Molpro version 2002.8 release notes from patch history of preceding version

27 Feb 2003 nec_parse.2002.7
Set up get_parse to recognize a range of sx/nec-superux platforms.

27 Feb 2003 dtraf_pgk.2002.7
Some minor corrections in drv2el_pgk

27 Feb 2003 batchs1.2002.7
Tuned new version of basis_integral_batchs.f
New kernel integral routine basis_shell_Jn (to be replaced later).
Correction for timing in dfexch.f

28 Feb 2003 parse_express2.2002.7
Bug(s) fixed: 476
stop some special input fields being run through express in the call to value

28 Feb 2003 configure_tru64.2002.7
Protect TRU64 /bin/sh from eval'ing ^ in regexp.

2 Mar 2003 dfit_restart.2002.7
To get restart with density fitting working, the following changes were made:
1.) in basis_data_dump and basis_data_load extend loops to include orbital basis
2.) add basis_data_dump in aoint.f after basis_ao_initialize
3.) add basis_ao_initialize at the end of basis_data_load
4.) call basis_data_load before density fitting calculation in scfini and dftrans_loc
5.) Some other fixes in basis_data_dump/load:
save of ioff missing, use of fixed character length variable in csave.

2 Mar 2003 dfit_defbas5.2002.7
Implements new subroutine load_basis, which loads all basis sets
for density fitting calculations. This can load basis sets directly
from the library, and so specification of fitting basis sets in
a basis block is no longer necessary (unless special input is needed)
Group command in basis block eliminated!
dfit_defbas3: fix mpp case. Also fix bug in basconv/binput.f (minl not broadcasted).
dfit_defbas4: fixes for hp. Undefined basis data must be nullified and skipped.
dfit_defbas5: further fix for hp. Set bases(iset)%n_attributes=0

3 Mar 2003 nohf.2002.7
Implements primitive version of hf optimization of non-orthogonal local orbitals

16 Mar 2003 rs2grad5.2002.7
RS2 gradients now fully working with core, inactive, and satet-averaged.
Add new testjobs for rs2 gradients.
Clean up print in cicon, musamc, and cirs2grad
Save multi occupations and state information in variables.
cpmcscf,rs2 input no longer needed for rs2 gradients
Fixes missing call to citime in cporb1

17 Mar 2003 dfti_mpp.2002.7
Bug(s) fixed: 483
Move reservation of grid data structures into replicated code block

20 Mar 2003 input_cvb.2002.7
Bug(s) fixed: 490
Removed redundant code causing problems with the most recent Sun
compiler.

21 Mar 2003 hfpol2.2002.7
Analytical polarizabilities for closed-shell Hartree-Fock
hfpol2: correct syntax error in cphf_general.f

21 Mar 2003 slater.2002.7
Initial implementation of slater integral package

25 Mar 2003 prepin_geom.2002.7
Bug(s) fixed: 492
Fix prepin to get geom={...} working.
Previously at least geome={...} was required.

25 Mar 2003 newJ.2002.7
Faster Coulomb integrals.

26 Mar 2003 repulsion_f.2002.7
Corrected Boy's function for some cases.

27 Mar 2003 pcmgrd.2002.7
Routines for computing gradient of potential on a grid.
Should be modified for cosmo gradients.

27 Mar 2003 lmp2_mpp.2002.7
Adds new parameter mppfac to affect switching between mpptype=1 and
mpptype=2 depending on the number of processors. With mppfac=1
automatic switching is disabled.

27 Mar 2003 mxv.2002.7
Bug(s) fixed: 491
Don't use MXMA and MXMB for matrix-vector multiplications. Converted
matrix-vector calls to use MXVA and MXVB.

28 Mar 2003 vxv.2002.7
Bug(s) fixed: 491
Don't use MXMA and MXMB for vector-vector multiplications. Converted
vector-vector calls to use new VXVA and VXVB subroutines.

2 Apr 2003 fujitsu_conf.2002.7
Lower optimization for Molcas/cmpcts.f for correct compilation on VPP.

3 Apr 2003 defbas.2002.7
Generalization of defbas.

3 Apr 2003 mxm_fujitsu.2002.7
Bug(s) fixed: 491
Further checks for illegal DGEMM input parameters for unix-fujitsu;
new environment variable, MOLPRO_MINDGF, to control calling of scalar
code for small matrices.

3 Apr 2003 defbas_binput.2002.7
binput and defbas in new form

5 Apr 2003 defbas_binput3.2002.7
Changes defbas, basconv, and load_basis to new structure.

5 Apr 2003 scfconv3.2002.7
Fixes convergence test in konvor (scfpr2.f)

5 Apr 2003 lsdflcc7.2002.7
implements linear scaling 4ext integrals in the context of DFLCC

5 Apr 2003 defbas_binput5.2002.7
Removes some test print
Sets default for in_attributes, in_type to blank

7 Apr 2003 thrden5.2002.7
Modifies energy thresholds. New threshold edens for energy per electron.
If given in input, old threshold energy has preference. By default,
edens only used for more than nedens electrons (nedens define in registry
file to be 50).

7 Apr 2003 procedures.2002.7
Implement new input structure using procedures
energy, gradient, optimize, hessian, and extrapolate.
Numerical gradients and hessian not finished.
So far only extrapolation using L^{-3}.

8 Apr 2003 glypep2_lmp2.2002.7
Add energy threshold to glypep2_lmp2_direct.test to avoid errors due
to new thresholds

8 Apr 2003 dflmp2g.2002.7
DF-LMP2 gradient, batched version

2003-04-08 165 2003 patchlevel_molver.2002.7
Molver side of patchlevel: shell script errors fixed

9 Apr 2003 restart4ext.2002.7
restart from 3-ext integral set possible (if 4-ext transfomation failed for some reason)

9 Apr 2003 drivers2.2002.7
Makes a driver for scfpro and separates this from driver.f.
First step in attempt to make driver.f more modular

2003-04-10 421 2003 molcas_info.2002.7
Bug(s) fixed: 485
molcas_info not set in src/util/aoint.f which caused c2h4_ksfreq
testjob to crash. Put all instances of setting molcas_info into
an include file.

