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 September 2014 and May 2016. These release notes are relative to modifications/improvements of ABINITv8.0 with respect to v7.10. Moreover, most of them are also described in the Computer Physics Communications 2016 ABINIT paper, doi:10.1016/j.cpc.2016.04.003 The list of contributors includes : F. Abreu, B. Amadon, G. Antonius, F. Arnardi, L. Baguet, J.-M. Beuken, J. Bieder, A. Bokhanchuk, E. Bousquet, F.Bruneval, D. Caliste, W. Chen, M. Côté, J.F. Danel, M. Delaveau, C. Espejo, A. Gerossier, M. Giantomassi, Y. Gillet, X. Gonze, F. Jollet, J. Laflamme Janssen, R. Laasner, A. Martin, C. Martins, A. Pasquarello, V. Planes, S. Poncé, Y. Pouillon, T. Rangel, V. Recoules, F. Ricci, G.-M. Rignanese, B. Rousseau, O. Rubel, A. Shukri, M. Torrent, M. Van Setten, B. van Troeye, M. Verstraete, D. Waroquiers, J. Wiktor, Bin Xu, 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 ... Xavier __________________________________________________________________ Version 8.0, released on May 16, 2016. Changes with respect to version 7.10 : 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 AND IMPORTANT REMARKS A.1 The automatic renaming of the output file when there is already a old output file present with the same name has been modified : now the name of the old file is changed instead of the name of the new file. (From J. Bieder). A.2 The ABINIT header for many binary files has been updated, and backward compatibility has not been maintained. This is a major move forward : the code developers can assume that the new content of the header is available for their development. So, the user should not be surprised not being able to read old files produced by ABINITv7 with ABINITv8. Either use ABINITv7.10.5 (the last official version of ABINITv7) with these files, or regenerate them with ABINITv8. Mostly, work by M. Giantomassi. A.3 Cleaning of ABINIT capabilities and input variables. Some deprecated capabilities of ABINIT have been suppressed, as well as the corresponding input variables, documentation, and tests. There has been a similar cleaning of some unfinished development. The source code has been accordingly cleaned. In particular, items related to the following have been removed : the CML interfacing, the old RPA or ACFD approach to Van Der Waals energy (that was only implemented at k=Gamma, and separate from the GW part and the RPA computation of total energy), the Thomas-Fermi Von Weiszacker preconditioning, some items related to orbital magnetism, reading of KSS files, the documentation about the unsupported OpenDX software, an old methodology to compute Wannier functions (oddly present in cut3D), the ETSF_IO related input variables, the Effective Energy Technique (EET), and the LOBPCG II algorithm. Suppression of input variables : etsfgroups, etsfmain, freqsusin, freqsuslo, fft_opt_lob, getkss, gw_eet, gw_eet_inclvkb, gw_eet_nband, gw_eet_scale, idyson, ikhxc, intexact, iprctfvw, irdkss, LaticeConstant, ldgapp, nbandsus, ndyson, nfreqsus, optfreqsus, suskxcrs, chkgwcomp, orbmag, prtcs, npweps, npwsigx, npwwfn, nsheps, nshsigx, nshwfn . The following input variables are now internal : npwwfn and npwecut. Suppression of tests : v2#57, v2#78-80, v3#68 and 69 (CML), v4#10, 47, 48, 49, v5#47, v67mbpt#10 and 28 (EET), seq#tsv5_101 This work was done by many ABINIT developers. A.4 Cleaning the use of libraries The use of FoX and Libpspio has been disabled. Fox is actually superseded by PSML. The use of ETSF-IO library is deprecated (should be replaced in the long term by the ESL library). Still, the use of netcdf and of the ETSF-IO standard is maintained (and will likely spread due to the COST EUSpec efforts). Work by Y. Pouillon and M. Giantomassi A.5 Changing input variable names Some input variable names, that were not fully appropriate, have been changed, breaking backwards compatibility. In particular : sciss is now dfpt_sciss soenergy is now mbpt_sciss