Molpro version 2002.9 release notes from patch history of preceding version

2003-12-11 libmol_comment.2002.8
Bug(s) fixed: 603
Fix problem with conflicting symbol names.

2003-12-11 molcas_dummy4.2002.8
Adds () to function allocdisk in molcas_dummy.f to avoid compilation error on hp
Also deletes gamod.F in src/mpp since actual version is in src/Molcas
dummy3: Now also deletes gamod.o to make sure the correct one is used.
dummy4: Also deletes gamod.o in src/Molcas to force recompilation.

2003-12-11 ecpinp.2002.8
Skip ecp input in kerni to avoid overflow of basis groups

2003-12-11 expiry_warn3.2002.8
Warn if licence is going to expire soon. Now controlled by environment
variable MOLPRO_EXPIRY_WARN setting the number of days.

2003-12-12 diag2_ia64.2002.8
Make info in diag2 static to avoid efc 7.1 compiler optimization bug.

2003-12-12 aosort_harmonize.2002.8
Bug(s) fixed: 607
A bit of address harmonization between processors in aosortv.f

2003-12-12 hfma_mpp.2002.8
fix order of calls to add_mpptim in hfma.f

2003-12-12 slater_include.2002.8
Several include files for src/slater missing

2003-12-14 field.2002.8
Fix bug (type deltee instead of deltae) in field.f

2003-12-17 eccprop.2002.8
Bug(s) fixed: 601,605
fix two bugs: 1a) iwhich has too small space, 1b) irep was
not set to 0, what resulted in ignoring call to cckext (now irep is
completely removed), 2) special treatement of mpp was removed from
LES for CCSD properties (before Jacobi was called in this case).

2003-12-17 unini2.2002.8
uninitialized variables

2003-12-18 molpro_cartesians.2002.8
Bug(s) fixed: 541
(Nearly) fixes the part of bug 541 that relates to density fitting methods failing
for cartesian AO basis sets. The error was that the cartesians in the integral program
are ordered lexicographically (xx,xy,xz,yy,yz,zz) which is different from the molpro order
(xx,yy,zz,xy,xz,yz). Also the normalizations are different. I am trying to reduce
the number of places that the order/normalization info is written by putting these
data in module basis_ao_info. The integrals produced by the integral batcher appear
now to be correct but methods such as df-lmp2 still fail for cartesian AO basis sets
when there is more than one group with angular momentum .ge.2.

2003-12-19 regions.2002.8
Implement restrict option for regions.

2003-12-19 aix_include.2002.8
replaced missing src/common/fexcp.h

2003-12-22 lattice_interface.2002.8
Bug(s) fixed: 503
New interface routines for lattice of point charges.
Not quite finished. Ultimately lattice data file will have format
<string>
<nlat>
x y z q igrad
...
with igrad=1 meaning do a gradient of the energy wrt this point, and igrad=0
meaning do no gradient for this point.

2003-12-24 f12andr12.2002.8
Part of the way to make Andy May's MP2-F12 work with R12 as well.

2003-12-31 ampr.2002.8
Molcas fix from Roland Lindh.
Fix two small bugs, 1) correct copy statement in Drv1El, and 2) fix signe error in ampr.f.

2004-01-05 alpha_build_explicit_dfmp2f12.2002.8
Fix build problem on alpha for explicit_dfmp2f12.f.

2004-01-05 opteron.2002.8
More fixes for Opteron; i4/i8 builds now supported with IFC and PGF90.

2004-01-06 iow_direct_loops.2002.8
Bug(s) fixed: 611
Wrap the calls to READ/WRITE in loops that proceed until either
the required number of bytes has been read/written, or an error
occurs.

2004-01-07 updui_trap_overflow.2002.8
Bug(s) fixed: 624
Trap infinite loop in multi/muint2.f:updui

2004-01-07 mp2dip_loc5.2002.8
Implements lmp2 density and gradient for local=1.
New vrt1 and modify vrt3. Now done in PAO basis.
New option do_vrt3 to avoid vrt3 using terms from mp2dmat_loc.
Simplified mp2dmat/mp2dmat_p for canonical MP2: Z-terms computed differently
Modify tvl2 (tvl2_loc) to account for external-internal part according to new reformulation.
Generalize mxmla, mxmlb to handle different matrix symmetries and adjust calls accordingly.
New Routine mp2pert to test cphf and cpl for first-order properties.
Added functinality to cpl: can now solve cpl or z-cpl equations,
see progman.ps (or codedoc cpl.f)
Fix core contribution in cphf_general.
Fix mpp bug in tvl_loc2

2004-01-07 cosmo.2002.8
Fix read statement in setamat.f (ioff -> dcoff)
Allocare iiatsp integer rather than real in cosmo_basis.f
Add dimension statements to cosmo_setamat.f

2004-01-07 opteron_patcher.2002.8
Minor fixes to patcher.c for i4/pgf90 build on Opteron.

2004-01-07 opteron3.2002.8
Replaces intrinsic functions which access q in basis routines
to avoid segementation faults. Changes marked with cpgf

2004-01-08 xyz.2002.8
Two minor changes to PUT output: (a) no longer include ENERGY= in the
default header, since this caused problems for xyz files read by Molpro;
(b) element symbols now in mixed case in xyz output (required by at least
WebMO).

2004-01-08 defbas_update.2002.8
Bug(s) fixed: 615
Update defbas to include Huzinaga basis sets (huzinaga, midi,
midi-bang mini and mini-scaled).

2004-01-08 pgf90_i4_casvb.2002.8
Lower optimization of casvb/symut_cvb.f and casvb/ci_cvb.f to get
casvb testjobs working for i4/pgf90 build.

2004-01-09 inv-mp2-r12.2002.8
Implements orbitally invariant MP2-R12/A' theory.

2004-01-09 optg_inactive.2002.8
Fixes case that numerical hessian is computed in the presence
of inactive variables. New pointer ipvar from active to zvarsym.
The following quantities are dimensioned nvar: zhessnum, zdipdiff, zgdone
The following quantities are dimensioned nvar: zgradvar, zgradlast, zhessvar

2004-01-12 pgf_rpath.2002.8
Adds directory of blas library to path to search for shared libraries

2004-01-14 pgf90_i4_casvb2.2002.8
Lower optimization of casvb/symut_cvb.f to get
casvb2 testjob working for i4/pgf90 mpp build.

2004-01-15 r12_improvements.2002.8
Improvements to the R12 codes, particularly new solver (superceding r12_svd), combined integrals for
F12 and R12 methods, change to dfasmbl to treat kinetic energy properly.

2004-01-15 basis_ao_info_dp.2002.8
Bug(s) fixed: 626
Added "implicit double precision" statement to module BASIS_AO_INFO to
ensure ROOT is a double precision array.

2004-01-17 perl3.2002.8
Improve perl detection in configure.

2004-01-19 batchs_memory.2002.8
Corrects underestimate of maximum memory requirements in batchs initialization.

2004-01-19 qsdpath2.2002.8
Fix some minor print problems to get more info about qsdpath calcs.

2004-01-23 ciref_sun.2002.8
Fix missing continuation mark in ciref.f.
Appearing only on SUN.

2004-01-23 newslapaf.2002.8
Bug(s) fixed: 606
Slapaf, at MOLCAS level 6.1

2004-01-26 molcas_vars.2002.8
Fix wrong calls to getvar in molcas_vars.f

2004-01-27 workingf12_3.2002.8
MP2-F12 working. Some other changes in the explicit codes.

2004-01-28 pipekm3.2002.8
Implement alternative pipek-mezey localization method in cpl.f (quadratic convergence, cplmethod=2)
New local options cpldel, cplmaxdl: the last "cpldel" functions of each type are deleted in PM localization
If cplmaxdl.ge.0, only functions up to this angular momentum are deleted.
In locali and lmp2/lccsd, new input: PIPEK,DELETE=cpldel,MAXDL=cplmaxdl,METHOD=cplmethod,SOLVE=cplsolve,
STEP=cplstep,ACC=cplacc,MAXIT=cplmaxit,THRESH=thrpip,SHIFT=cplshift
Save variable occscf in scfpr2.f

2004-01-29 f123el_2.2002.8
New code for explicit 3-electron integrals for density fitted MP2-F12 theories.

2004-01-29 hp_xml_no_opt.2002.8
There is a compiler error, when util/xml.f is compiled with optimisation on HP-UX 11.00.
util/xml.f is added to the no_opt list for HP-UX.

2004-02-02 lmp2_print.2002.8
Cleans up some print statements

2004-02-03 batchs2_mem.2002.8
Fixes erroneous memory request in batchs_init2.

2004-02-03 parse_i686_i4_compat.2002.8
New licence object for use with RedHat 9 and SuSE 9. Old licence object kept with different name to be used by older versions of glibc which fail to link with new object. utilities/get_parse no longer asks users if there is more than one available but instead uses newest.

2004-02-03 sse2.2002.8
Change detection for p4 archictecture to sse2 in flags in /proc/cpuinfo so that Petium Xeons and Pentium mobile processors are detected.

2004-02-04 dflmp2g_membugfix.2002.8
fixes memory bug in dflmp2 gradient
some other bug in slapaf call is also fixed

2004-02-05 f12.2002.8
Better density fitting for the density fitted 3-electron inytegrals.