11 Apr 2003 defbas_binput6.2002.7
Changes defbas, basconv to implement:
(a) both ' ' and '*' act as wild-card inputs for context, type, attributes
in basconv. If '*', then the actual values for defbas are returned, otherwise
with either ' ' or an actual value, the parameters are read-only.
(b) multiple contexts are specified in defbas. The programs that call basconv
should declare the single context that they actually are, and basconv will
find a match. If the program doesn't care about contexts (unlikely) a
wild-card can be used.

11 Apr 2003 grad_batch2.2002.7
New routines for doing batched gradients in the DF-LMP2 project.
The batcher is currently incomplete in that it will eventually
provide a gradient contribution and a batch of integrals -- this
restriction is specified in the documentation. +mends the mess
of grad_batch, which trashed the integral batcher.

12 Apr 2003 mp2_pol.2002.7
Preparation for mp2 polarizabilities

2003-04-14 213 2003 drivers_aix.2002.7
compilation of src/util/drv_scf.f failed on AIX because graddef (line 81) was not defined

2003-04-14 217 2003 basis_gradient_batchs_aix.2002.7
kcont not defined in src/basis/basis_gradient_batchs.f

15 Apr 2003 oldfork_cleanup.2002.7
Removed all references to conditional `oldfork' code.

25 Apr 2003 sigprof.2002.7
Don't handle SIGPROF signals.

28 Apr 2003 enest_dummy.2002.7
Ignore dummy atoms when computing enest

29 Apr 2003 defbas_extra_free_format.2002.7
Properly implement free format for extra lines in lib/defbas

29 Apr 2003 aug_cc_pv6z_corr.2002.7
Changed sorting order of primitives for He in aug-cc-pv6z for orbitals p to h

30 Apr 2003 dflmp2g_2.2002.7
New gradient batcher for DF-LMP2 gradients.
New GPT routine for use in the batcher.

30 Apr 2003 dft_orbital_hi.2002.7
Bug(s) fixed: 504
The DFT code is broken for h and i orbitals. This patch traps such
basis sets with a crash, rather than silently producing bad numbers.

1 May 2003 local_fulldom.2002.7
Fixes bug for case of full domains in local correlation

1 May 2003 enepart.2002.7
Slight modification of enepart. Print and save min and max distances

1 May 2003 putxml3.2002.7
Implement putxml command, which allows the storage of Molpro variables
into XML files.

2 May 2003 myriwrapper.2002.7
With the latest version of Myrinet's MPI library (mpich-1.2.5..9)
mpirun stopped to accept a combination of -np and -pg options. This
patch modifies molpro.c to avoid this combination. I also did
some myrinet related cleanup in configure and fixed a problem
with MPI job startup on systems with multiple network interfaces.

3 May 2003 cckint_print.2002.7
Print message NPASS IN TRSOR3 on iout rather than unit 6, so print is
not repreatedly seen in numerical geometry optimization

3 May 2003 extrapol_variable2.2002.7
Implements basis extrapolation using variables

5 May 2003 bench.2002.7
Adds energy thresholds to normal_dft and big_normal_ccsd to obtain same
results as before change of convergence thresholds.

2003-05-05 235 2003 molden_orbital_normalization.2002.7
Bug(s) fixed: 507
Correct orbital normalization factors for put,molden which were wrong for some
d and f functions

7 May 2003 linuxconf3.2002.7
Adds possibility to select blas and lapack libraries in linux.conf.
Gives atlas blas highest prio
Allow shared libraries (mkl)

8 May 2003 linuxconf3a.2002.7
Changes the name of variable libdir in linux.conf to libdirectory to avoid
the interference with the identically named variable in ./configure.

2003-05-08 170 2003 linuxconf3b.2002.7
disables questions when batch mode specified

8 May 2003 linuxconf3c.2002.7
Added Debian locations for ATLAS libraries.

9 May 2003 grad_unini.2002.7
Corrects bug and removes undefined variable in gradient batcher.

9 May 2003 explicit.2002.7
Fixes up a memory leak in the basis_eri test code, and puts down some more code for
DF-MP2-R12.

2003-05-11 157 2003 molverF.2002.7
Recognize .F files in develop directory for molver.f

13 May 2003 alpha_linux_J.2002.7
Bug(s) fixed: 495
Fixes crashing density fitting testjobs on alpha linux.

13 May 2003 blastest.2002.7
Fixes undefined variable ntrans in blastest.f, causing failure of mpptune/tuning

14 May 2003 dflmp2g_batched_8.2002.7
DF-LMP2 gradient, batched version. Blockwise bucketsort of A(i,r,P) to avoid paging
or discontinuous readback in later contraction steps involving this quantity,
3: prescreening bug fixed
4: more memory for direct fock matrix evaluation in mp2dip.f
5,6: some fixes
7: further fixes

2003-05-21 215 2003 enepart_variable.2002.7
Bug(s) fixed: 513
adding undefined variables to lib/variable.registry for
src/local/enepart.f

21 May 2003 dfsortorb.2002.7
change in dfsortorb, # orbitals now is SubR parameter

22 May 2003 rs2grad_fix3.2002.7
Fix in krafta to avoid attempt to read hf orbitals in rs2 case (crash if
no hf done). Print correct state for expectation values and gradients

22 May 2003 eom1.2002.7
1) correct nonlocal and local simulation restart, now restart from
wf that has different number of strong pairs is possible (parameter DYND=1)
2) different domains for singles and diagonal doubles corrected (changes
in eomloc.f, cosmetic in cckext.f)
3) varia added for true local program:
a) domains for Y_1(kj), procedures for local Y/Z_n(kj), y(ijk) etc.
b) added EOM specific K-op domains
c) new smallwe parameter allows to blow up only really big excitations
d) allcopy - an easy way to use the same domains in EOM as in CCSD

23 May 2003 eom1cor2.2002.7
Bug(s) fixed: 515,516
add logical first in 2 places in opdom_eom;
fix incorrect loop format in extract_loc, unitialized first in opdom_eom, and
format in diagofa_loc;
add printing control flags and remove some excessive printing

24 May 2003 patcher_printf.2002.7
Print help message with a printf per line.

26 May 2003 dflmp2g_3.2002.7
New gradient batcher for the DF-LMP2 gradients. Now produces gradient contributions
and integrals simultaneously. Updated gradient GPT code. Minor change to LMP2
gradients code to use new batcher.

