Copyright (C) 2008 ABINIT group (XG)
This file is distributed under the terms of the GNU General Public License, see ~abinit/COPYING or http://www.gnu.org/copyleft/gpl.txt .
For the initials of contributors, see ~abinit/doc/developers/contributors.txt .

Many thanks to the contributors to the ABINIT project
between February 2007 and October 2007. These release notes
are relative to improvements of ABINITv5.4 with respect to v5.3.

The list of contributors includes :
B. Amadon, P.-M. Anglade, M. Arai, F. Bottin, P. Boulanger,
F. Bruneval, D. Caliste, E. Cances, S. Caravati, M. Crack, T. Deutsch,
Luis Diaz, L. Genovese,
A. Ghasemi, M. Giantomassi, S. Goedecker, D. Hamann, N. Helbig, F. Jollet,
T. Nishimatsu, H. Mathis, S. Mazevet, A. Neelov, Y. Pouillon, C. Redaelli,
R. Shaltaf, M. Torrent,
M. Verstraete, Xinjie Wang, Zhigang Wu, G. Zerah, O. Zilberberg, J. Zwanziger

It is worth to read carefully
all the modifications that are mentioned in the present file,
and examine the links to help files or test cases ...
This might take some time ...
Please note the WARNINGS !

Xavier

__________________________________________________________________

Version 5.4, released on June 6, 2007.

Changes with respect to version 5.3 :

A. WARNINGS.
B. Most noticeable achievements (for users)
C. Changes in the package, for developers
D. Other changes (or on-going developments, not finalized)

__________________________________________________________________

A.  WARNINGS

A.1 A life cycle is now defined for each version : 14 months after the
    initial release of some version of ABINIT, there will be a warning,
    asking the user to switch to a more recent version. After 18 months,
    ABINIT will stop.

A.2 The input variable so_typat has been declared obsolete, and replaced
    by so_psp . Although the new variable so_psp should play a role similar
    to so_typat, there are two major differences :
    (1) so_psp is defined
    to affect each pseudopotential, and not each type of atom (this is to permit
    to use alchemical mixing of spin-orbit part, and to explore easily a dependence
    of the calculation on the strength of the spin-orbit coupling)
    (2) so_psp is to be set to 0 to suppress the treatment of the spin-orbit
    coupling, while so_typat had to be set to 1 . On the contrary, setting so_psp to 1
    means that spinorbit coupling will be used, with the correct formulation deduced
    by the content of the pseudopotential file.
    So, some input files with nspinor=2 and for which pseudopotential(s) with spin-orbit are used
    might have to be adapted with respect to v5.3 or earlier version .

A.3 The strain is now symmetrized, during the geometry optimization procedure.
    See the new routine strainsym.F90 . This means that
    the output of some geometry optimization might slightly differ from the
    previous versions, in which only the stress was symmetrized. In principle,
    this correction should fix some problems, where it was observed that
    the symmetry was lost during geometry optimization.

A.4 In case of GW calculation, there is a potential bad scaling (Npw^4) of the findggp.F90
    routine. If this happens, one should simply disactivate this routine, by inserting
    a "return" statement as the first executable instruction.

__________________________________________________________________

B.  Most noticeable achievements

B.1 Electric field gradient at nuclei site is now available.
    Coding from J. Zwanziger
    See test v5 # 51 , computation of EFG for stishovite.

B.2 A new format for pseudopotentials is available,
    pspcod=10, giving access to a library of pseudopotentials
    also used by CPMD.
    It is a slightly modified version of the existing HGH and GTH
    formats (pspcod=2 and 3), but there are some differences in the
    definition of the parameters.
    The library is available at
    http://cvs.berlios.de/cgi-bin/viewcvs.cgi/cp2k/potentials/Goedecker/abinit
    In this library, the already existing HGH and GTH pseudopotentials
    are provided ('pade' folder) in this new format, but there
    are also pseudopotentials for other XC flavors, like the PBE,
    and the HCTH-120 and HCTH-407 (see section B3).
    Tests are available v5#54 and 55 .
    Implementation of reading by S. Caravati, generation of psps by M. Crack.

B.3 The whole family of HCTH XC functionals is now available :
    in addition to HCTH-93, that was already present, the HCTH-120, HCTH_147
    and HCTH-407 are now available. A few pseudopotentials exist,
    see section B.2 . These functionals are thought to be particularly
    accurate for simulation of biological systems and water.
    ( Should be tested - test also the different HCTH functionals,
    120, 147 and 407, with the 93 and a similar pseudo (local, simply smoothed) for H,
    then use a tailored pseudo for HCTH 120 and 407 ).

B.3. The Wu and Cohen GGA XC functional is now available, use ixc=23 .
    Reference : Z. Wu and R.E. Cohen, Phys. Rev. B 73, 235116 (2006).
    See tests v5 # 28, 48 and 49.

B.4 The coupled k-point+Spin/Band/FFT parallelisation is available, for testing purposes.
    Tests should be set up.
    Work by F. Bottin anf G. Zerah.

