lp:~albertog/siesta/legacy-ldau
This branch contains an abandoned line of development focused on implementing the LDA+U functionality using experimental ideas for the expanded radial-function indexing needed to account for the LDA+U projectors.
In the end the new ideas were found not completely appropriate.
- Get this branch:
- bzr branch lp:~albertog/siesta/legacy-ldau
Branch merges
Branch information
Recent revisions
- 311. By Alberto Garcia
-
Stop if user-basis* is used. Proper inclusion of NETCDF_INCFLAGS.
* The code is not prepared to handle .ion or .ion.nc files. Now a
message is printed and the program stops.* The variable NETCDF_INCFLAGS was not properly handled in
Src/Makefile.modified:
Src/Makefile
Src/atom.f90
Src/atom_basis_io. F90 - 310. By Daniel Sanchez-Portal
-
Fix the calculation of lmxldaupj in reldauproj
Variable lmxldaupj was not well determined for a particular format of
the LDAU-bock input. As a consequence, LDAU projectors were not used
in some cases (in principle, quite randomly, but in practice this bug only
showed up when the requested projectors had a large l, like "f" projectors.modified:
Src/atom_basis_specs. f90 - 309. By Alberto Garcia
-
Trapping of out-of-bounds errors if nuo=0 in some nodes
* Modifications made to prevent an out of bounds error in diagg,
diagk, and diagk_file in parallel when nuo = 0 on some nodes. Reflects
changes in main trunk.* The call to optical in siesta_analysis has been modified to remove
the specific array element reference for Eold and Dold since this also
causes a similar out of bounds reference.(Julian Gale)
modified:
Src/diagg.F
Src/diagk.F
Src/diagk_file.F
Src/siesta_analysis. F - 308. By Alberto Garcia
-
Tentative fix for DM reuse bug
In some cases the re-use of the DM after a geometry change led to
large values of ddmax and instabilities in SCF convergence. The
tentative cause was the re-allocation of Dscf right after the call to
hsparse for the new geometry, and before new_dm, combined with a
"sparsity pattern change" in-place. When the re-allocation is deferred
to new_dm, and the plain "change_sparsity" routine is used, the
problem seems to disappear.Thanks to Oswaldo Dieguez for his sleuthing work.
More testing is needed.
Modified:
Src/Makefile
Src/new_dm.F
Src/state_init.F - 307. By Daniel Sanchez-Portal
-
Subroutine xc_check in atom.f90 modified
so the program does not stops when functionals
other than PBE or CA are requested.Files:
Src/atom.f90 - 306. By Daniel Sanchez-Portal
-
Subroutine xc_check in atom.f90 modified
so the program does not stops when functionals
other than PBE or CA are requested.Files:
Src/atom.f90 - 305. By Daniel Sanchez-Portal
-
A few input examples added to directory Examples/ldau,
and the corresponding pseudos to Examples/Vps.
FeO_shift.fdf is an example of an input file
to calculate the value of U from the variation
of the local occupationadded:
Examples/ldau/Fe_ ldau.fdf
Examples/ldau/FeO_ shift.fdf
Examples/Vps/O_opt. psf
Examples/Vps/Fe_ shift.psf - 304. By Daniel Sanchez-Portal
-
A few input examples added to directory Examples/ldau,
and the corresponding pseudos to Examples/Vps.
FeO_shift.fdf is an example of an input file
to calculate the value of U from the variation
of the local occupationadded:
Examples/ldau/Fe_ ldau.fdf
Examples/ldau/FeO_ shift.fdf
Examples/Vps/O_opt. psf
Examples/Vps/Fe_ shift.psf - 303. By Daniel Sanchez-Portal
-
Sorry, version info not modified in the last commit
modified:
version.info - 302. By Daniel Sanchez-Portal
-
Default value of LDAU.CutoffNorm changed. Miscelanea of small changes
* Default value of LDAU.CutoffNorm increased from 0.85 to 0.90
* Small correction in the format of write stament in subroutine write_subs
* Small change in lin_rad_copy to make a quick and dirty cure of the problem
found for the forces due to the undefined gradient of the s-orbital near the
origin due to inadequate boundary conditions for spline interpolation near 0* Small correction in the format of write stament in subroutine write_subs
modified:
Src/ldauproj_gen.F
Src/radial_lin.F90
Src/write_subs.F
Docs/siesta.tex
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)
- Stacked on:
- lp:siesta