27 May 2003 opdom_eom2.2002.7
Adds missing locgical declarations in opdom_eom.f

29 May 2003 uninis_eom.2002.7
correct some unitialized variables

29 May 2003 potint.2002.7
Fixes pot_nuc not to devide by numerical zero

29 May 2003 modelopt.2002.7
Bug(s) fixed: 508
When generating new Z-matrix coordinates, add the back transformation
of the step to the old coordinates, rather than doing a back
transformation of the new coordinates. This stops coordinates with
zero Hessian eigenvalues being clobbered.

2003-05-30 392 2003 displace.2002.7
Bug(s) fixed: 511
Improved model potential function in displace which selects symmetric
internal coordinates for geometry optimization. Previously there was
a problem with atoms of the same charge in inequivalent positions, leading
to wrongfully omitted coordinates.

2003-05-30 275 2003 zmatrix_overflow2.2002.7
Bug(s) fixed: 505
Trap overflow of zsymbol in call to gets.
IBM compiler did not like call of len(zsymbol(k,zncen)) in write statement.

2003-05-30 258 2003 displace_testjob.2002.7
Bug(s) fixed: 517
Updating energy threshold for h2o_numfreq testjob to take account of improvements due to patch displace.

11 Jun 2003 basis3.2002.7
Further attempt to fix hp crashes due to undefined bases(0)%name

11 Jun 2003 qsdpath4.2002.7
Various fixes for qsd reaction path optimization, in particular correct setting
of variables after program steps. New system variables qsdstep and qsdirc.

11 Jun 2003 extrapol4.2002.7
Implements various methods for basis set extrapolation

2003-06-16 1007 2003 casvb2a.2002.7
Bug(s) fixed: 490
Bug fixes & extensions to casvb.

Jun 17 2003 casvb2b.2002.7
Bug(s) fixed: 000
Avoids potential problems in multi for certain variational VB calculations
started without a prior CI vector

2003-06-18 278 2003 dfti_new.2002.7
Place-holding patch which merges the new DFT quadrature code with the old,
now controlled with Molpro variable dfti_new

21 Jun 2003 pspace2.2002.7
Use only input p-space configurations if select=0 (multi, csf).
Degenerate CSFs are now only added if select.gt.0
In mrci, new pspace option mxshlp (column 4 on pspace card):
max shell for p-space CSFs. This is sometimes useful to avoid
unwanted CSFs and force convergence to the desired state.
New mrci option cluster=2: as cluster=0, but invariant to
rotations between degenerate states (if included). The
correction is computed using g=sum_{i=1,nstatr} <Psi|Psi_ref(i)>^2

23 Jun 2003 fujitsu-ssl2.2002.7
Updates for new PRIMEPOWER compiler.

Jun 23 2003 muaug.2002.7
Bug(s) fixed: 000
Trival change to avoid problems due to an uninitialized array (single
excitation integrals) in Augmented Hessian routines (when constructing
the full hessian).

2003-06-23 805 2003 dft_grid_coarse.2002.7
Implement coarse grid for early SCF iterations

2003-06-24 buildpatch_date.2002.7
Bug(s) fixed: 523
More robust determination of patch date, especially for AIX

2003-06-26 148 2003 gamma.2002.7
Incomplete gamma function

2003-06-26 wiom_progman.2002.7
Replace sreibw.f by wiom.f in programmers' manual, following 2002.7.readm2

2003-06-26 casvb2c.2002.7
Minor fix for the case of variational direct-product wavefunctions,
and a few cosmetic changes.

2003-06-26 readm3.2002.7
sets string in readm (iom.F)
removes sreibw.f since identical to wiom.f

2003-06-27 mpputil.2002.7
Fixes bug in list of environment variables, causing crash in (or following) mxma_mpp

2003-06-27 lmp2r12dom2.2002.7
Preparation for lmp2-r12.
get lmp2-r12 (optional df-lmp2-r12) input working
explicit card in ciinp for additional options
r12 excitation domains

2003-06-30 mkl60.2002.7
Adds intel/mkl60/lib/32 to list of directories to search for blas

2003-07-01 dft_more3.2002.7
Some more dft stuff: (a) modularization of grid generation (more to come)
(b) dfti/gridorb interaction. mpp is temporarily broken.

2003-07-01 gridsort3.2002.7
Sorting dft grid into blocks, and some more bugs fixed in dftgrid

2003-07-02 athlonlapack.2002.7
Bug(s) fixed: 525
configure was not picking up libathlonlapack due to a couple of typos.

2003-07-02 gatools.2002.7
Introduces a new molpro precompiler flag: gatools.

2003-07-02 kexta.2002.7
Skips small integral blocks if threshold KEXTA is set

2003-07-03 depositpatch_chmod.2002.7
Does a chmod a+r or a+rx on files to be deposited so that web server import script can pick up the files

2003-07-04 system_variables3.2002.7
Bug(s) fixed: 526 531
Make read-only system variables for intrinsic function names.

2003-07-05 scfthresh.2002.7
1.) Add input option thresh in scf (as in multi, mrci).
2.) Integral threshold reduction and fine grid selection before calling hfma.
This fixes the convergence check. New threshold THRCOARSE
3.) Fix maxit input in case of restart
4.) Select fine grid immediately in case of restart.
Otherwise maxit input not obeyed in case of restart. Also better in geometry optimization.
5.) Testjobs adjusted.

2003-07-07 dft_orbmed3.2002.7
Get rid of orbmed scheme, and some other little fixes.
Computed dft grid now stored on file 4 to allow shared implementation.

2003-07-07 lmp2_mppacc.2002.7
Disables ga_acc if insufficient GA space available (according to -G option)
Prints max ga memory in mpp_open as determined by -G option.

2003-07-07 timing_mod.2002.7
timing include file had variable mod which clashes with internal variable.

2003-07-07 dfcoul_mpp.2002.7
parallelizes basis_A_contract_tri

2003-07-07 batch_general6.2002.7
A working version of the generalized batcher for local MP2-R12.
Also some small fixes to the local dftrans code, to mend the
nclass.ne.1 cases.

9 Jul 2003 pipek_mpp.2002.7
Use mxma_mpp in some places in pipekm and locstart.

2003-07-09 posinp2.2002.7
Detect top of file using line numbers.

