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 June 2007 and May 2008. These release notes
are relative to improvements of ABINITv5.5 with respect to v5.4.

The list of contributors includes :
B. Amadon, P.-M. Anglade, A. Bergman, F. Bottin, D. Caliste,
R. Caracas, H. Chunping, T. Deutsch, L. Diaz, L. Genovese,
A. Ghasemi, Ph. Ghosez, M. Giantomassi, S. Goedecker, D. Hamann,
N. Helbig, F. Jollet, E. Luppi, A. Neelov, T. Nishimatsu,
V. Olevano, Y. Pouillon, G.-M. Rignanese, L. Sindic, M. Torrent, V. Veniard,
M. Verstraete, X. Wang, 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 5.5, released on November 23, 2007.

Changes with respect to version 5.4 :

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 Since its very beginning of its availability inside ABINIT,
    the utility "optic" had not been correctly interfaced with
    the main ABINIT program.
    In particular, at the level of the linear optical response,
    the anti-resonant contribution was missing.
    Also, an incorrect factor for the matrix elements of the
    momentum operator was taken into account.
    These problems have been identified and corrected by E. Luppi and V. Veniard.

A.2 The ordering of symmetry operations, for GW calculations,
    has been modified, due to the implementation of non-symmorphic
    operations (by M. Giantomassi), and the needed coherence in mrgscr.
    As a consequence, the selection of the q points in the irreducible
    part of the Brillouin zone has changed. This might
    induce some backward incompatibility.

A.3 Two bugs have been fixed in the computation of forces in the PAW
    case, by M. Torrent.
    Numerical values might be affected at the level of 10d-3 Ha/Bohr
    or so. The previous (incorrect) behaviour might still be activated
    by changing the value of the abinit_prior55 variables in the subroutines
    pawxcm.F90 and nres2vres.F90, if needed for backwards checks ...

A.4 The default value for pawmixdg was documented to be 0, but was actually 1.
    This has been fixed: the default for pawmixdg is indeed 0 now.

A.5 The v5.5.1 default value for the input variable fftgw corresponds to a
    computation not completely identical to the references for v5.4
    The quantities computed within the GW approximation might have slightly
    changed.

A.6 Change of name for GW computations : spectral has become spmeth .

A.7 Change of name for treatment of Coulomb interation : icoultrtmt has become icoulomb .

A.8 The cpp flag -DMPI_FFT is not used anymore. It has been replaced
    by the use of the input variable paral_kgb , see B.2 .
    The use of MPI_IO has also been cleaned, and a new automatic test
    has been introduced, paral#Q (but not activated by default,
    because the MPI_IO library is not yet standard on all machines
    that have MPI).
    Work by F. Bottin and G. Zerah

A.9 All the lines of the header of the _DOS files now start with a '#' sign in the first colum

__________________________________________________________________

B.  Most noticeable achievements

B.1 The wavelet (BigDFT) part of ABINIT can be used for production.
    It corresponds also to the release of the version 1.0 of the
    BigDFT library.
    The BigDFT library is now compiled by default.
    The automatic tests in the directory tests/bigdft are integrated
    in the standard suite of tests, driven by
    make tests_acc
    Collinear spin polarisation has been implemented
    (see tests BigDFT #12 and #13).
    Other minor improvements include the implementation of Krack
    pseudo-potentials (see point B.2 of release notes v5.4) and the
    fractional occupation (fixed, not evolving during minimisation).
    See several input variables (to be listed). In particular,
    in case of Bigdft wavelet Poisson solver use, see nscforder.
    Effort from the BigDFT community: D. Caliste, L. Genovese,
    S. Goedecker, T. Deutsch, A. Neelov, A. Bergman, O. Zilberberg.

B.2 The coupled k-point+Spin/Band/FFT parallelisation can be used for production
    (although the user-friendliness might be improved).
    Related input variables : paral_kgb, npband, npfft, npkpt.
    Test paral#P4 is an example of a Band/FFT parallelism.
    Test paral#R4 is another example of a Band/FFT parallelism, with PAW.
    Work by F. Bottin and G. Zerah.

B.3 Berry phase can be computed with PAW.
    See automatic test v5#12 and 13.
    Work by P. Hermet, B. Amadon, F. Jollet and Ph. Ghosez.
    (Warning : in v5.5.2, still seg fault to be solved)

B.4 The response with respect to electric field, in the presence of
    spin-orbit coupling, can be computed, giving access
    to Born effective charges and dielectric tensor. Work by XG and M. Torrent.
    See automatic test v5#24
    (Warning : in v5.5.2, still some checks to be done ...)