iprcch is now densfor_pred nloalg has been split in two input variables : nloc_alg and nloc_mem accesswff is now iomode (the previous values used to select Fortran-IO, MPI-IO and ETSF-IO are unchanged while netcdf mode has been removed. Note that the policy used for IO has been changed. See docs) rfd2dk is now rf2_dkdk rfXatpol is now d3e_pertX_atpol (for X=1, 2, 3) rfXdir is now d3e_pertX_dir (for X=1, 2, 3) rfXelfd is now d3e_pertX_elfd (for X=1, 2, 3) rfXphon is now d3e_pertX_phon (for X=1, 2, 3) Work by X. Gonze M. Torrent and M. Giantomassi A.6 Other cleaning or renumbering. All the reference output files have been updated, because the reference machine has changed (the old "testf" has been replaced by the new "abiref"), and also due to the switch from ABINITv7 to ABINITv8. Some additional information (e.g. related to magnetization) is now present in the standard output file. The enable_clib option of configure has been removed because clib is now required. Some tests in tests/v7 have been reordered and renumbered (with respect to their numbering in Computer Physics Communications (2016)) : test v7#67 has become v7#16, tests v7#114-115 have become v7#47-48. Other renumberings : tests v3#90-95 have become v3#92-97, tests v3#100-101 have become v3#90-91, tests v7#100-101 have become v7#83-84, tests v7#110-113 have become v7#85-88. The "special" directory, that contains scripts and directories to perform a series of checks of ABINIT, not in the mainstream of the automatic tests (coding rules, build systems, etc), has been renamed "abichecks". The old "psps" directory has been removed (superseded by the pseudo-dojo project). The GW lessons of the tutorial have been cleaned, the input files are now using the standard default values in general. Many automatic tests have also been cleaned, thanks to the use of the default values (e.g. for the following input variables : asr, inclvkb, iscf, ntemper, prtdensph, pawxcdev, symchi, symdynmat, tempermin, temperinc). Although they can still be generated, KSS files are not used anymore in ABINIT (especially they are not used anymore in the GW part). Work by many ABINIT developers A.7 The default writing of status file has been disabled. By X. Gonze A.8 restartxf=1 has been removed By M. Giantomassi A.9 Linking with NetCDF is strongly suggested as particular features such as the computation of the temperature-dependence of the electronic structure and other post-processing tools (e.g. AbiPy) rely on files produced in this format. Note that Abinit8, supports netcdf4 + Hdf5 + MPI-IO Users who want to produce their results in netcdf format are strongly encouraged to link against an external library providing these capabilities. See ~abinit/doc/build/config-examples/ubu_gnu_4.9_mpich.ac for a typical configuration file. By M. Giantomassi A.10 Calculations done with nprocs greater than 1 will produce a file (__ABI_MPIABORTFILE__) in the working directory with the error message of the first processor that encounters the problem. This is particularly useful for runs done on many CPUs since log files are disabled. Note that the logic used by Abinit breaks down if users run multiple instances of abinit in the same working directory. By M. Giantomassi A.11 cut3d user interface has been changed. This modification could break third-party scripts written by users. Some capabilities of cut3d have been removed : Wannier (see above), densfileformat, DX . By M. Giantomassi A.12 New defaults and different behaviour in the GW section of ABINIT. * inclvkb now defaults to 2 (was 1 previously) * awtr defaults to 1 (was 0 previously) * getkss and irdkss have been removed. Use getwkf and irdwfk to start a GW calculation from the WFK file. * prtsuscep now defaults to 0 (was 1 previously) GW calculations using a cutoff in the Coulomb interaction must set prtsuscep == 1 in the screening part so that ABINIT can recompute the screening in the sigma part starting from the SUS file (polarizabilty) with