2004-02-09 dflmp2r12_new.2002.8
New implementation of df-lmp2-r12/1a theory

2004-02-06 gtg1.2002.8

2004-02-10 lccsd_loc_r12en.2002.8
Fixes print of lmp2-r12 strong pair energy contribution

2004-02-11 mp2r12_orb.2002.8
Change mp2-r12 to read orbitals using cread rather than read_orb
to enable the use of localized orbitals

2004-02-11 locali.2002.8
Fixes bug concerning core orbital input in locali

2004-02-11 molver_syntax.2002.8
Replacing non-compliant shell syntax in bin/molver

2004-02-12 scan_inp2.2002.8
Modifications of scan_inp, scan_inp1, scan_inpg.
Allow several sets to be scanned, set names sepatated by colon.
Error exit if parameter is not found in given sets
Add scan_inp_nocheck which ignores errors
Modify calls accordingly

2004-02-12 basis_eqv.2002.8
Fix use of .eq. with logicals in src/basis/basis_froz_gem.f

2004-02-12 memtest.2002.8
Bug(s) fixed: 629
Various testjobs need more than 2 million words of memory on 64 bit systems when maxval=512

2004-02-12 dflmp2r12_new6.2002.8
New implementation of df-lmp2-r12/1a theory
Energy partioning for r12 pair energies
New options to define close pairs

2004-02-12 opteron_numforces.2002.8
Lower optimization of opt/numforces.f; fixes i4 build testjob failures
on opteron (both IFC and PGF90).

2004-02-12 df3elv.2002.8
Implements a simple DF algorithm for the 3-electron integrals in the V term of MP2-F12.
Also implements (at last) robust fitting for the commutator integrals in MP2-F12 theory.

2004-02-16 molcas_vars3.2002.8
fixes a bug in the setting of molcas variables

2004-02-17 scfini2.2002.8
Use tight thresholds from the beginning if start orbitals are found

2004-02-17 dfparams.2002.8
New input parameters invert_scf, invert_mp2, invert_r12 as well as
locfit_scf, locfit_mp2, locfit_r12 with specific defaults for the
corresponding methods. The parameters invert and locfit, respectively,
Take these defaults but can be used to change them in local input.

2004-02-18 dflmp2r12_new11.2002.8
fixes in df_fit_loc (reduced memory requirement)
Also overwrite half transformed integrals after sorting.

2004-02-18 grid_orbital_value_alpha2.2002.8
Bug(s) fixed: 630
Avoid calculation of 0**0 in grid_orbital_value as this leads to a
floating point exception (or NaN) on alpha. Treat some cases of
non-zero/zero combinations of x, y, and z in x**i 2002.8.grid_orbital_value_alpha.sh.gz 2002.8.grid_orbital_value_alpha2.sh Makefile base basediff buildpatch codedoc copyin cshrc depend depositpatch form_dftfreq.test grid_orbital_value_alpha.def grid_orbital_value_alpha2.def gridorb.diff gridorb.f gridorb.f.count gridorb.f.keep gridorb.log gridorb.o h2o_dfit.test h2o_dft.test lib libmolpro.a makepatchdef molpro molpro.exe molpro.f molpro.log molpro.o molpro.rc old-code.txt patchdiff profile test.log y**j 2002.8.grid_orbital_value_alpha.sh.gz 2002.8.grid_orbital_value_alpha2.sh Makefile base basediff buildpatch codedoc copyin cshrc depend depositpatch form_dftfreq.test grid_orbital_value_alpha.def grid_orbital_value_alpha2.def gridorb.diff gridorb.f gridorb.f.count gridorb.f.keep gridorb.log gridorb.o h2o_dfit.test h2o_dft.test lib libmolpro.a makepatchdef molpro molpro.exe molpro.f molpro.log molpro.o molpro.rc old-code.txt patchdiff profile test.log z**k with
special code.

2004-02-24 mxmabn3.2002.8
Extend mxman, mxmbn to use degmm; also generate mxman_dgm and mxmbn_dgm for general
multiplication.
Fixes preprocessor problem occuring if blas3 is not set.

2004-02-24 fitmltp.2002.8
Add missing d0 in a data statement for pi on line 572

2004-02-24 patcher_local2.2002.8
Bug(s) fixed: 642
If patcher fails to get contents file and a previous version exists locally then use that one.

2004-02-24 r12_output.2002.8
Some extra output from MP2-R12. Use explicit,print=3 to get maximum debugging output.

2004-02-24 molpro_net.2002.8
Make depositpatch send to ftp.molpro.net instead of molpro.bham.ac.uk

2004-02-25 dfexch_nrec.2002.8
Increas max number of records in dfexch.f

2004-02-26 r12core.2002.8
Frozen cores for MP2-R12 and MP2-F12

2004-02-27 ioff_dom.2002.8
Adds variable ioff_dom to common/clocal. This should always be set
to the highest domain number, and is used as offset for further domains, e.g.,
in r12dom.f.
Change ioff_dom to ioffdom in dtraf_lmp2.f and eomloc.f to avoid compilation
errors. Use consistent offsets in dftrans_loc.f

2004-02-28 dflmp2r12_2a3.2002.8
Implements lmp2-r12/2a
New implementation of AO/RI domains.
Clean up ccmp2_loc3 (remove all r12 stuff).
Clean up common/cexplicit

2004-02-29 mp2r12fix3.2002.8
Fix bug in explicit_dfmp2f12.f (address pointer IB used as do index).
Set variables with final energies.
In explicit_dfmp2r12.f set additional variables EMP2_R12A and
EMP2_R12 to be consistent with lmp2-r12
Use non-local df-hf reference energy if available
Temporary fixes memory problem in explicit_3idx_ints.f: reserve more
memory than requested by basis_3idx_batchs_init2

2004-02-29 libmol_plusd.2002.8
Additional basis sets with hard d-functions for Si-Ar

2004-03-02 undef_unit.2002.8
Fixes missing definitions of character string unit for the getvar calls,
and also corrects non-robust fitting of FJ integrals in the F12 program.

2004-03-03 makedepend_f90.2002.8
Fix up makedepend.c to deal with filename suffixes longer than 1 byte,
eg f90

2004-03-03 ftn_f90.2002.8
bin/ftn now recognizes .f90 suffix. Still to do: special compiler options
for free-form on systems where .f90 file suffix is not enough.

2004-03-03 cmlgeom7.2002.8
First go at reading in CML geometry. (Fix for non-standard comment characters by NTW)

2004-03-03 ibm_f90.2002.8
Bug(s) fixed: 577
Uses value of F90 rather than FC in CONFIG when compiling .f90 files. Controlled by -f90 flag passed to bin/ftn.

2004-03-04 xml.2002.8
XML output stuff after Phil Couch visit 3/3/04

2004-03-05 distpairs3.2002.8
Modifies subroutine distpair and isolates regions stuff in a separate routine.
Define close pairs by default. RMAXT(IMAXT) now alias for CLOSEP(ICLOSE).

2004-03-05 ftconv90.2002.8
Generalize ftconv to deal with free-format f90, with trigger !ftc.
cstart -> !ftc if
celse -> !ftc else
cend -> !ftc end
c; -> !ftc ;

2004-03-08 even_geminal_fix.2002.8
Fixes the exponents, and hence coefficients for even quantities
of gaussians used to make a even tempered gaussian geminal.
Also changes formatting of f12 program for consistency with r12 program.

2004-03-08 h_functions.2002.8
Bug(s) fixed: 541
A hack that corrects the problems with h and i functions in the Coulomb
part of the Fock builder by changing the (wrong) ao-specific transformation
with the correct general transformation. The limitation this imposes is that
the general transformation works only with nosym. This is no problem for the
Current use in HF theory, because symmetry is not implemented for exchange
either, but may be a problem in KS calculations that use density fitting.
More discussion can be found under the associated bug.

2004-03-10 dflmp2g_furtherfixes4.2002.8
some further fixes and tweaks in dflmp2g

2004-03-10 mp2r12fix4.2002.8
Fix string variable in call to readm. Must not be a constant
since returned by readm (crash on hp).

2004-03-11 patch_hpux.2002.8
Bug(s) fixed: 651
Fixing buildpatch tar options for HP-UX and makepatchdef to work when PWD is not set.

2004-03-11 mp2pol_ifppos.2002.8
Bug fixed: ccsd/mp2pol.log:Warning 11010 : At line 394 in procedure mp2pol Uninitialized variable 'ifppos'

2004-03-11 redef.2002.8
Removes a redefined variable that causes compilation failure under intel ifort version 8.

2004-03-11 unini3.2002.8
Fixes various undefined variables.
Fixes print of pair energies in ccmp2_loc, ccmp2_loc3

2004-03-11 cidft2.2002.8
CI and DFT are coupled splitting the 2e-interaction into a short range and long range part

2004-03-12 cosmo2.2002.8
New version of cosmo from Dr. Diedenhofen

2004-03-12 xml_configure.2002.8
Bug(s) fixed: 648
First part of changes to configure for libxml2. New flags to configure:
To use xml:
./configure -xml
To turn off xml (at moment you also need to revert patch cml_geom7):
./configure -noxml

2004-03-15 r12-long-range.2002.8
Fixes erroneous integrals in long-range cases in r12, and makes the
default integral threshold THRINT, into zero. Non zero values cause
problems at long-range. This is to be investigated. Also make a
trivial change to the input processing.

2004-03-17 mcfailsafe.2002.8
Implements failsafe option in multi. Sets harder convergence
thresholds in internal optimization. Starts each iteration with
ci (like diagci), and fully converges internal optimization
up to 12 iterations.

2004-03-18 lmp2_restart.2002.8
operator list have to be rebuilt in case of restart, since no longer saved
Match orbital domains and reorder orbitals accordingly (new option checkdom)

2004-03-18 3idx_ints_fix.2002.8
fixes incorrect syntax for readm in explicit_3idx_ints.f used in
mp2-r12 and mp2-f12 programs

2004-03-19 makedepend_rindex.2002.8
Include strings.h for rindex for systems where this is not implicit.

2004-03-19 thrdscf.2002.8
Modify integral threshold in dscf and in dfhf.
Avoid blow up in dfhf for diffuse basis sets.

2004-03-21 ftcflush.2002.8
Adds util/util.o to the ftcflush list (to recompile if olddiag2 is set)

2004-03-22 libmol_install.2002.8
Bug(s) fixed: 657
Cope with libmol files containing parentheses during "make install".

