Copyright (C) 2010-2024 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 May 2009 and January 2010. These release notes
are relative to improvements of ABINITv6.0 with respect to v5.8.
The list of contributors includes :
D. Adams, S. Albrecht, B. Amadon, P.M. Anglade, A. Berger, J.-M. Beuken, S. Blackburn,
F. Bottin, P. Boulanger, D. Caliste, M. Cazzaniga, V. Chevrier,
M. Cote, M. Cotelo, F. Da Pieve, M. Delaveau, T. Deutsch, L. Diaz, G.A. Franco, M. Gatti, M. Giantomassi,
X. Gonze, Chuping Hu, A. Jacques, F. Jollet, G. Jomard, S. Leroux, A. Lherbier, M. Mancini,
N. Manini, V. Olevano, M. Oliveira, G. Onida, Y. Pouillon, T. Rangel, V. Recoules, L. Reining,
G.-M. Rignanese, F. Sottile, M. Stankovski, M. Torrent, M. Verstraete, D. Waroquiers, 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 6.0, released on January 18, 2010.
Changes with respect to version 5.8 :
0. Foreword
A. WARNINGS.
B. Most noticeable achievements (for users)
C. Changes in the package, for developers
D. Other changes (or on-going developments, not finalized)
__________________________________________________________________
0. Foreword
Internal changes are quite large between ABINITv5 and ABINITv6.
Many developers have concentrated to clean and simplify
the package, as well as to make it more robust (many more automatic tests, see D.14).
Some of these modifications are visible to the
users, many are not. In particular, the automatic testing
has been considerably improved thanks to the setup of a farm of
machines, on which all the automatic tests are done nightly.
The consequences of the many modifications will be felt in
the forthcoming months, with likely an increased development rate.
Concerning the user's point of view, although already some
noticeable achievements are listed in section B, several of the
projects started during last year are not yet finalized
(described in section D).
In order to have a reasonable time gap between v5.8 and v6.0
it was however important not to wait more than a half year.
So, keep posted, ABINITv6.x is rather exciting !
__________________________________________________________________
A. WARNINGS AND IMPORTANT REMARKS
A.1 The build system has been strongly modified, and actually simplified : it builds
one executable "abinit", and not separately a "abinis" or a "abinip".
"abinit" will be a sequential executable or a parallel executable, depending
on whether MPI libraries have been enabled during the build.
Although most of the automatic tests execute with an executable
that is sequential as well as parallel, there is now a separate
test directory tests/seq to host specifically tests for the sequential executable,
while the already existing tests/paral still hosts specifically tests
for the parallel executable.
If you want to build both a sequential and a parallel version of
abinit, you can create 2 build dirs and run configure with different
parameters in each one of them. If you wish to differentiate the
executables at install-time, you may use for instance the
"--program-suffix" option of configure. See the output of
"./configure --help" for details.
(Work done by Yann Pouillon)
A.2 The license of Abinit is now GNU GPL v3+.
__________________________________________________________________
B. Most noticeable achievements
B.1 The computation of the derivative with respect to the wavevector (d/dk - DDK perturbation)
has been implemented, in the PAW case. This gives access to the first derivative of the eigenvalues
with respect to wavevector, and prepares the electric field response. See also D.2.
Automatic test Tv5#05, #30.
Work done by J. Zwanziger + M Torrent.
B.2 The computation of Raman scattering intensities (for norm-conserving psps)
has been very much improved :
(1) It is now possible in the spin-polarized LDA case (work by F. Da Pieve)
See the test case Tv6#62
(2) It is now possible with a grid of k points that has only one k-point per direction (work by S. Leroux)
See the test case Tv6#65
(3) It is parallelized over k points and tested (work by S. Leroux)
See the test case Tparal#E
B.3 Berry phase calculation of the polarisation.
The constraint to have at least two k points per direction has been removed (work by S. Leroux, needed for B.2)
See the test case Tv6#06
B.4 The computation of the second derivative of the eigenvalues with respect to atomic displacements
giving access to the temperature dependence of the band-gap (in the DFT) can be used in production
(for norm-conserving psps).
Input variables for ABINIT : eig2rf, bdeigrf
Automatic test : Tv5#26, 27, 28 and Tv6#61
Work by P. Boulanger with M. Coté and X. Gonze.
B.5 LDA+U + spin-orbit has been implemented. Presently only the FLL version of double counting is valid (usepawu=1)
Automatic test : Tv5#16 and 17
Work done by B. Amadon and M. Torrent
B.6 The computation of the positron lifetime and annihilation rate has been implemented within the PAW,
using the self-consistent two-component DFT. Actually, this part (already working for Norm-conserving PP),
has been rewritten from scratch, with added functionalities : full self-consistency,
force and stress computation, GGA on the electron-positron correlation (as well as other usual XC functionals).
Input variables : ixcpositron, positron, posnstep, posocc, postoldfe .
Automatic test Tv5#07
Work by M. Torrent and G. Jomard.
B.7 GW calculation using different treatments of the exchange for the core electrons
has been implemented : either the core contribution is approximated with vxc
or the core term is approximated with the Fock operator inside the PAW spheres.