the truncated interaction. Same restriction applies when gwgamma=1 is used. * npweps, npwsigx, npwwfn are now internal variables. The basis set for GW calculations should be defined via ecuteps, ecutsigx, ecutwfn * kptopt = 0 should not be used to produce the WFK used in GW calculations. By M. Giantomassi __________________________________________________________________ B. Most noticeable achievements B.1 Implementation of the Van der Waals D3 functional proposed by Grimme (vdw_xc=6) and also the modified flavour by Becke and Johnson (vdw_xc=7). Computation of the ground state energy, force and stresses, but also the dynamical matrices at arbitrary q points (DFPT). Elastic tensor and internal strain are available as well in DFPT. There is no specific contribution to derivatives with respect to electric field of the Van der Waals D3 functional (so the computation reduces to the one of the underlying DFT functional only), so these derivatives are available as well. The dynamical matrices, elastic tensor and internal strain have also been implemented for the Van der Waals D2 functional, previously made available (vdw_xc=5) See B. Van Troeye, M. Torrent, and X. Gonze. Phys. Rev. B93, 144304 (2016) for the dynamical matrices. Another paper is in preparation, concerning the elastic tensor and internal strain. Input variables : vdw_xc , vdw_tol, vdw_tol_3bt Tests : v7#97 to 99. Work by Benoit Van Troeye, XG, and M. Torrent. B.2 The efforts related to the generation of pseudopotentials have lead to well-tested pseudopotential tables for both PAW and Norm-conserving cases (GGA-PBE), available at http://www.abinit.org/downloads/atomic-data-files . The ABINIT package v8.0 and beyond is compatibible with these new pseudopotentials. See the description of both efforts in the forthcoming paper in Computer Physics Communication ABINIT2016, a preprint is available at http://www.abinit.org/about/ABINIT16.pdf . See also K. Lejaeghere et al, Science 351 aad3000 (2016). Work by M. Torrent, F. Jollet, M. Giantomassi, M. van Setten, G.-M. Rignanese, X. Gonze. B.3 The efforts related to Dynamical Mean Field Theory (DMFT) have lead to an operational framework combined with the ab initio computation of the U value for projected local orbital Wannier functions, thanks to cRPA (see several previous release_notes of ABINITv7). A tutorial has been written, describing the DFT+DMFT calculation on SrVO3. See ~abinit/doc/tutorial/lesson_dmft.html (or http://www.abinit.org/doc/helpfiles/for-v8.0/tutorial/lesson_dmft.html) See also ~abinit/tests/tutoparal/Input/tdmft_1.in and tdmft_2.in . A new input variable dmftctqmc_basis has been introduced. ABINIT has been interfaced with TRIQS/CTHYB library (Toolbox for Research on Interacting Quantum Systems). See https://triqs.ipht.cnrs.fr/1.4/applications/cthyb/. See directory src/01_triqs_ext This library provides a rotationally invariant CTHYB solver for the Anderson model, complementary to the density-density CTHYB solver available in ABINIT. Work by B. Amadon and V. Planes B.4 Bethe-Salpeter equation : implementation of the Rohlfing-Louie approach (Phys. Rev. B62, 4927 (2000)), coupled with the Haydock recursion algorithm, and different (more complex) interpolation schemes, also coupled with the Haydock algorithm. The simple Rohlfing-Louie scheme apparently seems to perform better than the more sophisticated approaches. These different algorithms have been described in Y. Gillet, M. Giantomassi and X. Gonze, Computer Physics Communications, 203, 83 (2016). Related input variables : bs_interp_* variables, bs_interp_m3_width, bs_method, bs_interp_rl_nb. Related tests : v67mbpt#30 to 35. Work by Y. Gillet and M. Giantomassi . B.5 The new main executable fold2Bloch allows to extract, from supercell computations (that include e.g. defects), the components (and weight) of wavefunctions in the underlying small primitive cell. See O. Rubel, Phys. Rev. B 90, 115202 (2014). Tests : tutorial#tfold1bloch_1 