2004-03-22 rilmp2g_clean.2002.8
Removes some debug print in ri_lmp2g.f

2004-03-23 dfhf3.2002.8
Modify memory management in dfexch.f. Do localization and new domains
in final HF iteration. Separate files rhfpro.f and uhfpro.f for convenience.
Automatic save/restart of cphf solution in cphf_mp2.
Automatic adjustment of hf and cphf convergence thresholds in
geometry optimization, depending on OPTCONV.

2004-03-26 extdom5.2002.8
Modifies algorithm for extending domains. Separate extension
of strong and close pairs. Print extended domains.
New parameter chgminp: Only atoms with charge greater than
this are used to determine pair type (strong, weak etc).
(This was hard coded 0.5 so far. New default value is 0.2).
New default value for chgmax (BP procedure is used to select
atoms below this charge). New 0.2, old 0.4. This might cause
inclusion of more atoms in orbital domains for conjugated
systems and changes the domains for form_dual.test
Fix for local=1 in case that close pairs are present:
set ipend_proc=min(ipend_proc,npp)
in truncate_pairlist to avoid overwriting problem.
Add local=1 test to propane_lccsd
Add testjob propane_lmp4.test

2004-03-23 gtg2.2002.8

2004-03-26 dflmp2g_furtherfixes5.2002.8
dynamic setting fo rdomaux_scfgrad depending on OPTCONV

2004-03-28 makedepend_use_comma.2002.8
Extend makedepend to handle
use modulename, a=>b
by ignoring everything from the comma onwards

2004-04-01 save_data.2002.8
Renames subroutine data to save_data to avoid duplicate names.

2004-04-01 cvectr.2002.8
Make common/cvectr an include file to avoid warning messages about changing lengths

2004-04-01 natorb2.2002.8
Fixes bug in qcisd affecting calculation of natural orbitals (missing
orbitals in ccys.f). Sets option icphf=1 if natorb or dm input is present.
Compute and print natural orbitals also in case that natorb directive is given
without arguments.

2004-04-01 rhfpro_dual2.2002.8
Add some code to generate proper canonical orbitals and
orbital energies in case of maxit,1 (for dual basis calculations).
Fixes wrong MP2 energies in such cases.
Also eliminates in this case recomputation of dfhf-energy without
local fitting if local fitting is disabled.

2004-04-02 h_functions3.2002.8
Bug(s) fixed: 541
A hack that corrects the problems with h and i functions in the Coulomb part of the Fock builder by changing the (wrong) ao-specific transformation with the correct general transformation. The limitation this imposes is that the general transformation works only with nosym. This is no problem for the Current use in HF theory, because symmetry is not implemented for exchange either, but may be a problem in KS calculations that use density fitting. More discussion can be found under the associated bug. Addition by hjw: Use new code for nsk=1, old code for nsk>1.
And now make an error in the unsupported case, ie, when nsk>1 and h or i functions are present.

2004-04-02 ccmp2_loc2.2002.8
Comment ipend_proc=min(ipend_proc,npp) on line 1208 of ccmp2_loc.
This caused trouble with tests form_optloc and loc_eom3 in mpp case.
Set close pairs to weak (ipair(5,ip)=2 if iter.eq.maxitcls

2004-04-02 fix659.2002.8
Bug(s) fixed: 659
bugfix lccsd dommerge

2004-04-03 prepin_echoproc.2002.8
Modify prepin to echo procedures when they are defined. Procedures in molproi.rc
are not echoed.

2004-04-03 restart_variables.2002.8
Recover Variables in case that file contains only the variable record (500).
Previously, the file was considered empty in this case and the variables were lost.

2004-04-04 scf_iniocc.2002.8
Tries to fix occasional wrong occupation in hf after restart.
Checks consistency of initial occupations with number of electrons.
Consistent use of CHARGE in putocc/getocc.

2004-04-06 relintens.2002.8
Avoid division by zero if max intensity is zero

2004-04-06 dftigrad_clean.2002.8
Remove debug print in dftigrad.f

2004-04-07 hpux1131.2002.8
changes to utilies/hp.conf for version 11.33 of HP-UX

2004-04-08 make_parallel.2002.8
Changing the order in which directories in src/ are compiled so that src/Molcas and src/slater are compiled first. This speeds up compilation when make is done in parallel

2004-04-08 fix621.2002.8
Bug(s) fixed: 621
bug fix for bug621 (bug in bucketsort of d4ext transformation)

2004-04-20 noxml.2002.8
Bug(s) fixed: 648
Added util/xmlstubs.h which will allow code to compile without including xml libraries.
Changed configure so that -noxml is default when first run, then uses previous choice.
Changing between xml/noxml causes recompilation of object files.

2004-04-20 hfexchange.2002.8
Implements option to print exchange energy in HF (defined as tr(D*K) without factor 1/2).
Saves this in variable e_exchange. Can be used to optimize DFHF fitting bases.

2004-04-22 fci_all.2002.8
Undocumented option for fci (all) to print out all eigenvalues and eigenvectors.

2004-04-23 cidft_fock.2002.8
Modifies subroutine optr (cicore.f) to read full fock matrix in
ccsd/dft case

2004-04-26 compress.2002.8
Fix error in readcmp occuring in the very special case that the
last record has less then 6 words left for the labels indicating
end of file. Now consistenly with writec_term changed to less
than 4 words (only 3 needed for labels, 1 reserved to round up to
multiple of real*8)

2004-04-26 get_guess.2002.8
Bug(s) fixed: 667
Fixes overwriting error in case that the guess orbital basis set
is larger than the current basis set.

2004-04-29 dfdftsapt6.2002.8
implementation of DFT-SAPT with density fitting by A. Hesselmann, G. Jansen, and M.Schuetz
now with patch headers for the .f90 files...

2004-04-30 opteron_blas2.2002.8
Bug(s) fixed: 670
Locates optimised blas on opteron systems

2004-04-30 olddiag2b.2002.8
Bug(s) fixed: 670
set ftcflag olddiag2 if amd64/i4/atlas
remember previous ftcflag settings

2004-05-04 batchs_poisson.2002.8
Some improvements in basis_integral_batchs.f for poisson case. Avoid
gpt-multiplication with empty blocks. Add timings and print these in
dfexch.f (gprint,cpu=1 inly).

2004-05-05 dfdftsapt10.2002.8
possibility to use MP2 fit basis for dispersion terms is implememted

2004-05-07 opteron_errormsg.2002.8
Gives error message when configure can't decide which compiler to use on opteron

2004-05-10 opteron_getmachine3.2002.8
This patch determines compiler and integer size on opteron. Defaults to i4, defaults to ifc for i4, pgf for i8. If no flags specified on subsequent configure then old values remembered.

2004-05-12 cpl_lenv.2002.8
Rename lenv to lenu to avoid overwriting of common variable lenv

2004-05-12 lapack3.2002.8
Bug(s) fixed: 664
New lapack subroutines in src/blas for sapt

2004-05-12 fredbug.2002.8
new threshold for Freds potentialy buggy 3-idx prescreening on primitive level

2004-05-12 cardiff.2002.8
Address and copyright changes; remove info

2004-05-14 r12_memory5.2002.8
Fixes excessive memory use on kinetic energy integral symmetrization.
Corrects bug in paging in DF-MP2-R12 program.
Allows use of Cholesky decomposition/back-substituion instead
of matrix inversion/multiplication. Invertion controlled by
dfit,invert=1 (inversion) or dfit,invert=0 (cholesky).
Inserts a printout of the condition number of the matrix to be
inverted/cholesky-ed.
Adds two lapack routines for back-substitution and to compute
the condition number in lapack2.f. The routines are DPOTRS and
DPOCON and came from netlib.

2004-05-14 d1_diag.2002.8
Bug(s) fixed: 652
Implemented D1(CCSD) diagnostic for closed-shell case. This is defined
By Janssen and Nielsen in CPL 290 423-430 (1998).

2004-05-17 batchs_poisson2.2002.8
Some further improvements of basis_integral_batchs.f
Explicit code in ao_contr; modified sort; reduce overhead to produce klist

2004-05-17 lapack2.2002.8
Adds a few routines to lapack2.f needed by other routines I added in patch r12_memory5.

2004-05-18 enest.2002.8
Disable energy estimate for 1-electron case. Also add noenest directive
to disable energy check

2004-05-21 extrapol.2002.8
Add parameters pc, pr for (l+pc)**x extrapolation

2004-05-21 fix675.2002.8
fix for bug 675

2004-05-21 erel_var.2002.8
Fix setting of variable erel in case of multiple state calculations

2004-05-21 dfexch2.2002.8
Improvements of dfexch.f:
Product screening in local fitting and assembly.
Connectivity can be used for fitting domains.
Use sparsity in second half transformation for non-local fit of energy.
Use triangularity for non-local fit of energy.

2004-05-25 sapt_memfix.2002.8
memory fix by Andreas for df-dft-sapt

2004-05-28 head.2002.8
Bug(s) fixed: 671
The latest GNU coreutils 5.2.1 has, in it's infinite wisdom, made 'head -1' and 'tail -1' obsolete.
One consequence of this is that patches made without this patch applied will not work on these systems as there is an embedded tail command.

2004-06-04 casvb_mem1.2002.8
Trivial change to memory handling. Might matter for -i8 and/or large N.

2004-06-04 dfdftsapt13.2002.8
further improvements by A. Hesselmann

2004-06-05 d1_diag2.2002.8
Bug(s) fixed: 652
Minor change in logic.

2004-05-17 d1_diag_open_shell.2002.8
Bug(s) fixed: 652
Implemented D1(CCSD) diagnostic for open-shell case. This is defined
By Lee in CPL 372 362-367 (2003).

2004-06-06 matrop_doc.2002.8
Fix misprint in matrop manual. COUL -> EXCH

2004-06-07 grad_lattice2.2002.8
enable XFILED option in Molcas to get correct lattice contribution to the gradient.

