Copyright (C) 2002 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 .
Many thanks to the following contributors to the ABINIT project
between November 2001 and February 2002:
J.-M. Beuken, J.-F. Brie're, M. Cote', R. Caracas,
G. De Cock, T. Deutsch, Ph. Ghosez, F. Jollet,
B. Magne, M. Mikami, V. Olevano, J.-Y. Raty, V. Recoules, A. Roy,
L. Sindic, S. Tausendpfund, M. Torrent, M. Verstraete,
M. Veithen, G. Zerah

I think 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 3.3, released on March 8, 2002.

Changes with respect to version 3.2 :

A. Changes for the installation.
B. Most noticeable changes in the package
C. WARNINGS.
D. Other changes.

__________________________________________________________________

A.  News concerning the installation.

A1. Files for installation on MAC under OS X are available
    (~abinit/Machine_dept_files/MAC/makefile_macros.macosx).
    The files previously needed for previous versions of MacOS
    have been deleted.
    (JMB + XG)

A2. The automatic testing has been made much more portable, so that
    the fldiff files are usually much easier to decipher (XG)
    New commands to make the automatic tests are available :
    make tests_allseq
    make tests_dev
    (Just type   make dev   to have a brief explanation,
     or see the end of the section "How to make the other tests ?"
     in the installation notes for v3.3)

A3. New directories have been created :
    Src_3paw (for Projector Augmented Wave routines - FJ and MT),
    Src_7lwf (for Lattice Wannier Functions, RC).

A4. The abinit Web site is now located at https://www.abinit.org

__________________________________________________________________

B.  Most noticeable changes.

B1. A MPI band-by-band parallelism for the ground state calculations
    has been coded by Aline Roy.
    See the input variables nbdblock (should be equal to the
    number of processors the user wishes for this
    parallelisation), and wfoptalg (that should be equal to 1).
    The code must have been compiled with the
    MPI flags in the makefile_macros for this option to be available.
    This parallelization can be used at the same time as the MPI k-point
    parallelisation, the MPI spin-parallelisation, or the OpenMP
    parallelisation over plane wave coefficients.

    WARNING : although this is an exciting new feature,
    the algorithm used in this parallelisation will likely be efficient
    only for a quite small number of processors (lower than 10 ?).
    Also, we have NOT worked on the decrease of the memory needs.
    Another type of algorithm is presently under study by Christophe Audouze
    and Francois Alouges.

    However, we hope that it will be a very useful parallelisation
    as soon as the parallelisation over k points is not interesting
    (especially when nkpt=1 !!). Further developments are likely
    in v3.4 . Still, if you want to be testers of this
    possibility, please go on ...

B2. Improvements for Time-Dependent DFT for excited states (iscf=-1).
    The option mkmem==0 allows now to store
    the real space wavefunctions on disk : it is possible to
    use TDDFT for much bigger systems than previously. This has been
    done by J.-Y. Raty, who has tested SiH clusters with about 70 atoms.
    Also, the possibility to restart the calculation has been implemented.
    This has also been done by J.-Y. Raty.
    The computation of Cauchy coefficients has been implemented
    and a new test of TDDFT, for N2, is available (both by XG).

B3. New GGAS are available (however, corresponding pseudopotentials
    should still be generated) :
    - ixc=14 => revPBE of Y. Zhang and W. Yang, Phys. Rev. Lett. 80, 890 (1998)
    - ixc=15 => RPBE of B. Hammer, L.B. Hansen and J.K. Norskov,
                 Phys. Rev. B 59, 7413 (1999)
    - ixc=16 => HCTH functional of F.A. Hamprecht, A.J. Cohen,
                 D.J. Tozer and N.C. Handy, J. Chem. Phys. 109, 6264 (1998).
    The latter should be particularly accurate for chemical
    applications (outperforms the B3LYP in the energetics of
    small molecules). (XG)