and tfold1bloch_2 Related tutorial : lesson_fold2Bloch. Contribution by Anton Bokhanchuk (supervised by Oleg Rubel). B.6 Direct computation of the DFT effective mass tensor, in the framework of DFPT (ddk perturbation). A paper is accepted for publication in Phys. Rev. B. Work for NCPP as well as PAW. Also spin-polarized and with spin-orbit interaction (the latter only in the case of PAW). Input variables : efmas (turn on effective mass tensor calculation), efmas_bands, efmas_calc_dirs, efmas_deg, efmas_deg_tol, efmas_dim, efmas_dirs, efmas_n_dirs, efmas_ntheta Tests : v7#80 (Norm-conserving), v7#81 (PAW), v7#82 (PAW with spin-orbit).. Work by Jonathan Laflamme Janssen, Marc Torrent and XG, with contributions from Y. Gillet, S. Poncé, A. Martin . B.7 GW Lanczos-Sternheimer This new methodology to perform GW calculations is described in Phys. Rev. B 91, 125120 (2015). At present, however, it is implemented for system with only one k-point. New input variables : gwls_sternheimer_kmax gwls_kmax_complement gwls_kmax_poles gwls_kmax_analytic gwls_kmax_numeric gwls_list_proj_freq gwls_nseeds gwls_n_proj_freq gwls_recycle gwls_first_seed gwls_model_parameter gwls_second_model_parameter gwls_npt_gauss_quad gwls_dielectric_model gwls_print_debug gwls_band_index gwls_exchange gwls_correlation Related tests : paral#77, v67mbpt#15 . Work by Jonathan Laflamme Janssen, B. Rousseau and Michel Côté. B.8 Implementation of the strain perturbation in PAW+DFPT, and associated computation of elastic constants, internal strains and piezoelectric coefficients. See tests v7#95, 96 Also, associated debugging of Born effective charge and phonon calculations in DFPT+PAW. Work by A. Martin and M. Torrent. B.9 Electron positron Doppler broadening can be computed in PAW. Input variable : posdoppler Test : v7#35 Work by J. Wiktor and M. Torrent B.10 Achievements related to the treatment of the Fock operator, and hybrid functionals : Hartree-Fock is available with PAW (self-consistent, based on plane waves). However, forces and strains are not yet available. The Spencer-Alavi scheme if used for Coulomb operator Test v7#70. The HSE functional is available in the case of norm-conserving pseudopotentials, via libXC. Note that forces are available in this case (not yet the strains). The PBE0 functional is also available in the case of norm-conserving pseudopotentials (via libXC as well as via ixc=41/42). Work by C. Martins, F. Arnardi, M. Torrent and F. Jollet. B.11 Different improvements in the computation of the temperature-dependence of the electronic structure (and zero-point motion effect), as listed below. Implementation of the calculation of the spectral function and dynamical lifetimes in ElectronPhononCoupling. See G. Antonius, et al, Phys. Rev. B 92, 085137 (2015). Implementation of the correction to impose the charge neutrality sum rule also on the matrix elements of the electron-phonon, in case of IR-active insulators, see S. Poncé et al, J. Chem. Phys. 143, 102813 (2015). Also, improvement of the flow, by the possibility of separate treatment of all q wavevectors. Also, print Fan and DDW decomposition inside _EP.nc . Implementation of g2F Eliashberg function and PDOS . Creation of GKK.nc files (much better from the memory point of view) Related input variables : getddb, irdddb, wtq . Test v7#57, 58, 59 . Spin-dependent version of the temperature-dependence of the gap, tests in v7#100, 101. Parallelized electron-phonon change of electronic structure. Printing of the mode-by-mode decomposition of the Fan and DDW terms. Work by S. Poncé, G. Antonius and Y. Gillet. B.12 Implementation of inelastic variational approximation for the computation of the thermoelectricity (Seebeck effect). See tests v7#90, 91, 92, 93, 94. However, due to instabilities, the diff on the output with respect to a reference has been disabled ... Work by Bin Xu and M. Verstraete B.13 The electronic contribution to the linear electro-optic coefficients