2004-06-07 multi_dmsav3.2002.8
Save density matrices only of DM directive is given. DM,1 saves
also the spin density matrices. Update manual accordingly.
Add DM directive to testjobs fci.test and hybrid_dft_ci.test

2004-06-07 gtg3.2002.8
Reads ipair_neglect in gemctl_size; removes maxjedomsize in geminal
files
gemctl_drv.f src/local/
geminal.f src/local/

2004-06-07 cref_f90.2002.8
common/cref can compile inside .f90

2004-06-07 codedoc_f90.2002.8
Changes to codedoc: (a) support for .f90 files (b) pdf output with -P

2004-06-07 altix_reshape3.2002.8
reshape intrinsic on altix machine broken for src/util/displace.f:220
rewrote with do loops

2004-06-07 wrbig3ext4.2002.8
writem_big/redm_big for 3-ext integrals, this time with
all include files included

2004-06-08 qmc2.2002.8
Bug(s) fixed: 678 679
Quantum Monte Carlo stub.
qmc2: (a) remove external rand/srand (fixes bug 678?) ; (b) forall->do (fixes bug 679).

2004-06-09 dflmp2_maxrec.2002.8
Increases max number of records and buffer length in dftrans_loc.f

2004-06-09 grad_lattice3.2002.8
Some fixes for lattice_gradients

2004-06-09 ciexit_variables.2002.8
1.) Variables for properties printed in the same order as energies, independent
of reordering of roots (state with lowest energy is first).
2.) set variable COEFF even if csf given on puncsf card does not match or is below threshold

2004-06-10 local_chgmax6.2002.8
Increases chgmax back to its original value of 0.4
This affects the orbital domains; with diffuse basis sets, smaller
domains will result/

2004-06-14 dfexch_mpp.2002.8
Fixes a bug in df-hf for mpp case
Removes test print in ccmp2_loc3.f

2004-06-14 globalop_undef.2002.8
Fixes undefined variable iout in lines 925 and 950 of globalop.f

2004-06-15 multi_maxdav.2002.8
Increases maxdav to 50 in multi

2004-06-15 dfexch_mxvl.2002.8
Error exit in dfexch if nocc.gt.mxvl

2004-06-15 f90_preprocess.2002.8
Bug(s) fixed: 678
Enable automatic ftc processing of .f90 files

2004-06-15 qmc_main2.2002.8
Bug(s) fixed: 789
qmc: use f90 intrinsic random_number

2004-06-16 mrci_printdav.2002.8
New print option dav:davidson in mrci to print cluster corrected energies in each iteration.
Modify setkey* in parse.f to handle aliases in key list (separated by colons).

2004-06-17 input_seward.2002.8
some MOLCAS bug causing error in recprt

2004-06-17 triples_grad.2002.8
First working implementation of the QCISD(T) gradient.

2004-06-18 df3ext_3.2002.8
all 3-external integrals in lccsd done by density fitting

2004-06-23 cipt2.2002.8
Reduces convergence thresholds in cinit.f/ci_cipt2b; print warning
message about which state is optimized only in multi-state case.
More sensible error message in intrude case in cirs2iic.f.

2004-06-23 largefiles.2002.8
Change utilities/linux.conf to set large file support for >= 2.4 kernel. User no longer asked for preference.

2004-06-23 bench.2002.8
Modify target energies of normal_dft and big_direct_lmp2.bigcom.
Changes in normal_dft due to new dft implementation (to be checked).
Changes in big_direct_lmp2.bigcom due to modified HF convergence.
Timings cannot be compared with 2002.65 since number of iterations changed!

2004-06-24 cidft3.2002.8
Additional input options in ciinp.f SRCDFT and SRXCDFT for cc/dft coupling

2004-06-24 triples_grad_mpp2.2002.8
Bug(s) fixed: 683
Disable mpp when calculating QCISD(T) gradient terms.

2004-06-25 global_inp.2002.8
Fixes global_inp, determine parameter types from registry

2004-06-25 opteron_parse.2002.8
New opteron licence object

2004-06-26 prop_qm.2002.8
Allocate all memory needed for making quadrupole moment operators
before calls to mltpl_general. Fixes overwriting problem
in big cases.

2004-06-27 string_utils.2002.8
Add missing double precision declarations to get_integ and get_float

2004-06-27 virtmo.2002.8
Increases threshold for linear depenedencies in virtmo

2004-06-28 ftcdel2.2002.8
changing from configure -xml to configure -noxml requires deleting an object from library. List of objects in FTCDEL in src/GNUmakefile

2004-06-28 atomic_mpham.2002.8
Code for the nuclear potential weighted moeller plesset hamiltonian.
wmpham needs to be specified in order to compile with this option

2004-06-29 mp2pol_loc1_v1.2002.8
Implementation of the MP2 polarisabilities
for local=1 domsel=1.0

2004-07-06 parse_ia64.2002.8
New itanium licence object

2004-07-07 f77version.2002.8
Report correct compiler for intel fortran compiler 8

2004-07-07 3eigrid3.2002.8
More on 3-electron integrals on the grid, including a test code for building
2-electron integrals and doing MP2.

2004-07-07 explicit_ints.2002.8
New interface for explicitly correlated integrals, and some code for the new R12-wof method.

2004-07-07 ccdft4.2002.8
Coupling of CC/CI LDA/PBE XC/C (DFT exchange/exact exchange) is now implemented,
subroutines have a clearer structure

2004-07-07 ccdft5.2002.8
Fixes a bug in dftfun_exerfpbe.f

2004-07-08 ccdft_chirs.2002.8
Fixes bug, isrxcdft had not been declared

2004-07-08 isrxcdft_decbug.2002.8
Double declaration of isrxcdft removed

2004-07-07 triples_grad2.2002.8
Further work on QCISD(T) gradient implementation. In particular, all
terms are calculated in a restricted summation over i, j, and k. New
testjob (geometry optimization of H2O).

2004-07-09 connectivity_format.2002.8
Minor change in printing format in connectivity.f

2004-07-09 r12wof.2002.8
New stuff for R12-WOF method.

2004-07-13 amd64_ifc_i8_blas4.2002.8
Select correct blas for an opteron ifc i8 build

2004-07-14 userprog2.2002.8
Call userprog with KEYWORD USERPROG

2004-07-15 easy_qmmm.2002.8
Get (fractional) point charges into the main geometry

2004-07-16 triples_grad_mpp3.2002.8
Integrate QCISD(T) gradient terms into existing mpp framework.

2004-07-19 r12disk.2002.8
Fixes bug that broke successive R12 calculations.

2004-07-19 df_basis.2002.8
In r12 calculation, use df_basis as fitting basis if given on explicit card.
If not given on explicit card, use basis (df_basis) as given on mp2/lmp2 card or cfit/dfit cards.
Make df_basis an alias for basis (similarly for basis_mp2, and all other fitting basis types)

2004-07-20 xml_basis.2002.8
Some further tidying on xml basis output. Namespace removed. Association xlink.

2004-07-20 sapt_regfix.2002.8
registry fix for ks-sapt

2004-07-21 mkl_ia32.2002.8
mkl_ia32 is faster on a xeon than mkl_p4 and has been made the default blas.

2004-07-21 dflmp2r12_project2.2002.8
Implements local projectors for lmp2-r12. Activated with projects=-1

2004-07-21 domstart2.2002.8
Remove modifications of ciinp.f and util.f made in patch domstart.
Instead, implement restart of atom domains to ccdomain; record
is now given as local option DOMSTART=record

2004-07-22 basis_explicit_declarations.2002.8
This patch fixes undeclared variable ioff_grp in basis_integrals
and variable iKij in explicit_mp2

2004-07-22 triples_aix.2002.8
Fix an undefined logical in src/ccsd/triples.f which fails with IBM compiler.

2004-07-22 blkbinsrt_fix.2002.8
fix in BlkBinSort memory allocation

2004-07-23 geminal_fitting.2002.8
A new set of options for the basis_froz_gem routine to allow
different fitting criteria for the frozen geminal

2004-07-23 pwof.2002.8
implements more of the projected WOF method, using quadrature for integrals.

2004-07-29 sharedliblinux.2002.8
When building lib/libmolpro.a from scratch a loop over directories is now used to avoid a bug in bash filename expansion

2004-07-29 r12_memory6.2002.8
Reserve more memory for shell routines in explicit_3idx_ints.f
Comment test print in explicit_mp2.f

2004-07-29 df3ext_cached.2002.8
some caching for df3ext contractions

2004-07-29 kincorr1.2002.8
kincorr stuff as at end 2003

2004-07-30 fixpol.2002.8
Fix of uninitialized variables in mp2pol

2004-07-30 cml_size.2002.8
Fix compilation problem with SIZE function in cml.f90
on opteron PGF90 (5.2, i8).

2004-07-30 basis_froz_gem_derf.2002.8
Replace derf by dderf in basis_froz_gem.f

2004-07-30 dgetrf_info.2002.8
Set info=0 before call to dgetrf. Otherwise crash on opteron pgf90/5.2

2004-08-01 mp2fit_new3.2002.8
New mp2 fitting basis sets from Weigend. Reoptimized VDZ/AVDZ. New V5Z/AV5Z and
CVnZ, ACVnZ. Replace two testjobs since VDZ fitting basis changed.

2004-08-02 fixpol2.2002.8
Another fix of uninitialized variables in mp2pol

2004-08-02 ccmp2en_loc_alpha.2002.8
Correct callers of ccmp2en_loc; fixes some lccsd testjobs on alpha.

2004-08-02 mp2ymat_alpha.2002.8
Initialize scratch array to avoid floating point exception on alpha.

2004-08-03 dfexch_alpha.2002.8
Initialize q(iorb) array to avoid floating point exception on alpha.

2004-08-05 vscf.2002.8
New VSCF and VCI program

2004-08-07 nbo2.2002.8
Prepares for nbo interface.

2004-08-07 cosmo_setstr.2002.8
Replace constant by variable mxs in call to setstr

2004-08-07 pathf90.2002.8
Changes for pathf90 compiler

2004-08-09 aoint_matrix.2002.8
Replace intrinsic function maxval by loop to avoid crash in alloc.
Unclear if this crash is due to earlier overwriting or not.