B.5 Computation of Electric Field Gradient is available
    within LDA+U. See automatic test v5 # 53.
    Also in the spin-polarized case.
    Coding from J. Zwanziger, with help from M. Torrent.
    Also, additional test for EFG test v5 # 52 (Stishovite GGA)

B.6 Large improvement in GW computations (v5.5.1 to be documented),
    including cut offs for specific geometries.
    New automatic tests to be set up.
    New input Variables nshadd, icutcoul, rcut.
    Work by M. Giantomassi.

B.7 PAW can now be used to compute optics properties.
    New routine linear_optics_paw , as an option to conducti .
    To be documented with automatic tests.
    Work by J. Zwanziger.

B.8 The m-resolved density of states is available.
    Coded by T. Nishimatsu.
    Automatic tests to be provided (Tv4#35, 38, 39 will be modified
    if lines 476 and later are activated in tetrahedron.F90),
    domain of use should be examined.

B.9 NetCDF and ETSF_IO libraries are now correctly integrated in the build system.
    (v5.5.4 : already tested for four different compilers on PC, should still
     be tested for more platforms).

__________________________________________________________________

C. Changes for the developers


C.1 Introduction of the abiauty script by LSindic .
    The indentation of the F90 files is now automatic.

C.2 The new precisions 'gwp' and 'gwpc' have been introduced, see defs_basis.F90 .
    They will allow to do GW calculations either in single precision
    or in double precision, to make the automatic tests more accurate.

C.3 The default length for file names has been changed from 132 to 264 ,
    see defs_basis.F90 .

C.4 The values for the fundamental constants have been changed
    from the 2002 adjustment to the 2006 adjustment,
    see defs_basis.F90 .

C.5 The test cases bigdft#06, 07 and 08 have been desactivated, since they
    required the activation of the ETSF_IO library, that is not yet in production.

C.6 The platform that produces the references is now PC Intel + Ifort 8.1 .

C.7 The possibility of NOT using HAVE_FORTRAN_INTERFACES has been desactivated
    and the build system has been simplified accordingly (Y. Pouillon, T. Deutsch)

__________________________________________________________________

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

D.1 New input variables ndivsm (for : number of division of the
    smallest segment) for the construction of band structures.
    This alternative to ndivk is easier to use.
    (v5.5.1 : no automatic test yet)
    Coding from Matteo Giantomassi

D.2 NMR chemical shielding factor is in development.
    See the new input variables prtcs and corecs (diamagnetic
    contribution from core to be defined by the user).
    Coding from J. Zwanziger

D.3 New input variable dmatpuopt.
    To define the expression of the density matrix in LDA+U (PAW)
    Automatic test needed, doc needed.
    Coding by B. Amadon.

D.4 Reduced matrix functional theory coding is going on.
    Work by N. Helbig. New input variable : rdmnb.
    To be documented.

D.5 Work on the switch to bzr version control system, and associated documentation,
     by Y. Pouillon .

D.6 Improvements of the build system, including new automatic options,
    by Y. Pouillon.

D.7 Preparation of nightly builds for developpers using bzr, by Y. Pouillon.

D.8 Modifications of the doc needed to have ABINIT in Debian, by Y. Pouillon, D. Hamann,
    V. Olevano.

D.9 Use of rfelfd=3 , instead of rfuser for finite electric field
     computation of Born effective charges and dielectric tensor,
     by Xinjie Wang

D.10 Implementation of response functions within PAW. Work going on,
     by M. Torrent.

D.11 Spin-polarized is OK with prtsurf input variable, from M. Verstraete.

D.12 Improvement of NetCDF support by Y. Pouillon

D.13 Extension of the ODA mixing in the GGA case
     Coding by F. Jollet.

D.14 Prediction of the density for self-consistency, using D. Alfe's algorithm
     (iprcch>4).
     Improvement of the prediction of rhoij for self-consistency in case of PAW,
     by M. Torrent

D.15 The ETSF-IO library can be used for the KSS file.
     Coding by M. Giantomassi.

D.16 Documentation improvements by : Y. Pouillon

D.17 The analysis of maxima and minima of the density, spin-density and magnetization
     is now much more extensive than previously for collinear and non-collinear spin
     polarized cases. Work by Luis Diaz.

D.18 Bug fixes by : D. Caliste, R. Caracas, Hu Chunping, M. Cote, T. Deutsch,
     L. Sindic, M. Verstraete