in the independent particle approximation is available, following the framework of Hughes and Sipe (Phys. Rev. B, 53:10751, 1996). Moreover, the anti-resonant contributions have been added to all second-order susceptibilities, as well as the possibility to do scissor corrected calculations. Finally the optic postprocessor has been parallelized. The tutorials tutorespfn#toptic_1 to 4 have been updated. Test for the new routines are also available in v7#47-48. Work by Y. Gillet. B.14 Stopping power ( kinetic energy lost by an impinging charged particle by unit of path length) Implemented in the linear response approximation. Input variables : npel and pvelmax. See test v7#16 (initially test v7#67). Work by A. Shukri and F. Bruneval B.15 Quasi-particle self-consistent GW calculation, with the bootstrap kernel in W, has been developed. This is described in the publication W. Chen and A. Pasquarello, PRB 92, 041115(R) (2015). Use gwgamma -4 See test v67mbpt#36 . Also the Gygi-Baldereschi auxiliary function is available (icutcoul 7). Work by Wei Chen and A. Pasquarello. B.16 Magnetic nuclear dipoles can be introduced, to allow computation of the Nuclear Magnetic Shielding tensor through the converse method (change of magnetization due to magnetic nuclear dipole). Input variable : nucdipmom Test : v7#32 Work by J. Zwanziger B.17 Many of the ABINIT post-processors are now NetCDF-ready (can use NetCDF-based files). Also, for DFPT, modules to write / read in NetCDF have been introduced (in view of high-throughput). Please note the following changes: * The `-etsf.nc` extension has been removed from the netcdf file. The new name is created by appending ".nc" to the standard Abinit name. * The HIST file has been renamed HIST.nc. * The output of OUT.nc has been disabled when ndtset > 10 * The netcdcf files: - AE_ONSITE_WFK.nc - PS_WFK.nc - AE_WFK.nc - PS_ONSITE_WFK.nc produced by pawprtwf=1 have been removed and replaced with a single netcdf file with extension _PAWAVES.nc The different PAW terms are stored in the netcdf variables: ur_ae, ur_pw, ur_ps_onsite, ur_ae_onsite Work by M. Giantomassi and D. Waroquiers.. B.18 Bigdft + PAW work in parallel Test : bigdft_paral # 01 and 02, execute on 4 and 10 procs. Issues related to fitting of Gaussians have been fixed. Work by M. Torrent and T. Rangel B.19 MPI and GPU work together. Compatibility with MAGMA 1.5 has been achieved. Test gpu#05 . New input variable : gpu_device Work by M. Torrent and J.-M. Beuken. __________________________________________________________________ C. Changes for the developers C.1. Creation of separate build systems (incl. config and makefiles) for different directories, e.g. tests, and doc . Work by Y. Pouillon . C.2 (Beautification) The whole input variable documentation is now written using yml , in the doc/input_variables/abinit_vars.yml file A java editor allows to view and modify this file. The html is automatically generated from this yml database. Group effort lead by Y. Gillet . When possible, the pointers have been replaced by allocatable. The coverage of ABINIT has been improved as well. Numerous contributors : Bernard Amadon, Gabriel Antonius, Jean-Michel Beuken, Jordan Bieder, Fabien Bruneval, Michel Côté, Muriel Delaveau, Gregory Geneste, Matteo Giantomassi, Vincent Gosselin, François Jollet, Jonathan Laflamme Janssen, Aurelien Lherbier, Alexandre Martin, Micael Oliveira, Samuel Poncé, Yann Pouillon, Gian-Marco Rignanese, Marc Torrent, Michiel Van Setten, Matthieu Verstraete, Bin Xu, Joe Zwanziger C.3 Cleaning of the arguments of routines with a very long argument list, e.g. nonlop, passed through structured datatype arguments (e.g. gs_ham). Work by M. Torrent C.4 Upgraded Wannier90 fallback to its 2.0 version and Atompaw to its 4.0.0.12 version. (Y. Pouillon and M. Torrent) C.5 Got rid of src/quarantine/ to avoid stupid conflicts (Y. Pouillon) C.6 Migration of the Web site to Plone 4. Work by JM Beuken. C.7 Creation of the