2004-08-11 mp2fit_new4.2002.8
New MP2 fitting basis sets from C. Haettig.
cc-pV5Z for H-Ar and aug-cc-pV5Z for H, He, B-Ne, Al-Ar
(for Al-Ar the VnZ sets are identical with those for V(n+d)Z)
cc-pwCVXZ/cc-pwCV(X+d)Z for X=D,T,Q,5 ffor B-Ne and Al-Ar
cc-pVXZ with X=D,T,Q,5 for Li, Be, Na, and Mg
new optimimized cc-pVDZ for C, N, O F;
new optimized cc-pVTZ and cc-pVQZ for Ne
QZVPP for H-Kr
SVP, TZVP, TZVPP for Ne, Ar, Kr

2004-08-11 vscf1.2002.8
Several small bugfixes concerning the new VSCF program

2004-08-11 locorb.2002.8
Do not localize if locorb=-1, independent of value of icphf.
For some reason, localization was always done if icphf=1.
Also, disable restart staff entirely of locorb=-1.
In case local=1, canonical=1 sets locorb=-1 and ican=0

2004-08-13 basconv.2002.8
Attempt to fix various bugs in basconv/defbas:
(i) If context.ne.' ', it found the first fitting basis in defbas, independent
of type. This caused chaos if AUX or POISSON appeared in defbas before the normal MP2 fitting bases.
(ii) Input of the form spd,atom,basis/context now works also if basis is a short-name
(iii) In ciinp: call basis_data_dump after each loaded basis sets. Avoids error in df-ccsd case.

2004-08-14 readop_lop2.2002.8
Skip 2-electron angular momentum operators (lop2, lzlz etc) in
gexpec_inp and print a warning

2004-08-14 musamc_weights.2002.8
Bug(s) fixed: 653
Error exit if attempt is made to compute gradients with different
weights. Gradients for this case are wrong (unknown reason)

2004-08-15 rsjc_thresh3.2002.8
rsjc_thresh: Reduce thresholds in rsjd to 1.d-14. This avoids infinite loop in so.test on p4.
rsjc_thresh2: Multiply threshold by absolute value of smallest diagonal element to avoid infinite loop (bug 673).
rsjc_thresh3: Previous change caused chaos. Now threshold as original, but thrsh=max(thrsh,t1*dii), where dii
is absolute value of smallest diagonal element.

2004-08-16 readop_rel3.2002.8
readop returns MASSV+DARWIN for operator REL (EREL is an alias).
This makes subroutine cowan obsolete. Expectation values for REL
now work as for all other operators.

2004-08-16 project_dav2.2002.8
Implements Davidson correction for projected calculations.

2004-08-16 triples_grad3.2002.8
Further work on QCISD(T) gradient implementation. Disk-based
transformation of 3-ext integral derivative prefactor matrices for
when insufficient core is available.

2004-08-17 diabci.2002.8
SImple program for diabatization using CI-coefficients

2004-08-18 triples_grad4.2002.8
Fix use of uninitialized variable in t2tran_triples_add.

2004-08-19 bascop5.2002.8
New bascop. Merges new and old sets.
New basis_evaluate: evaluate basis sets before density fitting calcs.

2004-08-19 print_amplitudes.2002.8
Bug(s) fixed: 652
Print amplitudes greater in magnitude than THRPRI threshold.

2004-08-25 config_pathscale.2002.8
integrating pathscale compiler into configure process

2004-08-26 pgi_conf.2002.8
Better detection of the portland compiler.

2004-08-27 transfix.2002.8
memory problem in transform fixed (seward needs some more memory for some big cases)

2004-08-27 ltrip_3extcache.2002.8
new memory cacheing for 3-ext integrals in local triples

2004-09-01 character_f90.2002.8
Change all character statements in commons to f90 form

2004-09-03 xml_basis2.2002.8
Some further tidying on xml basis output.

2004-09-08 aix_vscf.2002.8
source file with ^M characters causes IBM xlf compiler to fail

2004-09-09 lsint.2002.8
Bug(s) fixed: 509
Some array dimensions fixed; use MAXBFN where appropriate, and check
that we have allocated enough memory in LS12; large arrays allocated
dynamically.

2004-09-09 pbs_nodelist.2002.8
Fix PBS_NODEFILE usage. If -N used when PBS_NODEFILE is set then exits with error

2004-09-12 basis_check5.2002.8
Impements new basis_check, including check of fitting basis sets.
Restart of fitting basis now works correctly. Also the
fitting basis is revaluated if it changes without call to int.
Changes of variables in basis input are trapped.
getdefbas recognizes context.
Default contexts jfit,jkfit,mp2fit,ccsdfit if these setnames are used.

2004-09-12 dftrans_lccsd2.2002.8
DF transformation 0-2 ext for lccsd. Clean up and use consistently load_fitbas, df_fitdomains etc.
Rename dftrans_drv.f and dftrans_loc.f to dftrans_lmp2_r12.f and dftrans_lmp2.f,
respectively, and modify to use the new utilities. Coulomb part and interface to ccv_loc4.f
are still missing.

2004-09-13 mpp_environment.2002.8
add MOLPRO_OUTPUT to list of distributed environment variables, so that
get good file names with MOLPRO_GA_DEBUG.

2004-09-13 triples_tasks.2002.8
Fix wrong number of tasks in triples.f ; initialize minbt, maxbt in triples
and tripu

2004-09-13 ciexit_puncsf.2002.8
Fixes problem in ciexit with saving the last coefficient using puncsf

2004-09-14 ga_3ext.2002.8
Put 3-external integrals in GA record

2004-09-15 dflmp2_regions2.2002.8
Modifies dftrans_lmp2, dftrans_lmp2_r12, and dftrans_lccsd, df_utils to
treat regions efficiently by transforming only the necessary orbitals.
Orbital transformation matrix is compressed to contain only correlated
orbitals, and a pointer list listmo is used to point from the compressed
to the original orbitals (see subroutine df_listmo)
Also fixes bug in df_fitdom.

2004-09-15 sun_conf_primepower.2002.8
Updated configuration for the PRIMEPOWER platform (Ross Nobes).

2004-09-15 diabci2.2002.8
Simple program for diabatization. Uses 2x2 orthogonal matrices formed from CI coefficents.

2004-09-16 domstart3.2002.8
Just a small correction to allow for domain check in case one
uses domstart instead of start to recover domain definitions
in a local calculation.
files
ccdomain.f src/local/

2004-09-17 cosmo3.2002.8
Further corrections for cosmo from Michael Diedenhofen (24.8.04)

2004-09-17 dflmp2g_batchs2.2002.8
dflmp2 gradient via batchs2

2004-09-20 doc_module2.2002.8
Bug(s) fixed: 695
Leaves modules/doc out of documentation module

2004-09-20 doc_module3.2002.8
Bug(s) fixed: 695
Remove test for modules/doc from configure and GNUmakefile

2004-09-22 f12_more.2002.8
More code in the MP2-F12 project. Various frozen geminal models. Basis input for geminals.
New integral interface.

2004-09-23 triples_grad5.2002.8
Much more efficient evaluation of triples gradient terms implemented.

2004-09-24 aodom_fix.2002.8
accidental modification of global variable ndom in aodom.f fixed

2004-09-25 buildpatch_tail.2002.8

2004-09-26 pname_intsize2.2002.8
Bug(s) fixed: 672
Change bin/pname so that executable name includes integer size. Executable name is now molpro{s,p,x}_i{4,8}_$ARCH_$MPPLIB so that 32 and 64 bit opteron builds can be installed into the same directory. Change src/util/blastest.f to accomodate increase in filename length of rcfile.

2004-09-25 egrep.2002.8
Changing grep -e to egrep as not all systems have the -e flag

2004-09-27 dftrans_lccsd6.2002.8
Implementation of 0-2 ext transformation in lccsd.
New subroutine df_transform for all density fitting transformations.
modification in basis_integral_batchs2.f: entry init_klist can be used
to predefine the list of fitting functions, and this will then be obayed.
ccv_loc3 not yet modified.

2004-09-27 newr12.2002.8
New driver and general cleanup for explicit module. Also corrections to frozen geminals program.

2004-09-27 ri_geminals.2002.8
Beginning of code for RI geminals for Polly.

2004-09-27 veryclean.2002.8
If "make veryclean" is done in the base directory when no CONFIG file is present then the contents of the bin/ directory are no longer deleted.

2004-09-28 eom_progman.2002.8
description of ccsd/cc-eom properties program from T. Korona

2004-09-28 r12unini.2002.8
Fixes uninitialized variables.

2004-09-28 flushlog4.2002.8
open and close logfile when flush6 is called.

2004-09-28 dynpol2.2002.8
Includes code from D. Spelsberg to compute dynamic
polarizabilities with mrci

2004-09-29 gtg4.2002.8

2004-09-30 reg_inp.2002.8
Allows the user to input the atoms in the region card
by using numbers. These should agree with the order
of the given geometry.
files
local_init.f src/util/

2004-10-01 dftrans_lccsd2_dep.2002.8
Patches dftrans_lccsd2 used an rm command to delete some files and missed a depenceny to dflmp2_maxrec. A distribution built from scratch applies the patches in the wrong order. This patch deletes the files again.

2004-10-03 df_transform4.2002.8

2004-10-04 dftrans_lccsd16.2002.8
Implementation of 0-2 ext transformation in lccsd and for G(E).
ccv_loc3 is now working with direct and df.
ccv_loc3 can be used with local=4 by setting ccv_loc4=0.

2004-10-04 ncckext.2002.8
cckext for nstates at once: 1) changed cck3ext;
2) another ",1)" in all places where cckext is called
was added.

2004-10-05 load_basis_mod.2002.8
Modification of load_basis: setname now corresponds to the full basis name
if only a single default basis is used. Otherwise a unique basis name is
created as before. Some modifications of printing basis info in various places.