Input variables : gw_sigxcore
Requirements: External files with core wavefunctions generated with patched version of Atompaw.
Automatic test Tv6#66
Work by M. Giantomassi.
B.8 GW calculation with the extrapolar methodology has been implemented for PAW.
No automatic test yet.
Work done by M. Giantomassi.
B.9 Different files can now be generated :
- the all-electron wavefunctions, using the input variable pawprtwf (by M. Giantomassi)
(Note that one has to be enable the support for ETSF-IO at configure-time
as the _AW_WFK file is written using the NETCDF file format following the ETSF-IO specification for
wavefunctions in real space. Only available for sequential runs.)
Automatic test Tetsf_io#09
- the gradient and laplacian of the density (for norm-conserving pseudopotentials),
using the input variables prtgden and prtlden (by A. Lherbier)
Automatic test Tv1#21
- the spin-polarized ELF (Electron Localisation Function) (by A. Lherbier)
B.10 Implementation of the molecular dynamics stepper proposed by Blanes and Moanes,
a symplectic algorithm, see the input variable ionmov=14 .
Test Tv5#03
Work by P.-M. Anglade
__________________________________________________________________
C. Changes for the developers
C.1 A script to examine the output of the compilation with numerous debugging options
has been written by Jean-Michel Beuken, and is launched with the tests Tabirules#3 to 14.
C.2 Systematic use of the python INI format for environment variables, and other configuration files (Y. Pouillon)
C.3 Change in the build system and the package :
- Interfaces are now in a numbered directory (Y. Pouillon)
- Capabilities to test MPI_IO (Y. Pouillon, X. Gonze)
- Improvement for future Scalapack testing (Y. Pouillon)
- CUDA support (Y. Pouillon)
C.4 The indentation has been changed throughout most of the routines : now two blanks indentation
are used inside a do-loop of if clause .
C.5 The full Lapack library is now shipped with ABINIT.
C.6 The parareel algorithm has been suppressed (M. Torrent). This should open the possibility
to clean the MPI implementation throughout ABINIT.
__________________________________________________________________
D. Other changes
(or on-going developments, not yet finalized).
D.1. The capability to perform Dynamical Mean Field calculations is in progress,
by B. Amadon.
Related input variables (not operational) usedmft, dmftbandi, dmftbandf
Fake test, not documented : Tv6#07
D.2 The use of Density Functional Perturbation Theory with PAW
(phonons and electric field) is in progress,
by M. Torrent, F. Jollet and J. Zwanziger.
D.3. The capability to define "images", to perform (not yet implemented) Nudged Elastic Band method optimization
of paths, or the string method, or metadynamics, etc has been implemented,
but should still be optimized.
See the input variables dynimage, fxcartfactor, imgmov, nimage, npimage, ntimimage.
Automatic tests Tv6#21 and 22
By X. Gonze
D.4 A new optdriver level has been addded for performing self-consistent GW calculations.
Related input variables (not operational) gw_nstep, gw_sctype, gw_sctype_name, gw_toldfeig.
Work by M. Giantomassi
D.5 The Bethe-Salpeter equation (BSE) can be used to compute the optical spectra
of materials. The BSE explicitly treats the effect of excitons.
This is much more accurate than the optics utility (based on non-interacting particles).
Related input variables (not operational) optdriver=99 (to be changed later)
bs_algorithm bs_haydock_niter bs_haydock_tol bs_exchange_term bs_coulomb_term
bs_calctype bs_coupling bs_haydock_tol bs_eh_basis_set bs_eh_cutoff bs_freq_mesh .
Fake automatic tests : Tv6#91
Both norm-conserving and PAW implementations have been done.
The PAW one must be used with lot caution, and cannot be considered in production,
while the norm-conserving one is closer to production (not quite so, though,
more testing is needed). In any case, it is much too early for beginners to use the BSE,
at least a tutorial should be written by ABINIT developers.
The BSE routines now present in ABINIT originate from the EXC code,
by F. Sottile, V. Olevano, L. Reining, M. Gatti, G. Onida, S. Albrecht, with
the porting realized by M. Giantomassi.
D.6 Implementation of the GW_OPTIMAL algorithm, by A. Berger, is in progress,
with help for the porting by M. Giantomassi.
Related input variables (not operational yet) gw_optimal_level, gw_nband_optimal.
D.7 It is now possible to specify the tolerance for the identification of space group, given
a possible numerical inaccuracy of the atomic positions. (XG)
Input variable tolsym
Automatic test : Tv5#78 .
D.8 Finite electric field calculations with PAW are in progress (J. Zwanziger)
D.9 Important limitations of the space group finder have been removed.
This is for cases where the information from the cell geometry only
was leading to a holohedry that was different from the real holohedry, obtained
by taking into account the cell geometry and atomic positions. (XG)
D.10 The use of MPI-IO for parallel writing has been considerably improved, and is now truly
portable (MTorrent).
D.11 The capability to compute the U or J values, within the LDA+U (or GGA+U) approach
according to the method of Coccocioni and De Gironcoli, is under development.
Input variable : macro_uj, pawujat, pawujv .