LibPAW library, presently contained in src/42_libpaw It does not rely on defs_basis and abi_common.h . Can be used by BigDFT. Work by T. Rangel and M. Torrent (and D. Caliste for the BigDFT side). C.8 Preparation of the migration to Git/GitLab instead of bzr as version management system. Work by J.M. Beuken, with help by M. Giantomassi, Y. Pouillon, Y. Gillet and other core developers. C.9 Work on the memory profiling, replace the old machinery by new code. In practice : - Replace memcheck.py with abimem.py - Get rid of ABI_ALLOC_STAT_ABI, use ABI_STAT_ALLOC(arr, shape, ierr) - Remove ABI_STAT_CALLOC - Remove ABI_ALLOC_SIZE_ABI, use sizeof to support datatypes ifdef HAVE_MEM_PROFILING - Support np > 1 Work by M. Giantomassi and M. Torrent C.10 The python scripts for automatic testing are now compatible with python3 (incl. 3.5, 3.6, 3.7) Support for python < 2.6 has been removed. Work by M. Giantomassi C.11 Replaced many statement functions (obsolete) by internal functions Work by X. Gonze, F. Jollet, and others C.12 The test farm now supports a subset of F2003 (g95 had to be discarded). Work by JM Beuken and M. Giantomassi C.13 Support for ELPA 2015 has been added Work by M. Torrent. __________________________________________________________________ D. Other changes (or on-going developments, not yet finalized). D.1 Several shortcomings of the recognition of space groups, and automatic correction of inaccurate primitive lattice parameters using the tolsym input variable have been removed, so that ABINIT is now much more robust in this respect. See test v7#13 Work by X. Gonze D.2 Use of psml instead of fox (to read XML) Renaming tests/fox to tests/psml New pseudos files : 14Si.psml , 26fe.GGA.psml Work by M. Verstraete (also support by JM Beuken). D.3 The extrapolation of the wavefunction from one structural relaxation time step to the next one has been implemented. Input variable : extrapwf Test : v7#9 . Work by M. Torrent D.4 Generalisation of DFPT for non-collinear magnetic cases. Work by F. Ricci and E. Bousquet On-going. D.5 Several new input variables have been introduced, to allow and control the output of several quantities in specific files : prtgsr (for the GSR file), prtpsps (for the pseudopotential, in the netcdf format), prtvclmb (for the Coulomb potential), prtvpsp (for the local potential), prtwf_full (to generate a WFK file in the full Brillouin Zone after a run done in the irreducible wedge, useful to interface Abinit with post-processing tools that do not support symmetries.) See Tests v3#30, v6#04, v7#10, v7#12, v7#36, v7#12 and v7#36. Also, the new test mpiio#26 exercise IO routines in parallel with different combination of parameters in the GS eigensolvers while the new test mpiio#51 exercise IO routines in parallel with different combination of parameters in the DFPT eigensolvers Done mostly by M. Giantomassi and M. Verstraete D.6 Work on the distribution of MPI data (band and FFT) for the DFPT code, by M. Delaveau (supervision M. Torrent). D.7 Work on Vdw-DF. The new input variable prtvdw has been introduced, to allow storing a VdW-DF kernel after computation. See test vdwxc#03. Also, vdw_xc 14 (vdW-QHO-WF from Silvestrelli) can be used in periodic systems. Work by C. Espejo and Y. Pouillon D.8 Work on VWL+PAW, and disentangling of BigDFT and ABINIT Work by M. Torrent and D. Caliste. D.9 Implementation of the Beigi treatment of q-> 0 for surfaces Requires additional testing Work by Y. Gillet D.10 Implement Quantum Thermal Bath (QTB) formalism in Langevin PIMD Corresponds to imgmov=10 (to be documented, to be tested). Work by M. Torrent D.11 Implementation of prtdos 5 . This allows to output the spin-spin DOS in the nspinor=2 case. No test yet. Work by M. Verstraete D.12 Work on ABINIT GUI by F. Abreu and Y. Gillet D.13 Update of the tutorial map in doc/tutorial/welcome.html D.14 Work in progress on the Implementation of Raman intensity computation in PAW. Associated input variable