2004-10-05 transfix2.2002.8
fixes problems in transform with dfit

2004-10-05 doresid_loc4.2002.8
variable ISU was undefined in all2ex_loc4

2004-10-05 ncckext_cor4.2002.8
Remove xxxxxx, which was taken as a comment in intel compiler version 7, but
not in version 8 (doresid.f);
changes in dolresi.f required to handle a multi-state call to cckext

2004-10-05 pri_dflcc2.2002.8
print cleanup in df-lccsd

2004-10-05 develop_tools.2002.8
Delete object files correctly for .F and .f90 files. Improve detection of source files in develop directory when building patch

2004-10-06 dftrans_lccsd_doc.2002.8
Adds documentation to dftrans_lccsd.f. New common/cdfrec, holding
domain and record info for intergrals made in dftrans_lccsd.f

2004-10-06 maxjedomsize.2002.8
Fixes bug in computing maxjedomsize

2004-10-06 dflmp2_opt2.2002.8
Fixes missing fitting basis problem in df-lmp2 geometry optimization.
Fitting basis is regenerated for each optimization step. Currently
no dependence on igrdc

2004-10-06 triples_unini.2002.8
Fixes uninitialized variable iout at line 2276.
Replaces iout by 6 in error messages.

2004-10-06 cosmo_getstr.2002.8
Fixes calls to getstr. Replace returned arguments by variables

2004-10-07 config_pathscale2.2002.8
Compiler optimisation settings for pathscale compiler 1.3.

2004-10-08 triples_grad6.2002.8
Further work on QCISD(T) gradient implementation. More efficient
evaluation of N^7 terms.

2004-10-08 molcas_dummy_unit.2002.8
Fix calls to getvar in molcas_dummy so that unit is a variable not a constant

2004-10-08 df_conv_eom1.2002.8
DF for conventional (EOM)-CCSD for J(E),K(E)- and K(D)-type terms,
using DF subroutines of M. Schuetz for 3- and 4-ext integrals.

2004-10-09 old_slapaf.2002.8
Splitting old version of ddv from Molcas so that it can run in parallel with new version

2004-10-11 cceom_cor.2002.8

2004-10-11 df_eom_test.2002.8
testjob for df-eom-ccsd

2004-10-11 c3ext_unde.2002.8
include "common/cpar" to have the variable local defined

Created by Martin Schütz on 2004-10-12 sapt_test.2002.8
New Makefile for testjobs including dft-sapt test jobs

2004-10-12 explicit_mp2.2002.8
declare character str used in call to getvar to avoid segmentation fault

Created by Martin Schütz on 2004-10-13 sapt_induc3.2002.8
more efficicient method to compute induction terms

2004-10-13 testjobs_depend.2002.8
Bug(s) fixed: 706
Fix dependencies in testjobs/Makefile

2004-10-13 dftrans_lccsd_loc2.2002.8
Implements local fitting in dftrans_lccsd.
Fixes memory management in ccsd_loc for ccv_loc3 (but
still full Kij matrices used in ccv_loc3).
Cleans up some print,cpu stuff

Created by Martin Schütz on 2004-10-14 df_ltrip.2002.8
local triples with DF

2004-10-14 buildpatch_conflict2.2002.8
Quicker resolution of conflicts when applying patch. Should speed up the application of patches, especially those with many files.

2004-10-15 dplace.2002.8
SGI configuration changes and use of dplace within molpro.c to acheive better memory performance. Lowers optimisation on opt/frequencies.f util/zmatrix.f for intel itanium compiler version 8.

2004-10-14 PreMolcas.2002.8
Changes in preparation for Molcas updates

2004-10-15 Molcas01a.2002.8
Incorporating Molcas changes

2004-10-15 sapt_aix.2002.8
Trivial fix for src/sapt/dfdftsapt.f so that it compiles on AIX

Created by Martin Schütz on 2004-10-15 df_ltrip_fix1.2002.8
local triples with DF, fixes

2004-10-15 transfix3.2002.8
Fixes bug introduced by dftrans_lccsd_loc2; wrong check of icfit_ccsd

2004-10-17 syscon_memmax.2002.8
Increases memmax and memdef in syscon.f to enable debugging with large memory

Created by Martin Schütz on 2004-10-21 d4extfix.2002.8
fixes problem with bucketsort in d4ext

2004-10-22 patcher_config.2002.8
Fix buffer overflow in list of parent patches. Writes username and password to config file only if a successful download occurs.

2004-10-23 cosmo_mpp2.2002.8
Bug(s) fixed: 694
Fixes of cosmo for mpp

2004-10-23 bug713_triples_openmp.2002.8
Bug(s) fixed: 713
Fix zero-divide exception in openmp directives

2004-10-24 neg_pointers.2002.8
Allow negative pointers.

2004-10-25 PreMolcas01b.2002.8
Changes in preparation for Molcas01b

2004-10-25 171 2004 parse_x86_64_i4.2002.8
Opteron parse object

2004-10-25 ia64_intel8a.2002.8
Bug(s) fixed: 616
Some configuration changes so that configure works with version 8 of the intel compiler

2004-10-25 syscon_print.2002.8
Fixes incorrect print of secondary working directory if its name is longer than the primary one

2004-10-25 molpro_linux4.2002.8
Adds script molpro_linux which runs correct molpro on different linux hardware
Adds a program swapdisks which swaps scratch disks depending on molpro job count (stored in /tmp/MOLPRO_CONT)
Modify swapdisks so that the least used disk is selected as primary one.
Modify linux_molpro: -current added, -alpha corrected. Add path to check of VERSION.
Swapdisks called a second time after molpro finished.
This removes the current job from the list. Fixes wrong behaviour if several runs
follow each other in one batch job.

2004-10-26 corlsi.2002.8
Attempt to fix corlsi for the case of negative pointers

Created by Martin Schütz on 2004-10-26 fix698.2002.8
Bug(s) fixed: 698
all dgemm calls replaced by mxma/b in dft-sapt

2004-10-26 dummy_na2mem2.2002.8
Reverts patch dummy_na2mem since this problem has now been
fixed in patch Molcas01b.

2004-10-26 t2tran_trip.2002.8
Makes a simple modification in t2tran_triples_add to reduce I/O
formally from N^6 to N^4.
Reduces elapsed time from 113 to 44 sec for H2O/vqz on my laptop,
but is still very unefficient since 1 word at a time is read!

2004-10-28 casvb_report.2002.8
Fixes two potential bugs during reporting of CIWEIGHTS analysis.

2004-10-29 blkbinsort_fix3.2002.8
some fixes concerning type conversions, probably compiler bug iidentified

2004-10-29 d4extfix2.2002.8
memory fix in D4EXT

2004-10-29 cc3ext.2002.8
uses precomputed 3-ext integraÃls also for singles terms in case of
cfit_d3ext=2

2004-10-29 loccis.2002.8
Make local CIS: ground state diagonal (i,i) domains are
used for CIS excitations; operator domains and incomplete operator
lists can be used.

2004-11-01 mxmsu.2002.8
Bug(s) fixed: 595
Fix msmsu parameter overflow in geometry optimiser

2004-11-01 ia64_darwin.2002.8
Lower optimisation on Molcas/darwin.f for efc version 7

2004-10-29 dpocon.2002.8
Bug(s) fixed: 684 685
Fix 32 bit wrappers in 64 bit IBM version for dpocon call in explicit_3idx_ints.f

2004-11-02 dftrans_lccsd_robust13.2002.8
Implements robust fitting for 0,1,2-ext in dftrans_lccsd
Skips empty contributions in robust fitting and exploits
in dfasmbl_loc1.f if A and D are same quantities.

2004-11-02 opkdom.2002.8
The k-op domains were too small in some cases since the algorithm
in opkdom was not symmetric with respect to k and l. Also, up0dom
are now the united domains for strong pairs only, unless maxitcls.gt.1
(so far, these were united domains for strong and close pairs).

2004-11-03 depend_fh.2002.8
Bug(s) fixed: 721
If makedepend finds a file in same directory as a #include <file> it is added to depend file. Intended for molcas .fh include files

2004-11-03 dftrans_lccsd_defaults.2002.8
Makes robust local fitting with extended orbital domains the default in dftrans_lccsd

2004-11-04 buffread3.2002.8
buffered read for 3ext integrals in lccsd

2004-11-04 ftcflush2.2002.8
Adds lapack3.o to the ftcflush list in src/GNUmakefile

2004-11-04 mkl701.2002.8
Adds support for mkl701; also adds /software/linux/lib to libroot specials for Stuttgart

2004-11-04 Molcas01e.2002.8
Molcas tidying up.

2004-11-05 dflccsd_pathf90.2002.8
Fixes to get df-lccsd working with pathf90/i8:
1.) fix bug concerning linv (did not depend on intrel)
2.) Fixes in df3ext_lccsd: constants nams and iofs where passed but modified in the code -> crash!

2004-11-05 pathscale_makedepend.2002.8
Bug(s) fixed: 722
Change options to makedepend for pathscale compiler and version 8 of intel compiler so that module depencies treated correctly

2004-11-06 displace4.2002.8
Bug(s) fixed: 716 719
Make symmetric displacement coordinate algorithm more robust by
choosing a potential energy function that is closer to equilibrium
to avoid problems introduced by rotational contamination.
displace4: adjust results for h2o_numfreq.test, c2h4_ccsdfreq.test

2004-11-06 ccv_loc3c.2002.8
Further improvements of ccv_loc3 for local=4 case.
Now uses local KOPs and requires less memory and CPU.
loc3b fixes bug affecting loc_eom3.test.
Change ifil_k0ext from jopfil to kopfil

2004-11-06 diag2_dsyev.2002.8
Determine optimum work for dsyev in diag2 and try to allocate
it dynamically if fixed work array is too small.