2003-07-09 basis_smh.2002.7
Adds basis_smh to give A^{-1/2} for any matrix in any basis.

9 Jul 2003 cc3ext_loc.2002.7
isymj was put outside loop 200 before (my fault), and since this variable can be
changed inside this loop, wrong results are expected in cases where multipassing
was actually used. Now isymj is put on its right place.

2003-07-09 alaska_memory_clean3.2002.7
Bug(s) fixed: 534
Clean up memory stack after call to alaska otherwise later calls to
pricall crash in check_address

2003-07-10 lmp2r12_internals.2002.7
A routine (lmp2r12_internal_ops) which builds all of the purely internal objects
for (DF-)LMP2-R12 by local resolutions of the identity, and saves them on disk.
This code has never been executed! It does, however, compile :)

2003-07-11 dflmp2g_dfscf_2.2002.7
df-lmp2 gradient for df-scf reference

2003-07-11 dftgrad.2002.7
Some fixing of DFT gradient. The grid is now back on file 1; I wonder
if this is OK. I strongly suspect that something is not
quite right with the weight derivatives any more. The gradients look
slightly wrong.

2003-07-11 dftrans_new9.2002.7
Next release of dftrans_drv and subroutines for df-lmp2-r12. Now
all density fitted integrals in place!
This patches fixes some undefined variables.

2003-07-12 aoint_matrix_threshold3.2002.7
If all two-electron integrals in a matrix are smaller than gthresh,twoint
then AO_Integral_Matrix_Get returns an address of zero, indicating that there
is no matrix. Disable this for the reading of two electron integrals in cpps.f.

2003-07-15 project_rotations.2002.7
Bug(s) fixed: 496
Fix projection of rotations in frequencies

2003-07-15 r12internals2.2002.7
Lots of (though not all) bugs removed from local RI assembly.

2003-07-15 ga_cleanup.2002.7
This patch replaces spread GA and TCGMSG calls with their global_
equivalents. Now direct calls to GA are concentrated in driver.f,
gamod.F, globalop.f, and mpputil.f. Ideally a next step will reduce
this list to just globalop.f.

2003-07-15 load_basis.2002.7
Modification to allow loading of orbital basis sets

2003-07-15 lmp2r12dom7.2002.7
Add new domains to cexplicit,

2003-07-15 patchlevel_patcher_2.2002.7
Some bugfixes for patchlevel in patcher. Added flag -s --secure for transfer over ssl. More to follow.

2003-07-16 configure_perl5.2002.7
Check for first perl in configure which is at least version 5.0.0 and adds flags to CONFIG. Does not halt configure if perl is not found.

2003-07-16 configure_libxml2.2002.7
Adding libxml compiler flags in configure

2003-07-16 aoint_engeler.2002.7
Marco's stuff for scaled hamiltonians

2003-07-17 field_format.2002.7
Fixed format statements in src/util/field.f.

2003-07-20 basis_eri2.2002.7
Fixes some bugs in the r_12 del^2 4-index integrals needed by PJK.

2003-07-22 ga_cleanup2.2002.7
Bug(s) fixed: 544
This fixes two bugs that were introduced with patch 'ga_cleanup',
namely bug # 544 and a spelling error. Furthermore it reverts to
direct calls to TCGMSG routines rcv and snd in mpputil.f. Moving
these calls to globalop.f causes problems on Peter's box, that I
don't understand yet.

2003-07-25 triples_timing3.2002.7
Output timing information in triples calculation. Includes an estimate
of the percentage of work done after each i (2--nval).

2003-07-25 multi_fci_parallel3.2002.7
Bug(s) fixed: 545 546
Parallelization of determinant CI parts in multi. Bug 545: not right for
multistate; fixed in multi_fci_parallel2; bug 546: wrong state-average
gradient fixed with multi_fci_parallel3

2003-07-27 load_basis2.2002.7
Fixes for loading ri basis

2003-07-27 lmp2r12_new6.2002.7
New LMP2-R12 stuff.
Fix Hylleraas energy and bug in first half transformation using ri basis.

2003-07-28 patcher_xml_gz.2002.7
Implement patches.xml.gz

2003-07-28 basis_eri3.2002.7
Fixes some bugs in the r_12 del^2 4-index integrals needed by PJK.
basis_eri3: Fixes some more bugs in addressing, and now applies the KE
operator to 4th label (ie electron 2 ket).

2003-07-28 kincorr.2002.7
Placeholder for kincorr stuff

29 Jul 2003 eom3.2002.7
1. a new subroutine anawf, which makes a population analysis for an EOM singles wf,
analysis is switched on by setting eomprint,popul=1 (see h2o_eom.test)
2. in cc3ext_loc - add a possibility of out-of-core when J and K (E) do not fit into the memory
3. eompairdom3 - out-of-core for W matrices
4. local=4 case further developed
5. many small changes and cleaning
a) change default mergelmo to 1 and add mergelmo=0 in loc_eom.test to get the same results
b) change the place for setting itadr, jeadr, and keadr in ccsd_loc so that I can release
all other memory before calling cceom_loc
c) prepare to do EOM 1el properties - ccjext for other symmetries

2003-07-29 kext_4ext.2002.7
removes severe bug in kext_4ext

30 Jul 2003 eom3cor.2002.7
Bug(s) fixed: 549
correct a bug in eomorbdom2

2003-07-30 ibm_xlf8103.2002.7
Bug(s) fixed: 512
some testjobs broke with release 8.1.0.3 of xlf. Changed -O3 to -O3 -qstrict and added multi/mucpnr.f and multi/muaug.f to f_opt2

2003-07-31 configure_xeon.2002.7
configure on xeon now returns ARCHNAME=p4

2003-07-31 grad_batch_sig.2002.7
Attempted fix for the gradient batcher signature.

2003-08-06 aix_shmem.2002.7
Marks shared memory obselete

2003-08-06 erfc_variable.2002.7
Bug(s) fixed: 552
To access the complementary error function, the parser recognizes REFC
rather than ERFC. Set ERFC as a system variable for when parse.y gets
updated and recognizes ERFC.

2003-08-08 ibm_eom3.2002.7
doresid_loc.f fails to compile on aix due to two bugs. One undefined logical and one misplaced continue.

