ソフトウェア

https://launchpad.net/siesta/4.0/4.0.2

Release notes
The 4.0.2 version is a bug-fix release for the 4.0 release series.

BACKWARD COMPATIBILITY ISSUES

This release increases the size of the internal tables for two-center integrals used in
some matrix element calculations. This means that calculations are slightly more heavy,
but the accuracy is also superior. One can regain the *old* less accurate behaviour by
setting Compat.Matel.NRTAB to true in the fdf input file
(this is ONLY recommended for testing purposes).

Below we list the itemized changes in this release.

* Enabled ion.nc files for ghost atoms (#1738425)

* Enabled ghost atoms to read ion.nc files (#1736455)

* Forced libfdf to “die” when too long input strings are passed (#1728281)

* Monkhorst-Pack grids not properly shifted to [0;1[ when user specified large displacements (#1721479)

* Updated README content in Util directory (#1712319)

* Fixed building all utilities (#1712317)

* Fixed EIG file format for non-collinear spin (#1708634)

* Fixed possible segfault when using too large unit-cells (#1704370)

* Fixed ghost orbital energies (#1695130)

* Removed vpsa2bin and vpsb2asc codes (part of ATOM)

* Added new units in fdf (Hartree [Ha] and milli-Hartree [mHa])

* Added more tests

* Fixed a missing close when using Siesta in “master” mode

* Fixed several memory accounting errors and a couple of missing close
statements. Also removed all deprecated PASTE calls.

* Added compiler version information to the version.F90 file

* Fixed writing the Bessel ghost atom to ion.xml/nc files

* Extended information in PDOS xml files. Now the atomic number
as well as whether the orbital is a polarization shell is present

* Increased the default size of two-center integrals tabulation arrays from 128 to 1024.
This is a change that results in more accurate values (of basically everything). Set
‘Compat.Matel.NRTAB true’ to use previous value

* Added spin-monitoring in the SCF, for spin-polarized and non-collinear calculations
the total spin-moment is written

* Fixed non-collinear Mulliken populations in parallel

* Updated gnubands to the new code that has been present since the 4.0 release.
The new updated gnubands provides more functionalities

* Extended precision output in EIG, KP and PDOS files

* Fixed possible non-optimal DM initialization when unit-cell folding of orbital
connections is performed.

* Bugfix for writing out too many wavefunction coefficients,
see https://www.mail-archive.com/siesta-l@uam.es/msg10291.html