2004-11-06 basis_forces_dealloc.2002.8
Adds dealloc at the end of basis_sqr_forces_sqr_mem

2004-11-06 prepin_inquire.2002.8
Bug(s) fixed: 656
Inquire existence of include files in prepin.f

2004-11-06 scfpr2_enest.2002.8
Disable enery estimate for 1-electron systems

2004-11-07 syscon_tuning2.2002.8
Updates tuning parameters in syscon.f
Adds support for mkl, atlas, acml libraries and p3, p4, Athlon, and Opteron processors.
Adds subroutine cpuinfo

2004-11-07 makerc_blocksize.2002.8
Increases the default block size from 32 to 64

2004-11-08 opkdom4.2002.8
modifies opkdom again. Old results are recovered. Previous patch opkdom
was incorrect (larger domains than necessary).
This patch also improves the performance of ccv_loc3.f
Cleans up ccdomain: ioff_opkdom, ioff_opjdom, ioff_opjedom were used
in unipairdom if undefined (iopdom=0). Consistently modifies
ccv_loc3 and ccv_loc4 not to use any ioff_opkdom etc if iopdom=0.

2004-11-08 libdir2.2002.8
Sensible initial choice for location of global arrays library

2004-11-08 loc_eom.2002.8
improvements in local EOM-CCSD

2004-11-09 h2o_dfeom2.2002.8
Fixes error for the test option NODF for DF-EOM-CCSD.
Restores some lost functions of PSPACE>2 eomprint option.

2004-11-09 newr12ints2.2002.8
A completely new program for doing all sorts of R12 or F12 theories with
integrals called by name and automatically assembled. This is FRM's attempt
to clear up the horrible mess that he made with newr12ints. Apologies.

2004-11-11 numhess_mppx2.2002.8
Implements parallel evaluation of numherical hessian in mppx mode
Fixes bug in numforces (illegal symmetrization of hessian in eqsym_hess).
Modifies some testjobs due to very small numerical differences
caused by the above change.

2004-11-11 pname_mppx.2002.8
pname did not evaluate the program name correctly in the mppx case
since both MPP and MPPX are set in CONFIG, and it checked for MPP first,
using elseif for MPPX.

2004-11-11 molpro_linux_mppx2.2002.8
New option -mppx in molpro_linux. If this is given, the mppx version is used.

2004-11-11 configure_mppx2.2002.8
Installation directory names were wrong in the mppx case. mppx
must be checked before mpp!

2004-11-11 multi_dynweight.2002.8
Implements dynamical weighting in multi as described in J. Chem. Phys. {bf 120}, 7281 (2004)

2004-11-11 syscon_mppx2.2002.8
Appends processor number to logfile in mppx case
Gets cpuinfo only on processor 0 in order to avoid
problem with creating file cpuinfo

2004-11-14 scf_defocc.2002.8
Sets in closed-shell case iclos=iocc if OCC card is present. This
avoids error if the occ card changes in successive closed-shell calculations.

2004-11-14 Molcas_mppx.2002.8
Settings of mpp_state only in cstart mpp sections

2004-11-14 mppx_parallel.2002.8
Implements parallel execution of do-loops in mppx mode.
Parallel section starts with command 'parallel' and ends
with 'sync,variables..'. This synchronizes and broadcasts the
given variables. Only commands in 'strong' (driver.f) are skipped, i.e.,
variable settings and logical is performed on all processors consistently.
All settings of mpp_state must be within cstart mpp/cend
since mpp_state=2 in mppx mode throughout.

2004-11-15 ccv_loc3d.2002.8
Fixes memory allocation problem in ccsd_loc for special case H2

2004-11-15 leom_jhdom.2002.8
Version with domains for J(Hij)/K(Hij) operators works now.

2004-11-15 scfocc.2002.8
Improves algroithm for determineing occupation and closed shells.
Now works for oszillating cases like O3.

2004-11-16 mppx_parallel4.2002.8
Further modifications for parallel:
1.) modifies prepin to treat command={} as procedure and command{} as command block
2.) skips command block input in parallel mode if command is not executed
3.) Fixes typo in ccmp2_loc.f

2004-11-17 filsiz_3x4x_2.2002.8
computes file sizes for 2idx 2ext integrals for 3x,4x

2004-11-18 leom_next2.2002.8
Next improvements/error_fixes in local EOM:
1) error in yijk_loc if there is
a true sparse list for the k index.
2) eskipijkl is used now for alpha_ij,kl.
Remove some obsolete subroutines. Clean excessive debugs.
Put some debug printing subroutines to a separate file.
Move alpha_loc3 etc. from ccv_loc3.f to eom directory.

2004-11-19 cphf_thresh5.2002.8
Various modifications to avoid symmetry braking problems in geometry
optimizations and frequency calculations.

2004-11-19 dfasmbl_coul.2002.8
Fixes possible bug in dfasmbl_coul (variable nkk not initialized if integral batch is empty)

2004-11-19 missinginterface.2002.8
Fixes missing interface command in the explicit directory.

2004-11-19 lsints08.2002.8
lsints08 fixes two not inizialized variables
lsints07 fixes lsints06 conflict with mppx_parallel
lsints06 fixes compilation problem on opteron - all f90 pointer are removed
lsints05 changes respect to lsints04:
-> further cleaned up
-> all parallel
-> Fock routines
-> Corresponding changes to MRCI routines
SEE src/lsint/readme for more details
lsints04 fixes bugs of lsints03:
-> code cleaned up
-> fixed compilation bugs on IBM and HPUX
-> removed reading input by seward; now it is done through inisewm as standard
readme from lsints03:
--------------------------
lsints03 fixes bugs on IBM:
in drv2el_ls some variables not declared logical, were assigned logical values
in seward_ls, in the ss_get_sym , was unclear compilation error
readme from lsints02:
--------------------------
This patch completely replaces LSINT code by P. Palmieri. The old code can
be still used by specifying variable LSTYP=CADPAC, while default LSTYP=SEWARD
calls new routine. In the new routine there are no resctrictions on basis set
type. The new routine accepts all SEWARD keywords. General calling sequence
is
LSINT[,X][,Y][,Z];[ACC,CUT,TOL;]
here X, Y and Z (optional) specifies spin-orbit components to calculate,
ACC(optional) specifies CUT and TOL threshold.

2004-11-20 intface_mpp.2002.8
Skip intface except on processor 0

2004-11-21 numhess_mppx4.2002.8
Fixes uninitialized variable in numhess_mppx.f

2004-11-21 newvars.2002.8
Partly new implementation of getvar.f. Variable arrays can now be of mixed
type (i.e. some elements can be strings, other floating numbers). In assignments
of variables to other variables undefined elements are assumed to be zero.
Also, invalid expressions yield zero. This avoids error exits and is useful
in mppx mode. The format of the variable record changed, but downward
read compatibility is granted. The argument list of checkvar has changed,
and some files are modified accordingly.

2004-11-21 eom_alpha.2002.8
alpha intermediate with new lists in local EOM-CCSD

2004-11-22 pseudo_libmol2.2002.8
New Pseudo.libmol from H. Stoll, Oct. 2004
pseudo_libmol2 corrects some errors in the previous version.

2004-11-22 lsints09.2002.8
lsints09 fixes bug with TOL/CUT parameters in sodriver
lsints08 fixes two not inizialized variables
lsints07 fixes lsints06 conflict with mppx_parallel
lsints06 fixes compilation problem on opteron - all f90 pointer are removed
lsints05 changes respect to lsints04:
-> further cleaned up
-> all parallel
-> Fock routines
-> Corresponding changes to MRCI routines
SEE src/lsint/readme for more details
lsints04 fixes bugs of lsints03:
-> code cleaned up
-> fixed compilation bugs on IBM and HPUX
-> removed reading input by seward; now it is done through inisewm as standard
readme from lsints03:
--------------------------
lsints03 fixes bugs on IBM:
in drv2el_ls some variables not declared logical, were assigned logical values
in seward_ls, in the ss_get_sym , was unclear compilation error
readme from lsints02:
--------------------------
This patch completely replaces LSINT code by P. Palmieri. The old code can
be still used by specifying variable LSTYP=CADPAC, while default LSTYP=SEWARD
calls new routine. In the new routine there are no resctrictions on basis set
type. The new routine accepts all SEWARD keywords. General calling sequence
is
LSINT[,X][,Y][,Z];[ACC,CUT,TOL;]
here X, Y and Z (optional) specifies spin-orbit components to calculate,
ACC(optional) specifies CUT and TOL threshold.

2004-11-23 numhess_mppx7.2002.8
Modify print format in mppx mode. More digits for CPU.
Total CPU (user+sys) is now printed.
Append processor number to punch file (if specified in input)
Open file inp correctly in mppx mode

2004-11-23 global_err.2002.8
Fixes subroutine global_err: msg was undefined and print format incorrect.
Now include processor number in error message.

2004-11-23 f12bighack.2002.8
A major overhaul to the new integral program which, apart from just being
better, is flexible enough for the integrals needed for example in DF-RI
geminals.

2004-11-23 newvars3.2002.8

2004-11-24 mppx_restart2.2002.8
Implements restart possibility in mppx mode. Permanent files are only
allocated on processor 0, otherwise temp files are used. The content
of existing permanent files is broadcasted to all processors.

2004-11-24 cphf_general.2002.8
Avoid saving to negative record in cphf_general

2004-11-25 m2c.2002.8
If .mod file is created after .o file then default rule in make for Modula-2 files will cause build to fail

2004-11-25 mpp_aix.2002.8
Adds global_{create,destroy}_mutexes to common/ga.h

2004-11-25 numhess_mppx10.2002.8
Change iprocs to iprocx in globalop.f. Add lock and fence
routines. Modify numhess_mppx to use ga and implement
restart.

2004-11-26 licencekeyfile.2002.8
First attempt at separating licence key from CONFIG and configure

2004-11-26 licencekeyfilemaint.2002.8
adding extra files to parse out of distribution files

2004-11-28 leom_direct4.2002.8
Adaptations for direct local EOM-CCSD with domains:
1) change ccemat_loc (it takes T1 if nr_eom.le.0 and
U1(nr_eom) if nr_eom>0;
2) cckext called at the beginning of a new iteration now, since
G(H) is made in cckext for direct;
3) nr_eom set to -1 in eom_init
4) move all from doresid_loc4.f to doresid_loc.f

2004-11-30 patcher0700.2002.8
Please type one line for each file to be contained in the patch

2004-12-01 r12_merge.2002.8
Further merging of mp2-r12 methods to use new integral structure.
Also implements new version of f12 method when basis ri=ao.