2003-08-11 alpha_configure_xml.2002.7
Put call to xml2-config in backticks to avoid confusing /bin/sh on
Tru64.

2003-08-12 xml3.2002.7
Placeholder patch on the xml output stuff. Marco's putxml is now removed.

2003-08-14 ga_conflict.2002.7
Both casvb and Global Arrays 3.3 Beta used same standard code to get eigenvalues and eigenvectors which caused multiple definitions of symbols. Added a molpro_ prefix to these calls.

2003-08-20 dfti_print.2002.7
Comments test print of dwdx in dftigrad.f
comments istat=2 in dfti.f

2003-08-21 erestdom.2002.7
enable domain restart for local EOM

2003-08-22 readdump2.2002.7
New optiob OVL on start cards. Forces max overlap guess with previous
orbitals, even if basis dimension is the same

2003-08-22 dfmp2var.2002.7
Resets variable !DFMP2 to zero in non-density fitting calculation

2003-08-22 bench_direct.2002.7
Bug(s) fixed: 551
Modify MP4 target energy (due to patch scfthresh).

2003-08-23 dfexch_new4.2002.7
Proper implementation of df-hf. Initial iterations can be donbe with smaller
fitting basis and coarse thresholds. Final energy compute with large
fitting domains or without local fitting. More modular setup: Move
initialization of df-hf stuff from scfini to new subroutine dfhf_init.
Move various df utilities to new file df_utils.f
Move all density fitting utilities to new directory dfit

2003-08-23 test_vscf.2002.7
Add noproject option to vscf tests to avoid symmetry error

2003-08-23 cckext3.2002.7
New possibility for iflag: if negative, all operators are stored

2003-08-25 drv2el_scf_print.2002.7
Print screening info only with print,cpu=2 or higher

2003-08-27 qcisd_dipole2.2002.7
Bug(s) fixed: 553 556
Fixed problem with incorrect eigenvalues being passed to CPHF routine
during calculation of QCISD dipole moment after orbital rotation.

2003-08-28 lmp2-r12-new2.2002.7
Several corrections/updates to the LMP2-R12 project. Biggest change is removal of double counting of the MP2 energy.

2003-08-28 connectivity4.2002.7
Optionally use connectivity levels to define domains, pair type, domain
extensions etc. Nothing done yet for eom and triples.

2003-08-28 dflmp2g_dfscf_3.2002.7
df-lmp2 gradient, df-hf reference

2003-08-28 lmp2_canonical.2002.7
Adds "canonical" option for local calculations to use canonical
orbitals (full domains).

2003-08-29 dfhfenerg4.2002.7
New option cfit_final for selecting fitting option to be used for final scf energy

2003-08-29 fix_uninis.2002.7
Fixes some uninitialized variables

2003-08-31 lmp2_drv.2002.7
Separates lmp2 from the lccsd iteration loop. New subroutine lmp2_drv.
Moves vijm1_loc3 after cc3ext/cckext to simplify memory allocation.
Adjusts glycin_direct.test. Small energy change due to convergence in hf.

2003-09-01 ia64root_check.2002.7
Check that IA64ROOT is set before using it.

2003-09-02 uninis_eom2.2002.7
Bug(s) fixed: 560
initialize some variables

2003-09-03 bug561_2.2002.7
Bug(s) fixed: 561
fixes bug 561 (cphf problem in qci gradient)

2003-09-03 ccdomain_new.2002.7
Reorganizes and simplifies ccdomain. Make domains in logical order:
orbital domains, pair domains, operator domains, triples domains.
This is in preparation of new method to determine weak pairs, in which
operator domains will be determined after lmp2.
Domain restart modified, includes now connectivity
info. Operator domains are always remade after restart, using restarted
distance or connectivity parameters.
ccrest_loc/ccsav_loc moved from locorb.f to ccdomain.f,
and call to ccrest_loc also moved from getlmo to cidomain.

2003-09-04 sort_orbs.2002.7
Modify sort_orbs so that input orbitals are never overwritten

2003-09-04 dflmp2g_dfscf_5.2002.7
dflmp2 gradient 4 dfscf reference

2003-09-04 opteron2.2002.7
Basic configuration for Opteron machines. Build type depends on result
of `which pgf90`, e.g.:
/usr/pgi/linux86/5.0/bin/pgf90 => i4
/usr/pgi/linux86-64/5.0/bin/pgf90 => i8
and ARCHITECTURE provides "unix-linux-x86_64"; licence objects
included; minor fix for patcher.bootstrap.

2003-09-06 dfexch_rhf2.2002.7
Implements full fock matrix fitting for open-shell rhf

2003-09-08 ccsd_i3ext.2002.7
Restore the possibility to use option,i3ext=0 in CCSD even if the
3-external integrals have been computed in a previous calculation.

2003-09-08 riasmbl_ebc2.2002.7
Implements matrix A' for extended Brillouin approximation
New EXPLICIT option EBC (default 1).
A-A' is stored on new record.
Further fixes for df-lmp2-r12. Improve DIIS in lmp2.
Still not working correctly...

2003-09-08 dfscfg_3.2002.7
df-scf grad

2003-09-08 load_basis5.2002.7
Modifies load_basis so that full basis set names can be passed in.
If the name contains -context. the latter is deleted from the name.
Sets variable !DFBASIS_MP2 in dftrans_loc.f

2003-09-09 tran3_core_restart.2002.7
Bug(s) fixed: 558
In the case where transform finds the occupied orbitals are a subset of
the occupied orbitals for the already-transformed integrals, and 3-external
integrals are wanted, force transformation of integrals from the beginning
without reuse. This should eventually be replaced by the 3-external integral
code using MAPVAL.

2003-09-09 slapaf_fix.2002.7
geometry optimization using SLAPAF was not working with angsrom based
input. This is fixed now.

2003-09-10 gzip_not_compress.2002.7
Replace the use of uncompress by gzip -d since some Linux distributions
(notably Red Hat 9) don't have [un]compress.

2003-09-10 defbas_vdz_jkfit.2002.7
Adds defaults for VDZ-JKFIT. Use CTZ basis with highest angular momentum
functions omitted

2003-09-11 mp2pol2.2002.7
Implements anlytical mp2 polarizabilities.
Still core orbitals not working.

