GRASP0 UPDATE LOG


30/06/21 v9.9 cpb
COULG did not re-initialize L-spinor flag to .false. for l>0 for any 
subsequent (low-n) non-L-spinors. Note, low-n L-spinors are a good
approx for non-physical.

27/05/11 v9.8 nrb
OSCL/PRINTA: Test for zero f-value -> divide by zero len/vel.

04/03/11 v9.8 cpb
Adjust format statements to accommodate >999 levels, i.e. I3 ->I4.

19/10/09 v9.7 nrb
Minor.

09/10/09 v9.7 cpb
OSCL  4 gives magnetic sub-level resolution of radiative data.

25/09/09 v9.6 nrb
If no pre-existing MCP.DAT (Dirac-Coulomb Hamiltonian angular algebra)
file and no such file computed (ANG 8) then carry-out a pure SCF run.
E/AL are the only meaningful optimization options then.

23/09/09 v9.6 nrb
Tweaked Schmidt orthogonalization for safety with DARC.

18/09/08 v9.5 nrb
COUVEC and BREVEC not fully initialized (PHN).

17/09/08 v9.5 nrb
Partition diagonalization of H by Jp sub-blocks. Original code diagonalized
the entire H giving rise to small mixing coefficients between states of 
different Jp, particularly problematic for H-like ions due to energy
degeneracy. In particular, M_k radiative rates were rubbish if E_k were
computed at the same time.

19/05/08 v9.4 nrb
RRITZ got bypassed for s-states when non-zero ang. mom. present.

17/05/08 v9.4 nrb
Minor re-organization and tidy-up.

15/05/08 v9.4 nrb
Test for cases where L-spinor evaluation requested at too high an n-value.
(Numerical inaccuracy/failure.)

14/05/08 v9.4 nrb
For L-spinor (+CI) operation, initialize average occupation array (UCF)
to that of the first input configuration (which should be ground) for
subsequent default use by DARC to form static potential for continuum basis.
Original default was the average of all configs. With a large L-spinor 
basis, basically this leads to a highly-excited average config, which,
in turn, causes FINDER grief. Can/could be over-ruled by user specifying
(via OPT=43) explicit UCF in STG1D_ORB, but not really necessary as
easiest to default as above, then user is none-the-wiser.

18/03/08 v9.4 nrb
Major:
RRITZ generates the new L-spinor basis from the zero-order, 
c.f. Grant & Quiney (2000), called from INIT.
Minor:
DATSCF, allow read of FX, ISTATX, NPMIN0 from Record 11 (Z, RNT, H etc)
FX is factor for extending radial array - buffer space for RRITZ - default, 3.
ISTATX .ne. 0 uses static potential in RRITZ. Default: 0.
NPMIN0, include states from this princ. qu. no. in RRITZ. Default: all L-spinor.

30/01/08 v9.3 nrb
COULG: L-spinor large/small component energy norm sqrt(1 +/- E/c**2) 
is only correct to O(1/c**2). This is evident on examining Q(0)/P(0),
the second normalization condition (Int P**2 + Q**2 dr trivially o.k.)
The correct form is readily derived from analytic value of Q(0)/P(0).
Ratio was in error by up to 20% by Z=92. Correction has very little
effect on final energies though.
PROPG: Schmidt orthogonalize the power series coefficients (not needed
by one-electron problem).

17/01/08  v9.2 nrb
Re-worked PROPG for single pass orthogonalization

09/01/08  v9.1 nrb
Modified to generate (zero-order) L-spinor basis.
COULG, PROPG, ZEFR

Revision 8.5  2000/05/24 14:23:45  phn
*** empty log message ***

Revision 8.4  1999/06/04 23:05:24  phn
Improved output from BENA and OSCL.
OSCL option 8 gives a good debug print now.
OSCL prints line-strengths.

Revision 8.3  1999/04/29 15:10:58  phn
Removed INCLUDE statement for calen/quartz

Revision 8.2  1999/04/14 13:12:26  phn
Uses IMPLICIT NONE

Revision 8.1  1999/03/02 23:46:13  norrington
1. Dimensions are set by grasp0.inc
2. CALEN and QUARTZ are now included
   with names calen.for and quartz.for

Revision 7.8  1998/11/01 02:06:31  norrington
Changed MATOUT routine.

Revision 7.7  1998/05/19 14:09:46  norrington
Fixed bug in mak file.

Revision 7.6  1998/05/19 14:08:30  norrington
Used latest version of ftidy.

Revision 7.5  1998/04/30 11:37:12  norrington
1. added nearly all common blocks to routine AA
2. some variables changed to make the common blocks unique
3. IPRWF now writes to file: GRASP.SH (routine PRWF)

Revision 7.4  1998/04/24 12:43:37  norrington
Corrected bugs in COUMT2 and MATRIX.

Revision 7.3  1998/04/24 10:45:51  norrington
CRAY preprocessing now uses single precision LAPACK.

Revision 7.2  1998/04/24 10:33:25  norrington
1. PRWF now calls IN2CH3 rather than PRWFNM (now removed)
2. The index file is now called INDEX.

Revision 7.1  1998/04/07 12:36:54  norrington
*** empty log message ***