2006-01-27    <wdpgaara@lcmxaa.lc.ehu.es>

	* Added RPBE and revPBE exchange and correlation functionals.
          Updated manual. Put complete version (now 3.2.2) in header
	  and wrapup routines.

2002-09  Alberto Garcia  <wdpgaara@lcpxlz.lc.ehu.es>

	* Imported xc.f file from Siesta distribution. It fixes some
	problems with very small densities. It is the default in the
	makefile, but the old suite of xc files can be used, if
	desired. Pzxc.f has been fixed to avoid computing vxc for very
	small densities, but the fix in xc.f (by J.M. Soler) is in 
	principle better. Tagged as atom-3-2-1.

Thu Jul 25 15:48:48 2002  Alberto Garcia  <wdpgaara@lcpxbi.wm.lc.ehu.es>

	* Consolidated new version by removing old material from Samples.
	  Old pswatch program removed from Util.

2002-07-12  Alberto Garcia  <wdpgaara@lg.ehu.es>

	* Major revision with a view to ease analysis of pseudopotentials
	and streamline testing. Wrote a manual (Docs/atom.tex). 
	Incompatible changes:
	 - The "energy differences" section is now printed only after a
	   homogeneous series of calculations.
	 - The input file can contain comments, directives, and blank
           lines at the beginning and at the end.
	 - PSCHARGE file contains the true pseudo valence charge, not
	   the rescaled value.
	Other changes:
	 - Bug fixes for XC polarized routines (avoid crashes)
	 - Bug fix in tm2 in calculation of norm (negligible effect)	
	 - New implementation of "valence charge modify"
	 - More output files, including screened pseudopotentials.
	 - More plotting scripts (mainly gnuplot as lowest common
	   denominator)
	 - Output file rigged for easier extraction of data.
	 - Tutorial directory created (work in progress)
	 - ae.sh pg.sh pt.sh scripts re-designed.

Sun Sep 16 19:46:05 2001  Alberto Garcia  <wdpgaara@lg.ehu.es>

	* Pseudopotential files are now generated in both unformatted
	( VPSOUT) and formatted (VPSFMT) versions.

2000-04-10    <wdpgaara@node1.wm.lc.ehu.es>

	* Changed once more the management of the "compatibility
	strings". Went back to the tried and true "symbols" approach. Now
	it is possible to 

	%define COMPAT_UCB

	to revert back to the old parameters, and to change the GGA
	vs. LDA default for the core-correction scheme by using either of

	%define NEW_CC
	%define OLD_CC

	Introduced a version string, and set it to 3.1.

2000-02-21    <wdpgaara@node1.wm.lc.ehu.es>

	* Set the spacing to 0.01 in the real-space wavefunction plotting
	routines. Some nodes were not seen in the plots... 
	Added a few not very sophisticated GNUPLOT drivers in the Samples
	directory, for those without SM.

2000-02-16    <wdpgaara@node1.wm.lc.ehu.es>

	* Changed the defaults regarding the CC schemes. The old method is
	the default for LDA calculations, and the new one is used only for
	GGA calculations. This behavior can be overriden by specifying
	"mons-old-cc" or "mons-new-cc" as compatibility string. (See
	updated User.Guide.)

	* Implemented calculation of the Fourier transform of the
	pseudocore charge.

2000-02-10    <wdpgaara@node1.wm.lc.ehu.es>

	* Implemented new scheme for the calculation of the pseudocore
	charge, using a routine from Jose Luis Martins. The new method
	results in a larger pseudocore, so it is not clear whether to
	enable it by default. For now, it is indeed the default, unless
	the "compatibility string" (see compat_params.f) is set to
	"mons-old-cc" or "ucb".

	* Added a new macro "core" for the plotting of core charges in
	atom.sm. 

	* Note that the initializations mentioned in the next entry
	(Nov99) were actually done in a "real" subdirectory Pseudo/atom of
	the siesta cvs tree, instead of in the symbolic link to atom. They
	are merged now.

1999-11-19    <wdpgaara@node1.wm.lc.ehu.es>

	* Initialized variables (ddelta, fdold, ehart) in tm2.f, v0pp.f,
	and velect.f

Fri Nov 27 15:41:46 1998  Alberto Garcia  <wdpgaara@lcpxbi.wm.lc.ehu.es>

	* (Maider Machado) Added support for Becke-Lee-Yang-Parr exchange
	and correlation flavor. It is requested by the string "bl" in the
	input file.

Fri May 23 10:13:55 1997  Alberto Garcia  <wdpgaara@lcpxbi.wm.lc.ehu.es>

	* Put the Berkeley RCS sources under CVS control, and tagged them
	as RCSfiles.

	* Added some extra files to synch the module to ATM_1_0. Tagged as
	ATOM_1_0. Still no BHS, Contrib, Docs, or Utils subdirectories.
	The presence of Id and Log keyword flags made it appear as if
	there have been changes in every file...
	
	* Made some cosmetic changes (mainly use of 'unknown' as opening
	status for files and explicit setting of function values before
	exceptional returns). Added BHS, Contrib, and Docs. 
	Tagged as ATOM_1_1.

	* Implemented an interface to the Balbas-Soler XC package, and
	changed the default grid generation procedure. The old Berkeley
	behavior can be recovered by specifying the 'ucb' compatibility
	mode (implemented through the 'compat' files). Added an improved
	Util (including a ghostwatcher program from the UAM) and a Samples
	directory with execution and plotting scripts. Tagged as ATOM_1_2
	
************************
	The following changes were done at Berkeley starting from
	the Troullier-Martins code (actually, from a heavily modified
	version of it with declared variables, include files, and better
	structure).

Tue Jul 28 12:56:44 1992  Alberto Garcia  (alberto at savanna)

	* Modified the decision tree in atm.f dealing with
	charge and potential initialization for multiple runs.
	(Tip from somebody in Stuttgart via Christian Elsasser).

Thu May 21 16:58:41 1992  Alberto Garcia  (alberto at savanna)

	* Modified dsolv2. The core charge was always computed
	non-relativistically!.

Mon May 18 10:57:02 1992  Alberto Garcia  (alberto at savanna)

	* Modified dsolv2 to compute the core charge density in
	all cases (not only for core corr ps generation). Wrote	
	subroutine denplot to process the core and valence charge
	resulting from an all-electron calculation.

Wed Apr 22 20:23:33 1992  Alberto Garcia  (alberto at savanna)

	* Tentatively removed the code snippet in wrapup that
	  turns off relativity.

	* The logarithmic derivative of the pseudo-wave function is
	now computed in routine wrapup. A separate pseudopotential
	test is not required. (This is experimental).
	Logder now accepts character*2 flag (current values: 'AE', 'PS')
	to write the information to the appropriate file.
	
	Relativistic calculations are tricky: The pseudopotential is
	treated in a non-relativistic manner in wrapup, and that results
	in eigenvalue mismatches that in turn produce errors in the
	logarithmic derivative comparison.








	