2003-09-11 lccsd_new4.2002.7
New lccsd_loc for weak pair selection after lmp2.
Replaces reserv by reservem and specify record types for easier debugging.
Also clean up some print formats.

2003-09-12 multi_states2.2002.7
Increases max number of state symmetries. Makes dimensions in avci
dependent of parameters in cstate. Previous dimensions were
inconsistent. Adds check for max state symmetries and max states
in muinp1

2003-09-12 casvb_multi_states.2002.7
Minor fix to take account of change to mxstsy in common/cstate in
the multi_states patches

2003-09-12 fitbasis_blocks3.2002.7
Fixes bug in df_basis_blocks. This bug inhibited block merging.

2003-09-13 cirs2ps_new2.2002.7
It improves the single/pair block for RS2 reducing memory and
increasing speed.

2003-09-13 lccsd_listp2.2002.7
new subroutine make_listp in ccdomain.f
Makes listp for all pairs before call to setup1_3ext
to avoid error exit (temp fix)

2003-09-15 enepart_connect2.2002.7
Implements energy partitioning according to connectivity.
Fixes bug involving the use of wrong distance matrix for
3-external integrals

2003-09-15 lmp2_mppdiis2.2002.7
Fixes parallel diis in ccmp2_loc3

2003-09-16 cipt2_omp_io.2002.7
Fix of Kallies for cipt2 in OpenMP

2003-09-16 omp_squash2.2002.7
Suppress openmp in cigpp when cipt2 because (reported by Bernd Kallies, 7/03)
it causes problems with allyl_cipt2.test and o3_cipt2.test on IBM.

2003-09-16 ddv2.2002.7
Inlining fix for IBM xlf from patch ddv in 2002.4

2003-09-16 local3_discon.2002.7
Eliminates disconnected terms in lccsd, local=3. Old behaviour can be reproduced
with option,idiscon=1.
Removes use of Fc in local=3 case. Fc is no longer needed in any closed-shell ccsd.

2003-09-17 cpar_mxrectr.2002.7
Increases mxrectr in common/cpar to 20 to allow SO calculations with
up to 20 mrci wavefunctions

2003-09-19 driver_zstand2.2002.7
Modifies setting of zstand so that lmp2 etc is recognized in
geometry optimization (previously zstand was set to mp2 rather
than lmp2, and posinp than crashed)

2003-09-20 mkl61.2002.7
Adds support for mkl6.1

2003-09-21 riasmbl_ebc8.2002.7
Fixes for lmp2-r12.
New procedure ccliis in ccdiis for solving linear equations
Sets nonorm=2 and shellperm=.false. for r12 calculation.
Corrects calculation of A' in riasmbl_ext.f (RI done in MO basis).
He now fully working with any basis set in PAO and MO basis.

2003-09-23 r12_Bmat.2002.7
Corrects bug in the Fock trace part of B matrix in LMP2-R12 program.

2003-09-24 cphf_general.2002.7
Fixes the call to ccdiis_general in cphf_general

2003-09-25 mp2pol3.2002.7
Some cosmetic changes of the mp2 polarizabilities

2003-09-25 merge_orbdom.2002.7
Fixes a bug in orbital merging: lone pairs were not merged

2003-09-26 d4ext_faster.2002.7
improved I/O in d4ext and bug fix in BlkBinSort

2003-09-27 load_basis6.2002.7
Eliminates any angular momentum restrictions present in AO basis
from default fitting basis. Make sure that at least d-functions for H
and f-functions otherwise are retained in fitting basis when
reducing max angular momentum in initial df-hf iterations.
Set default for internal basis name to context (usually jkfit or mp2fit)
rather than to the full name of the first basis set loaded.
Disables case iprintg(1)=1 in basis.f (chaos!).
Trunctates basis library names to 16 characters before call to get_lib_bas

2003-09-27 dflccsd.2002.7
Fixes basis load for df-lccsd. Removes some debug print and
cleans up cpu prints. Adds context ccsdfit in defbas for all mp2fit bases.
Still to be done: temp fitting bases should be destroyed at the end of lccsd.

2003-09-29 basis_contraction_pretty.2002.7
Bug(s) fixed: 576
basis_contraction needed pretty=.false. on every entry

2003-09-29 d4ext_faster_2.2002.7
some bug in dtraf_4ext.f fixed

2003-10-01 cidps_orthp.2002.7
Bug(s) fixed: 572
Fix mistake in cidps.f over selection where possible of non-orthogonal
pairs for transition density. Now npp1.ne.nblkp1 instead of npp.ne.nblkp.

2003-10-01 mscaspt2_and_stuff_for_cipt2.2002.7
This patch implements:
1) Multistate CASPT2 for RS2 (not RS2c yet)
following Finley et al. CPL,288,(1998),299
Full description in file MultiStateCaspt2.f
2) Printing of reference coefficients before and after cipt2 calc.
Since both 1 and 2 require modifications of cipro, I put
them both in the same patch even though they are unrelated.

2003-10-01 alpha_build_load_basis.2002.7
Remove double declaration of logical variable FIRST in subroutine
LOAD_SET.

2003-10-01 fujitsu_build_ri_lmp2g.2002.7
Changed some variables to INTENT(IN).

2003-10-01 imax3x.2002.7
alternative to rmax3x, based on bond distance criterion

2003-10-06 cosmo1.2002.7
Preparation of cosmo iteration (MP2)

2003-10-06 mpptim.2002.7
Replaces all calls to citime in ccsd/lccsd by add_mpptim.
Implement stack in mpptim.f, and print for each routine incremental
times only. If nprocs=1, print long form only for iprint,cpu=2 and higher.
Store and print timings for some intermediates only at high print level.

2003-10-06 contract_doc.2002.7
Tries to clarify documention for specification of contraction coefficients.

2003-10-07 eomprop3.2002.7
EOM-CCSD densities (expectation value type) now available

2003-10-08 timper.2002.7
Missing type declaration for variable timper added. The f90 compiler
on CUSS' Sun seems to be the first to really complain about this
issue.

2003-10-08 lmp2_new2.2002.7
Major revision and cleanup of ccmp2_loc3.
Separation of lmp2 and ccsd update.