B4. Response functions are available for different GGAs :
    PBE and associated functionals : ixc=11,12,14,and 15.
    The tests have been quite intensive.
    It has been checked with spin-polarized or not,
    non-linear XC correction or not, also q=0 or not, or electric field.
    See Test_v3, test cases #7,8,16,17,18, 62-66.
    The memory needs have not yet been optimized.
    (XG)

B5. The localization tensor
    (see C. Sgiarovello, M. Peressi, and R. Resta, Phys. Rev. B 64,
    115200 (2001)), has been coded by M. Veithen,
    as well as the band-by-band decompositions of the Born
    effective charges (see Ph. Ghosez and X. Gonze,
     J. Phys.: Condens. Matter 12, 9179 (2000)), and the
    localization tensor (see M. Veithen, X. Gonze, and Ph. Ghosez,
    AIP proceedings, to be published).
    However, at the level of the localisation tensor, no symmetry
    is yet used, so that its use is not straightforward,
    contact Marek Veithen if you want to use it.

B6. The computation of the optical conductivity tensor using
    the Kubo-Greenwood formula has been coded by V. Recoules (in
    collaboration with G. Zerah). It is available as a new utility :
    conducti . See Test_v3, cases 68 (ABINIT preparation)
    and 69 (conducti run). A comprehensive documentation
    is still to be written.

__________________________________________________________________

C.  WARNINGS

C1. Switch to 1998 Recommended Values of the Fundamental Constants.
    This typically changed the values of many quantities
    by one part per 1.0d5 . See ~abinit/src/*/defs_basis.f
    By the way, the use of conversion factors defined in defs_basis.f
    has been enforced in several routines. (XG)

C2. Two differences in the protocol for developpers, see the
    ~abinit/doc/developers/collab document.
    1) The list of routines allocated to one developper is not
       needed anymore : the use of CVS is sufficiently effective
       at merging time. Still the concept of "task" remains :
       two developers should work on the same functionality
       of ABINIT
    2) The delivery of the output of the automatic tests is now
       mandatory. To help this, the automatic tests
       can be executed now using the command
       make tests_dev
       in the ~abinit directory. This command produces
       a "summary_tests.tar.gz" to be mailed with the modified
       routines, comments, new automatic tests, new documentation,
       etc ...

__________________________________________________________________

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

D1. A package (in development) towards the use of autoconf
    has been contributed by STausendpfund.

D2. JF Briere (in collaboration with M Cote) has developped
    the use of the Data Explorer (IBM) freeware to analyze the results
    of ABINIT, via cut3d, especially the analysis of wavefunction
    files.

D3. The ROBODOC list is now established automatically, and
    the script abirules can be used to check or to enforce
    more features than previously
    (e.g. the change of the copyright year 2001 -> 2002
    has been done automatically).
    (contributed by Luc Sindic)

D4. Test 8 has been created (spin-parallel) (XG)

D5. The work on the GW part of ABINIT continues : many
    small modifications or bug elimination (Valerio Olevano)

D6. Different input variables, whose dimension are energies or
    lengths, can now be input with an alternative unit than atomic units.
    (XG)

    For example, acell can now be specified in Angstrom.
    This is explained in ~abinit/doc/users/abinis_help.html , section 3.2,
    see the description of the 'ENERGY' and 'LENGTH' characteristics
    of input variables. See also the input variables acell,
    ecut, tsmear, ..., and the tests cases Test_v3#52 and 53.
    Also, logical and character values in input files
    are now allowed in response to selected token.
    See also Test_v3#52 and 53. Until now, the latter is not particularly
    useful, and the documentation is still lacking.
    However, this is a step towards being able to read parts of the
    SIESTA and GAUSSIAN input files.

D7. Implementation of lattice Wannier functions was started,
    by Razvan Caracas. See the Src_7lwf directory, and Src_main/lwf.f .

D8. More URLs in Infos/other_codes and Infos/Miscellaneous/Crystals,
    from MMikami

D9. Bugs removed : G. DeCock, M. Verstraete, M. Mikami, G.-M. Rignanese,
    T. Deutsch and some of the above-mentioned collaborators.