rf2_dkdk By L. Baguet and M. Torrent. D.15 Work in progress, concerning the capability to read DFPT potentials, in order to compute the phonon linewidths and the electron-phonon coupling parameters in a fine grid. Activated using optdriver=7. Some new input variables have been introduced, but also many input variables of anaddb have to be "reused" inside the main ABINIT for that purpose, and have thus been introduced (possibly with a prefix) and documented : asr, chneut, ddb_ngqpt, dipdip, eph_intmeth, eph_extrael, eph_fermie, eph_fsmear, eph_fsewin, eph_mustar, eph_ngqpt_fine, ph_intmeth, ph_ndivsm, ph_ngqpt, ph_nqpath, ph_nqshift, ph_qpath, ph_qshift, ph_smear, ph_wstep, symdynmat. See the tests v7#85 to 88. Work done by M. Giantomassi. D.16 Different treatments are now possible for the invalid frequencies possibly obtained with the GW Hybertsen-Louie plasmon pole model. See input variable gw_invalid_freq, and test v67mbpt#41 . From T. Rangel D.17 The non-linear core correction can now be treated in G-space, which might be more accurate than the real-space treatment, depending on the pseudopotential. See input variable nc_xccc_gspace, and test v7#10, 11 and 45. From M. Giantomassi D.18 A new criterion for stopping the geometry optimization (ntime loop) has been introduced : the maximum change of energy, with input variable tolmxde. See test v7#15 . Work from D. Waroquiers D.19 Work is continuing at the level of high-temperature DFT computations with orbitals, based on Thomas-Fermi approach (and corrections). See the new test v7#11, and input variables tfkinfunc, tfw_toldfe, ixc=50. Work by JF Danel and M. Torrent. D.20 The merge utility mrgscr has now the capability of q point recovery. See test v3#100 and 101. Work by M. Giantomassi . D.21 mrgscr can now merge partial screening files produced with different frequencies. See test v67mbpt#37, 38, 39. Work by M. Giantomassi D.22 GW computations with istwfk /=1 , as well as using norm-conserving pseudopotentials with multiple projectors can now be done. inclvkb == 2 is not supported yet. See test v67mbpt#40 Work by M. Giantomassi D.23 Implemented capability to check MD5 sum for pseudopotentials. Work by M. Giantomassi and Y. Pouillon D.24 APPA library : minor improvements from A. Martin. D.25 Anaddb utility : frequencies with non analytical contributions added to anaddb.nc, by G. Petretto D.26 A new value for prtvclmb is introduced. prtvclmb=2. By F. Bruneval. D.27 In the case of different images of the original system (e.g. string method, path-integral molecular dynamics, etc ...), several quantities related to LDA+U can be different in different images : the "U" value, as well as the occupation matrix (dmatpawu). Work by M. Torrent. D.28 A new value for prtwf is introduced: prtwf = -1 Abinit won't write the WFK (1WFK) file if prtwf == -1 and the calculation is converged. Implemented in the GS and in the DFPT part. From M. Giantomassi D.29 New command line option added to abinit: --timelimit days-hours-minutes-seconds The code will exit the iterative algorithm if the calculation cannot be completed within the given time limit. Available for GS, structural relaxations, and DFPT with one perturbation. This option should not be used in conjunction with multiple datasets. Use `abinit --help` to get the list of supported options. From M. Giantomassi D.30 New check on the correct editing of the doc/*/*yml documents (forbidding \n as carriage return) has been introduced. See test abirules#19 . From X. Gonze D.31 Introduce miniBZ technique in GW, to sample the Brillouin zone around the divergence From Fabien Bruneval D.32 Numerous miscellaneous additional bug fixes (to the sources, as well as to the build system, including patches for the fallbacks), and improvements of documentation by : G. Antonius, F. Bruneval, D. Caliste, M. Delaveau, C. Espejo M. Giantomassi, Y. Gillet, Raul Laasner (on the forum), Y. Pouillon, V. Recoules, M. Verstraete, M. Torrent, Bin Xu