2003-10-09 local3_better9.2002.7
Improves use of locality for local=3. Eliminates ltest(6) option (slow).
Fixes severe bugs in oprod_loc (case that result matrix is full and
vectors are local).
New subroutine transf2 in matrix.f

2003-10-09 lmp2sing2.2002.7
New option mp2singl to enable singles in lmp2.
Scan dfit and explicit options on command card.
New Option locinc for df-hf (localization each locinc iteration).

2003-10-09 r12_ttilde.2002.7
Computes r12 energy contribution using tilde t

2003-10-10 cosmo2.2002.7
Further cosmo stuff: mp2, modifications for cphf etc (working)
HF gradients (not yet working)

2003-10-12 dflmp2g_basis2.2002.7
Another Revision of load basis. Now basis sets and setnames separated.
Basis sets for df-mp2, df-lccsd generated in ciinp.

2003-10-12 lmp2g_nlint.2002.7
Remove allocation of memory for all-internal 2-el. ints. in ccmp2_loc3

2003-10-13 patcher_bootstrap_delobj.2002.7
Bug(s) fixed: 582
Delete object files after making patcher.bootstrap to avoid conflicts

2003-10-13 configure_libxml3.2002.7
Make $MOLPRO_LIBXML2_PREFIX also a prefix for running xml2-config.

2003-10-14 dflmp2g_dfscf_9.2002.7
replacement of general fock matrix construction on CPHF-Z RHS for better efficiency

2003-10-15 dflmp2g_cleanup2.2002.7
Some cleanup of df-lmp2g

2003-10-15 binput_variable.2002.7
Bug(s) fixed: 585
Avoids replacement of Atomic symbol by variable value in getdefbas
(e.g. variable i and Iodine atom).
Avoids creation of a large number of basgrp (nbasgrp) in kerni
in case of ecp input.

2003-10-16 ccdomain_for_cis.2002.7
correct a call for a simplified opkdom for CIS:
if(icis.eq.0) back to if(icis.gt.0)

2003-10-16 pseudo_libmol2.2002.7
Update of pseudo.libmol. Includes correlation consistent basis sets
for small core potentials.

2003-10-17 cpl_iterative6.2002.7
Implements direct solver for CPL, strongly reducing memory.

2003-10-18 gasum_mpptim.2002.7
Fixes crash with print,cpu=1 due to recursive call of prical.
(Problem occured in mpp case only: Stack overflow for GLOBAL_SUM_INT)

2003-10-19 epropcor8.2002.7
change order of loop 2000 in eomprop;
add some corrections in ldip.f (save density matrices only if dm>0,
and transition d.m. only if additionally eomprop>2);
XXX_mpptim in dolresi,ldip,ccsdp,cceom;
correct memory release in Htu; correct out-of-core u0cal;
use 3-external integrals for 2 left-eom intermediates instead of using
cckext and ccjext;
add thresh=1.d-16 in co_cordip.test testjob to avoid no convergence error in
mp2_cphf - no idea, why there was no error in previous versions!

2003-10-20 restdom.2002.7
Option not to relocalize the orbital in case of domain restart.
In this case use max overlap of new to old orbitals.

20 Oct 2003 hinputbug.2002.7
The dimensions of the arrays for the semilocal PP parameters were enlarged to h and i projectors

2003-10-21 dfhf_fitenergy10.2002.7
Implements non-local fit for final energy without construction of fock matrix.
Reference energy for closed-shell single-reference correlation methods taken from
variable !DFHFENERG in this case.
Now also second coulomb integral evaluation eliminated.
dfhf_fitenergy5: fix for mpp
dfhf_fitenergy6: New parameter locorb, allows to use localized orbitals in nonlocal fit.
dfhf_fitenergy9: Orbital statistics in df_sort_orb

2003-10-21 lib_makefile.2002.7
lib/Makefile used to delete cnvrt.exe after making ABDATA and RSYSW. This caused it to be remade on make install. If you are doing make install on a root_squash nfs directory then this is very irritating. Hopefully it was not deleted for a good reason.

2003-10-23 mp2pol4.2002.7
Fixes minor bug in mp2pol.f (missing character).
Comments out very hard debug thresholds for cphf.

2003-10-23 local_domains.2002.7
Add defaults idist, ivdist in multp case.
Make verydist=15/ivdist=8 default in all local calculations.

2003-10-23 cpl_iterative7.2002.7
Fix problem with rundant rotations in cpl. This might not fix
cases in which more than 2 rotations contribute to a linear dependence.

2003-10-23 ppwarning2.2002.7
Prints a warning in the case of large contributions within the
sum over products of angular and radial pseudopotential integrals

2003-10-27 dflmp2g_dfscf_11.2002.7
improved CPHF, DF Fockbuild directly in mixed occMO/AO basis

2003-10-28 mp2pol5.2002.7
improved mp2 polarizabilities, mp2dip is used for the zero order relaxed density

2003-10-31 geom_print.2002.7
Fixes format statements for printing bond distances (apparently messed up by xml patches)

2003-10-31 froz_gem3.2002.7
New integrals for geminal MP2 program df-mp2-f12.

2003-10-31 mp2pol6.2002.7
All components of the polarizability tensor are calculated now.

2003-11-03 darwin_conf.2002.7
Updated darwin.conf and getmachine to look for IBM xlf compilers.

2003-11-03 readop_multipole_nps.2002.7
Bug(s) fixed: 521
Add missing np setting for special multipole branch in nam_op.
Also add missing fac setting for this (hjw).

2003-11-04 multi_startorb_opt.2002.7
Discard start card during geometry optimization. Always use orbitals
of previous step. This avoids wrong occupation if start orbitals
from RHF are used.

2003-11-05 ciref_sym.2002.7
Checks symmetry of reference vector when using select

2003-11-12 epair3.2002.7
1) correct ini_ipair1
2) remove all instances of the use of ipaira,ipairb,listpa,listpb of a fixed
dimension (dependent of mxpar) from a common/eom and replace them by jpaira
etc. of a varying dimension. Remove ipaira etc. from
common/eom and therefore save some memory.

2003-11-13 mp2-f12_5.2002.7
New 3-index integrals routine for MP2-F12 and density fitting assembly routines.