Automatic tests : Tv5#38, Tv6#41
Work by D. Adams, with B. Amadon
D.12 Work on meta-GGA going on, by M. Oliveira and A. Lherbier .
D.13 Work on the recursion approach to high-temperature DFT calculations, implementation of
GPU capabilities, by M. Mancini.
D.14 New tests have been introduced to demonstrate the use of input variables or capabilities
introduced earlier, but not documented/tested. Some tests were modified for the same goal.
- Tetsf_io#21 for nctime (XG)
- the whole tests/mpiio directory for MPI_IO testing (FBottin+XG),
- Tparal#J for tddft in parallel, including spin (Chunping Hu + XG)
- Tparal#L for Berry phase in parallel over k points, and finite-electric field in parallel over k points (XG)
- Tv1#25 to test effmass (XG)
- Tv1#43 to test objbax, and objbro (XG)
- Tv1#69 to test td_maxenei, td_mexcit (XG)
- Tv1#85 to test natfixx, iatfixx (XG)
- Tv2#15 to test iavfrq (MVerstra)
- Tv3#22 to test genafm (XG)
- Tv4#13 to test nbandsus, suskxcrs (XG)
- Tv4#17 to test bxctmindg, iboxcut, mqgriddg (MTorrent)
- Tv5#2 : to test ionmov=5 (GAFranco)
- Tv5#4 : to test getvel (GAFranco)
- Tv5#6 : pawnzlm, pawstgylm, pawusecp, pawlmix, pawlcutd, pawnhatxc, pawntheta, pawnphi (MTorrent)
- Tv5#11 : pawprtden (MTorrent)
- Tv5#17 : pawcpxocci, spnorbscl (MTorrent)
- Tv5#18 : exchmix (FJollet)
- Tv5#30 : rfmgfd (JZwanziger)
- Tv5#40 : irdden, prtdosm (MTorrent)
- Tv5#46 : prtspcur (MVerstraete)
- Tv5#50 : prtelf (ALherbier)
- Tv5#73, 74 : iprecl=75, 79 (PMAnglade)
- Tv5#75,76 : recgratio, recrcut,recefermi, rectesteg (MMancini)
- Tv5#77 : strprecon (JZwanziger)
- Tv5#79 : prtxml (DCaliste)
- Tv5#80 : tphysela (MVerstra)
- Tv5#82 : irdddk (JZwanziger)
- Tv5#90, 92, 93, 94 : for ANADDB, in the electron-phonon case,
symgkq, telphint, ep_b_min, ep_b_max, mustar, elph_fermie, ep_nqpt,
ep_qptlist, qgrid_type (MVerstraete)
- Tv5#91 : a2fsmear, dossmear (MVerstraete)
- Tv5#99 : rifcsph (MVerstraete)
- Tv6#01 and 02 : qprtrb, vprtrb (XG) - But does not work correctly ?!
- Tv6#04 and 05 : fftcache (XG)
- Tv6#14 : ndivsm, prtfsurf (MGiantomassi)
- Tv6#81 : nomegasrd, omegasrdmax, nshsigx, nshwfn (MGiantomassi)
- Tv6#82 : nomegasi, omegasimaxi, spbroad (MGiantomassi)
- Tv6#83 : gw_nqlwl, gw_qlwl, vcutgeo (MGiantomassi)
- Tv6#85 : gwgamma, irdsuscep, irdqps, irdkss, irdscr (MGiantomassi)
D.15 Use of FFTW3 library is under development (by M. Giantomassi).
D.16 The different input files for the tutorial LDA+U (tests/tutorial/Input/tldau*.in) are now automatically tested.
D.17 : QPSCGW + PAW is still under testing due to some regressions introduced in the previous merges.
one-shot COHSEX + PAW is available for production, self-consistency with COHSEX + PAW is under testing.
No automatic test yet.
D.18 Parallelism over atoms in PAW is progressing (F. Bottin and M. Torrent)
D.19 ABINIT is able to read the pseudopotential .upf format, used by PWSCF (implemented by M. Verstraete)
D.20 New example configuration files in doc/config/build_examples, from JM Beuken.
D.21 New "commodity" or development input variables : prtangst, prtxcart, prtxred, maxnsym, normpawu,
usekden, prtxml.
D.22 Numerous miscellaneous bug fixes (to the sources, as well as to the build system, including the plug-ins),
and improvements of documentation by D. Adams, B. Amadon, P.M. Anglade, J.-M. Beuken, S. Blackburn,
F. Bottin, P. Boulanger, D. Caliste, M. Cazzaniga, V. Chevrier,
M. Cote, M. Cotelo, M. Delaveau, T. Deutsch, L. Diaz, G.A. Franco, M. Giantomassi,
X. Gonze, Chuping Hu, A. Jacques, F. Jollet, G. Jomard, M. Mancini, N. Manini, G. Onida,
Y. Pouillon, T. Rangel, V. Recoules, G.-M. Rignanese, M. Stankovski, M. Verstraete, M. Torrent, D. Waroquiers, J. Zwanziger
D.23 The distribution of processors in the FFT/Band/k point parallelism can be found automatically. Work by F. Jollet.
(Auatomatic test to be provided).