B.5 Spectral method for set-up of the frequency-dependent matrices, in the GW calculation.
    Tests should be set up
    GW can take advantage of non-symmorphic symmetry operations. See test v5#64.
    Work by M. Giantomassi

B.6 Many new features or improvements in PAW (some additional tests should be provided) :
     - Added possibility to print partial DOS in PAW (contrib. from S. Mazevet + M. Torrent)
     - new feature: integrated density inside atomic spheres
     - new feature: keeping density matrix constant during the first
         SCF iterations for PAW+U (from M.Torrent & B.Amadon)
     - Some inputs for tests have been changed in order to take into account new prtdensph keyword
         (see also new test v5 #11)

B.7 Computation of optical quantities, from the utility conducti in the PAW case :
    frequency-dependent dielectric tensor (both real and imaginary part),
    index of refraction, conductivity, reflectivity, absorption
    A test is available : v5#9 and 10.
    (S. Mazevet and F. Jollet)

B.8 Possibility to add a jellium slab (and also bulk), by S. Caravati.
    See tests v5#56-59 . New input variables : jellslab,
    slabwsrad, slabzbeg, slabzend .

B.9 Complete cubic-scaling implementation of the wavelet part of
    ABINIT (BigDFT project). This version includes, the ground state
    computation (energy and forces), the finite size corrections to the
    total energy, restart on wavefunctions following the ETSF norm and geometry
    relaxation using BFGS. Molecular dynamic is also available for
    test purposes. Memory occupation optimisation is still under development.
     Effort from the BigDFT community : S. Goedecker, L. Genovese, T. Deutsch,
     A. Neelov, D. Caliste, A. Ghasemi, O. Zilberberg.

B.10 Born effective charges and dielectric tensor calculations can now be computed
     in a finite electric field.
     For the time being, such computations are activated by setting
     rfuser to 1 . This should be changed in a future version of ABINIT.
     See automatic test v5#23.
     Contribution by Xinjie Wang.


__________________________________________________________________

C. Changes for the developers

C.1 See D.2 . The variable names being now checked, when a new variable
    is added, the list of admitted variables in chkvars.F90 must be updated.

C.2 Creation of a new subdirectory of src : src/21rdm , for developments
    related to Reduced-Density Matrix functional Theory, by N. Helbig
    helped by M. Giantomassi..

C.3 A new directory has been created, to host the Wannier90 library.
    See D.3

C.4 The Poisson solver library is now outside of ABINIT within its
    parent library: BigDFT. Thus, using icoultrtmt = 1 needs to
    compile with --enbale-bigdft. The directory 12poisson has been
    removed and replaced by 62_poisson because the XC is also computed
    now by the Poisson solver routine (since rhor is directly available).

__________________________________________________________________

D.  Other changes
(or on-going developments, not yet finalized).

D.1 New automatic test Tv5#43 : test Bismuth atom, spin-polarized,
    with collinear settings, as well as spinor settings.

D.2 The existence of variable names mentioned in the input file is now checked,
    so that typos are identified.

D.3 Interfacing with Wannier 90 done, by B. Amadon . Use prtwant=2 .
    Not yet in production

D.4 Berry phase for PAW, by F. Jollet (see mail)
    Not yet in production

D.5 ODA mixing scheme, by E. Cances and F. Jollet (see mail)
    Not yet in production

D.6 Update to the upstream up-coming version of the ETSF-IO library, by
    D. Caliste. This change allow to define precise variables into the
    output files. It also brings the support for experimental wavelet
    basis set. See KNOWN_PROBLEM #P54.11.

D.7 Work in progress : computation of the temperature dependence of the band gap
    THe new input variable ieig2rf has been defined.
    (P. Boulanger)

D.8 Parallelism over bands for GW computations has been reinstalled. Test N is OK again.
    A minor problem KNOWN_PROBLEM #P53.27 is still there, though .
    Work by M. Giantomassi

D.9 Many bug fixes and improvements for the build system, from Y. Pouillon

D.10 Possibility to have plug-ins, from Y. Pouillon.
     Wannier90 library activation follow this path.

D.11 Cleaning of ABINIT concerning the use of precision, by P.M. Anglade .

D.12 Work on improved preconditioners, by P.M. Anglade .

D.13 M. Giantomassi provided Test v5#63 for GW calculations with antiferromagnetic
     symmetries.

D.14 Work on the switch to bzr version control system, and associated documentation,
     by Y. Pouillon .

D.15 Description of path in reciprocal space (kptbound input variable)
     for the rhombohedral lattice (in doc/input_variables/vargs.html)

D.16 For PAW, it is now possible to develop the XC term in the spherical part of the
     energy up to 2nd moments. See the input variable pawxcdev.
     Implemented by M. Torrent.

D.17 Bug fixes by : F. Bruneval, M. Giantomassi, D. Hamann, M. Torrent, Y. Pouillon,
     M. Verstraete

D.18 Cut3d has been (partly) generalized to non-collinear spin. Work by Luis Diaz.
     See tests v5 # 32 and 33 .

D.19 abirules.pl insert automatically the release date, from L. Sindic

D.20 Parallelism over k points for quasi-particle self-consistent GW has been enabled
     by Don Hamann. A new test case (case O) has been created in the suite of parallel tests).

D.21 Bug fixes by : M. Arai, F. Bruneval, M. Giantomassi, D. Hamann, M. Torrent, Y. Pouillon,
     M. Verstraete, L. Sindic, T. Nishimatsu, H. Mathis, C. Redaelli

D.22 Implementation of the reversible integrator of Martyna et al,
     see ionmov=13 for more explanations,
     by G. Zerah. New input variables : nnos , qmass , bmass , vmass .