2003-11-13 r12_disk.2002.7
Disk-based transformations etc so that DF-MP2-R12 works for large systems,
and simple H2O test job for DF-MP2-R12.

2003-11-15 enepartp.2002.7
Add a print of number of pairs in energy partitioning
Print both partitioning types, if possible.
EPDIST contains a required (set by USE_DIST) partitioning type.

2003-11-17 molproc_xml.2002.7
Implement molpro.c parsing of -X aka --xml-output (default) and
--no-xml-output to control whether or not molpro output is an xml document.

2003-11-17 xml9.2002.7
Some mor work on the orbitals, and testing with a transformation sheet to give
Molden file. This more or less works, except f functions not yet permuted.
xml5: proper interface block in xml.f
xml6: regular output now compile-optionally tagged instead of comment, but this
not the default.
xml7: earlier start of xml headers; fix atom labelling if geometry reordered.
xml8: by default, output file is not any more xml (need molpro -X).
xml9: resolve conflict with 2002.7.optg_hess

2003-11-17 optg_hess.2002.7
Fixes some problems in optimization with precomputed analytical hessian.
hessian command now working for hf. In case of xyz input, correct scaling
of hessian to ang.

2003-11-14 lapack_init.2002.7
Bug(s) fixed: 591
Initialize INFO variables to zero to work around a bug in MKL/LAPACK
on IA64, where the top 32 bits are untouched, and random data in these
can lead to apparent failure of LAPACK routines. Include ESSL function
DGEEV in seward_essl.f (cf. DGEGV) -- XEIGEN, ORTRAN and ORTHES are
also needed, but note that BALBAK, HQR and HQR2 are provided by
src/casvb/eis_cvb.f. LAPACK function DGEEV in src/blas/lapack2.f is no
longer compiled.

2003-11-18 dflmp2g_locfit1.2002.7
implementation of local fit domains in CPHF

2003-11-19 mp2pol7.2002.7
Improvement of the MP2 polarizabilities

2003-11-19 putxyz.2002.7
Adds zmethod/basis to comment line in xyz. Apart from being useful,
this avoids that ENERGY= is treated as variable when including the
xyz file in geometry input.

2003-11-23 mp2pol8.2002.7
Fix the error of the offdiagonal polarisabilities.

2003-11-24 erel_variables.2002.7
Bug(s) fixed: 593
Properly fill variable EREL in the case of mcscf;expec,rel.
Properly fill the variables MASSV, DARWIN in other expec,rel cases.

2003-11-26 regions.2002.7
Fixes some problems with regions

2003-11-26 wrapper_makefile.2002.7
Takes the value of WRAPPER_tcgmsg rather than first instance of parallel in PATH

2003-11-27 extend_orbdom.2002.7
Fixes bug in makeing orbital domain extensions using connectivity.
The bug caused all orbital domains to be extended.

2003-11-27 rpm_key.2002.7
Substitutes the value of MOLPRO_KEY as licence key when installing rpm

2003-11-30 restdom4.2002.7
Fixes severe bug introduced by patch restdom. Bug could lead to wrong orbital swapping.

2003-11-30 pipek_thresh.2002.7
Inreases threshold to detect small orbital rotations in first pm iteration from 1.d-6
to 1.d-4. This avoids incomplete localization in c2h4 with direct option
due to very slightly symmetry broken canonical orbitals.

2003-11-30 loctrip_print2.2002.7
Adds missing common tapes to drv_34ext_loc.f

2003-11-30 c2h4_loctest.2002.7
Additional tests for lmp2 and lccsd(t) for c2h4 (since localization is critical)

2003-12-01 lapack_init_doc.2002.7
Bug(s) fixed: 591
Document required initialization of INFO variables in LAPACK routines.

2003-12-01 patcher_sort3.2002.7
Bug(s) fixed: 468
Sorts patchlist in patcher so that it is written to PATCHLEVELS. This is then read by molver which reports it if matches installed patches.

2003-12-01 unini_vars.2002.7
Fix uninitialized variables in various places

2003-12-01 dftigrad_fix.2002.7
Fixes undefined npoa in dftigrad.f

2003-12-02 dfutils.2002.7
Adds missing implicit statements to load_basis routines in df_utils.
Fixes basis_inttypes: must use trim(mat) in index function

2003-12-02 prtran_str.2002.7
Introduces local variable typu in prtran to avoid crash in upcase

2003-12-02 unini_ints.2002.7
Fixes some undefined variables.

2003-12-02 molver_awk.2002.7
When patchlist is too long, splitting patches into lines for molver.f crashes awk on some machines.

2003-12-02 gridorb_debug.2002.7
Synchronization of gridorb.f (debugging output only)

2003-12-04 closelog.2002.7
Clise logfile when error is encountered (fehler, termxy)

2003-12-05 libmol_comment.2002.7
Bug(s) fixed: 579
Output comments from libmol files via the web interface.

2003-12-05 mp2-f12_6.2002.7
New code for MP2-F12 with a fixed geminal. Also small fixes to MP2-R12.

2003-12-05 dumpenergy.2002.7
Save df-hf energy in dump record for use as reference energy in later correlation calculations

2003-12-06 dfttests.2002.7
Modifies some testjobs since dft results slightly different

2003-12-06 cosmo3.2002.7
Changes for cosmo gradients and additional testjobs. Currently, symmetry
not properly working (aty least different results than without sym)

2003-12-08 dfg_sao_consolidate.2002.7
Bug(s) fixed: 557
Stick symmetry-adapted DFT derivative stuff back into main branch with
option driven by variable dfg_sao

2003-12-08 closelog2.2002.7
Close logfile when error is encountered (fehler, termxy)

2003-12-08 hetero_final.2002.7
Scaled hamiltonian implementation: heteronuclear, closed and open shell

2003-12-08 dflmp2g_locfit2.2002.7
further improvements

2003-12-08 ppgrwarning.2002.7
Prints a warning in the case of large contributions within the
sum over products of angular and radial pseudopotential integrals
(for gradients)

2003-12-08 aoint_fix_alpha.2002.7

2003-12-08 hfpol_fix.2002.7
Adds missing argument to call of cphf_general

2003-12-08 dfcoul_mppfix.2002.7
Attempt to fix mpp problem in basis_integrals.f