_uj=3 (and nsppol=2) Standand procedure to determine J from potential shift on spin channel 1 on atom pawujat and response on spin channel 2 (experimental).

In case of non-standard, blocked algorithms for the optimization of the wavefunctions (that is, if wfoptalg=1 or wfoptalg=4):

• if wfoptalg=1, nbdblock defines the number of bands (or states) in a block, and, by the same token, the maximal number of processors that can be used for that type of parallelism.
• if wfoptalg=4, nbdblock defines the number of blocks (the number of bands in the block is then nband/nbdblock ), at variance with the case wfoptalg=1.

When nctime is non-zero, the molecular dynamics information is output in NetCDF format, every nctime time step. Here is the content of an example file :

netcdf md32.outH_moldyn1 {
dimensions:
        time = UNLIMITED ; // (11 currently)
        DimTensor = 6 ;
        DimCoord = 3 ;
        NbAtoms = 32 ;
        DimVector = 3 ;
        DimScalar = 1 ;
variables:
        double E_pot(time) ;
                E_pot:units = "hartree" ;
        double E_kin(time) ;
                E_kin:units = "hartree" ;
        double Stress(time, DimTensor) ;
                Stress:units = "hartree/Bohr^3" ;
        double Position(time, DimCoord, NbAtoms) ;
                Position:units = "Bohr" ;
        double Celerity(time, DimCoord, NbAtoms) ;
                Celerity:units = "Bohr/(atomic time unit)" ;
        double PrimitiveVector1(DimVector) ;
        double PrimitiveVector2(DimVector) ;
        double PrimitiveVector3(DimVector) ;
        double Cell_Volume(DimScalar) ;
                Cell_Volume:units = "Bohr^3" ;
}

• ndyson=-1 : let the code decide how many points to use (presently, 3 points for idyson=1 or 3, and 9 points for idyson=2)
• ndyson=0 : only compute the non-interacting and fully-interacting susceptibility matrices.
• ndyson>0 : use ndyson more points in ]0,1[

• normpawu=0 : leave projector
• normpawu=1 : renormalize

• tfkinfunc=1 : Thomas-Fermi kinetic functional (explicit functional of the density) is used instead of Kohn-Sham kinetic energy functional (implicit functional of the density through Kohn-Sham wavefunctions).
  
• tfkinfunc=2 : the Recursion Method is used in order to compute electronic density, entropy, Fermi energy and eigenvalues energy. This method computes the density without computing any orbital, is efficient at high temperature, with a efficient parallelization (almost perfect scalability). When that option is in use, the ecut input variable is no longer a convergence parameter ; ngfft becomes the main convergence parameter: you should adapt ecut for the ngfft grid you need (it is not yet automatically computed). Other convergence parameter are for the energetic values: recnrec, recptrott, recnpath. Since the convergence of the self-consistent cycle is determined directly by the convergence of the density: toldfe, toldff tolrff, tolvrs, tolwfr are not used, and are replaced by rectolden; the energetic values, except for the fermi energy, are only computed during the latest SFC cycle : the output file will show a jump of the total energy at the end, but it is not because of a bad convergence behavior. Computational speed can be improved by the use of recrcut and recgratio. The recursion method has not be tested in the case of non cubic cell or with the use of symmetries. In the recursion method the following variables are set to:
  useylm=1,
  userec=1.

P_lmn(r)=Y_lm(r)*p_ln(r)

• 0: no correction.
• 1: apply vdW-DF1 (DRSLL) from Dion et al.
  doi:10.1103/PhysRevLett.92.246401
• 2: apply vdw-DF2 (LMKLL) from Lee et al.
  arXiv:1003.5255v1
• 10: evaluate the vdW correlation energy from maximally localized Wannier functions, as proposed by P. L. Silvestrelli. doi:10.1103/PhysRevLett.100.053002

• (vprtrb(1)+I*vprtrb(2))/2 at G=qprtrb and
• (vprtrb(1)-I*vprtrb(2))/2 at G=-qprtrb (see qprtrb also).

• wfoptalg=0 : standard state-by-state conjugate gradient algorithm, with no possibility to parallelize over the states;
• wfoptalg=1 : blocked conjugate gradient algorithm, with possibility to parallelize over the states (or bands), but at the expense of a few more operations when a block of states has been optimized separately, to obtain a coherent set of wavefunctions. The number of states in a block is defined in nbdblock
• wfoptalg=2 : minimisation of the residual with respect to different shifts, in order to cover the whole set of occupied bands, with possibility to parallelize over blocks of states (or bands). The number of states in a block is defined in nbdblock. THIS IS STILL IN DEVELOPMENT.
• wfoptalg=3 : minimisation of the residual with respect to a shift. Available only in the non-self-consistent case iscf=-2, in order to find eigenvalues and wavefunctions close to a prescribed value.
• wfoptalg=4 : Locally Optimal Block Preconditioned Conjugate Gradient (lobpcg) method of Knyazev. Reference : A.V. Knyazev, "Toward the Optimal Preconditioned Eigensolver : Locally Optimal Block Preconditioned Conjugate Gradient Method". Siam Journal on Scientific Computing 23, pp517-541 (2001). The implementation (by G. Zerah), rests on the matlab program by Knyazev. For more information see Knyazev page.
• wfoptalg=10 : (for PAW) standard state-by-state conjugate gradient algorithm, with no possibility to parallelize over the states, but modified scheme described in Kresse, Furthmuller, PRB 54, 11169 (1996) (modified kinetic energy, modified preconditionning, minimal orthogonalization, ...) ;
• wfoptalg=11 : blocked conjugate gradient algorithm, with possibility to parallelize over the states (or bands), but at the expense of a few more operations when a block of states has been optimized separately, to obtain a coherent set of wavefunctions. The number of states in a block is defined in nbdblock. Modified scheme described in Kresse, Furthmuller, see wfoptalg=10.
• wfoptalg=14 : Locally Optimal Block Preconditioned Conjugate Gradient (lobpcg) method of Knyazev. Reference : A.V. Knyazev, "Toward the Optimal Preconditioned Eigensolver : Locally Optimal Block Preconditioned Conjugate Gradient Method". Siam Journal on Scientific Computing 23, pp517-541 (2001). The implementation (by G. Zerah), rests on the matlab program by Knyazev. For more information see Knyazev page. At odds with wfoptalg=4, the preconditionning of the block vectors does not depend on the kinetic energy of each band. In addition, the orthogonalization after the LOBPCG algorithm is no more performed. The first modification increases the convergency and the second one the efficiency.