2004-12-02 ump2_direct2.2002.8
Exit with error message if ump2 or rmp2 is attempted in direct mode (not implemented)
Exit with error if any other ci or ccsd calculation is attempted with UHF orbitals.
This error exit can be avoided by specifying orbital,type=uhf, and corresponding
instructions are printed before the error exit.

2004-12-02 mscaspt2_gradients2.2002.8
The patch implements a new multireference/multistate caspt2 code
and also implements analytical gradients for it.
It also fixes a few bugs in the old
singlereference/multistate caspt2 code (Roos approach).
Code works also in presence of level shift (extra Lagrangian
multipliers are added).
Present problems:
1) The situation singlestate/multireference is very inefficient.
Strictly speaking this situation would not require CPCASPT2
calculation. The present code however DOES run it: the result is
of course a set of lagrangian multipliers that are identical to the
first order wavefunction coefficients.
2) The code does NOT correct singlestate/singlereference calculations
for level shift effects even though the code is practically all there.
Consider the following situation.
A CASSCF wavefunction is produced averaging together several states of
symmetry S1 and 1 state of symmetry S2. One can optimize with level shift
correction any state of symmetry S1. But one cannot optimize the state in
symmetry S2 with level shift correction because that corresponds to a
singlestate/singlereference calculation (old code). One must always
allow for mixing in all symmetries which means, at least 2 states in each symmetry.
4) Recalculated and printed total energy in 'cirsamat_ms' is wrong
even though gradients are correct
5) Incorrect CASSCF dipole moment is printed in 're2den'. MS-CASPT2
dipole moment in 're2den' is correct.
6) Input formats for both MSCASPT2 approaches and gradients options
are described in MultiStateCaspt2.f

2004-12-02 acml2.2002.8
Add preferred acml blas option for opteron

2004-12-02 newvars4.2002.8
Fixes undef iout. Replace iout by 6 for error messages.

2004-12-02 fix_unini2.2002.8
fixes potentially dangerous uninitialized variable (thanks Joachim)

2004-12-03 H_tzvp.2002.8
H tzvp ahlrichs basis

2004-12-03 f12bighackdep2.2002.8
patch f12bighack did not have dependencies for deleted files

2004-12-04 mscaspt2_test.2002.8
Remove test h2o_mscaspt2_opt.test from testjobs/Makefile until mpp problem is fixed

2004-12-06 mpp_print.2002.8
SPLIT WORK and DISTRIBUTION info printed for only iprint(39).ge.3
rather than iprint(39).ge.1

2004-12-06 patcher_token.2002.8
retrieve licence token from website

2004-12-07 connectivity_ecp2.2002.8
Modifies connectivity.f to propertly handle atoms with ecps.
Must get charge from natmic, not from zcharge or charg.
Fix problem with special case fitcom_ccsd=3

2004-12-07 config_token.2002.8
Retrieve licence key from website during configure if not in cache. Remove key from molpro.rc

2004-12-11 optg_fix2.2002.8
Fix bug in optg: variable iroot used as pointer and for root!

2004-12-11 dflmp2g_defaults4.2002.8
Modify and correct settings for df-lmp2 gradients:
1.) Pass thresholds and other parameters in variables to force program, since otherwise input is lost
2.) local fitting in cphf and scfgrd disabled if locfit=0 in scf or lmp2 input
3.) locfit_cphf and locfit_scfgrd are aliases for CFIT_DOMAUX_CPHF and CFIT_DOMAUX_SCFGRD
4.) Different thresholds possible for cphf and scfgrd
5.) harder default screening thresholds for gradients and frequencies, also new default rdomaux_scfgrd=5
6.) new subroutines cfit_def1 and cfit_def2 to set defaults and final parameters for dfit
7.) Default threshold for cphf is 1.d-6 again.
8.) Default for rdomaux_cphf set to 3.0; was 0.0, and this caused errors even for h2o. But
still the program is incorrect for locfit_cphf=1 in larger cases, independent of rdomaux_cphf!

2004-12-13 token_frontend2.2002.8
If not set on command line then molpro.c reads licence token from location in lib or home directory

2004-12-13 ccsd_loc_undef.2002.8
Fixes opdom->iopdom in one place of ccsd_loc (affects eom only)

2004-12-13 msrs2_unini.2002.8
Fixes undefined variables in 2 places

2004-12-17 no_gpt.2002.8
removes the allocation of the gpt stuff in df-dft-sapt and df-lmp2 gradient

2004-12-18 mkl72.2002.8
Adds support for mkl72 to linux.conf

2004-12-18 dgemm_zero2.2002.8
Initializes result matrix of dgemm to zero (in mxm.f and mxmdgm.f)
Also does this in blas4.f

2004-12-20 lmp2_maxit.2002.8
Do not do hylleraas step if maxiti is reached in diis case (non-dynamical update)

2004-12-21 atdens_ecp4.2002.8
Implements density starting guess for ecp. Density guess only up to Kr,
and only ecp10 and ecp28 implemented so far.
Modify dummyecp.test to force starting with eigenvectors of h0.
Program does currently not trap case that some atoms have generally
contracted bases, others not.
Add support for heavy atoms (until element 118)

2004-12-22 bug740fix.2002.8
Bug(s) fixed: 740
fixes symmetry breaking problem of df-lmp2 gradient (local fit domains)

2004-12-23 newvars5.2002.8
Fixes bug in getvar.f causing creating an excessive number of variable values

2004-12-24 jkeij_2.2002.8
simultaneous construction of J(eij) and k(eij), if enough memory

2004-12-27 iotest.2002.8
Implements keywords twoonly and noseek in iobench

2004-12-27 fix744.2002.8
Bug(s) fixed: 744
fixes problem with negative pointer

2004-12-27 subr_stack.2002.8
subr stack for pricall/pridone and friends now done within a module,
since the subr stack was apparently contaminated at some occasions, i.e.,

2004-12-27 checkcommands.2002.8
Implements command.registry and input checking

2004-12-27 aronegr.2002.8
Adds missing call to add_mpptim, avoid error exit in ecp gradient calculation with print,cpu

2004-12-29 tripmemfix.2002.8
memory fix for df-lccsd(t)

2004-12-30 licenser.2002.8
Licenser updates

2005-01-03 scfocc2.2002.8
Fixes bug in scfocc

2005-01-04 orbdom_format.2002.8
Adds missing comma to format statement in orbdom1

2005-01-04 ri_lmp2g_format.2002.8
Adds a missing comma to format at line 2775

2005-01-04 driver_new5.2002.8
driver.f completely rewritten and modularized

2005-01-05 develop_token.2002.8
adjust develop/profile and develop/cshrc to get licence token from
$HOME/.molpro/token

2005-01-05 field2.2002.8
Bug(s) fixed: 746
field,dmz,0 now really does give a field strength of zero, just like
dip,,,0
in contrast to field,dmz or dip; which give the default strength (1).

2005-01-06 libmol_crash_mpp.2002.8
Bug(s) fixed: 743
Proper error exit for mpp when basis library entry not found

2005-01-11 checktoken2.2002.8
Checks that downloaded token contains required stanzas and has not expired

2005-01-11 checkopt3.2002.8
Implements modular processing and prechecking of options.
Options on command lines now processed in get_commandset.
Setup direct.registry, optg.registry, freq.registry, ksopt.registry, aoint.registry, locopt.registry
(most not yet used except for checking)

2005-01-11 testjobs_order.2002.8
Reorder sequence of tests in testjobs/Makefile to do (i) local and lccsd earlier and
(ii) longtest last (and in this the longest last).
Add lccsdtest to the normal test list.

2005-01-11 for.2002.8
Bug(s) fixed: 752
Fixes to buildpatch/makepatchdef: a) common/cvec now matched by makepatchdef,
and handled properly in buildpatch; b) header rewriting now more robuts, and
copes with missing version number or multiple spaces.

2005-01-12 dft_options2.2002.8
Uses ksopt.registry to scan for valid functionals. Uses get_inpl to access
command-line input for functionals. Fixes problems with mixed functional
and density fitting input. Some minor print changes in direct_init and getvar.

2005-01-12 basediff2.2002.8
Bug(s) fixed: 752
Circumvent problem with output from cd command in basediff

2005-01-12 vfi.2002.8
New 3-index integrals program, VFI. Many restrictions still exist
so general use is not yet recommended. Improvements and docs to
follow...

2005-01-12 casvb_bug754.2002.8
Bug(s) fixed: 754
Fix illegal branch into an elseif block due to typo. Label should
have been "12" instead of "2", but the two lines made identical
assignments.

2005-01-12 casvb_commandset.2002.8
Adds commandset='CASVB' definitions to commands.registry, with associated
changes to casvb input. Trivial update to readcommand.f. Replaces two
explicit BACKSPACEs with appropriate calls. Updated casvb2 testjob.
Updated documentation will follow soon.

2005-01-12 dc.2002.8
check for calculator program dc needed by configure

2005-01-12 fix756.2002.8
Bug(s) fixed: 756
Fix an error appearing with NEC f90 compiler

2005-01-12 casvb_commandset2.2002.8
Moves some misplaced files to their correct locations. Sorry for the previous
error.

2005-01-12 casvb_doc2.2002.8
Updated documentation following recent changes to input structure.

2005-01-13 subr_stack_revert.2002.8
Reverts changes make in patch subr_stack since these are unnecessary and do not work with pgf90

2005-01-13 dftfunc_inp2.2002.8
Bug(s) fixed: 663
Uses ksopt.registry to scan for valid functionals. Uses get_inpl to access
command-line input for functionals. Fixes problems with mixed functional
and density fitting input.
Moves dftini to subroutine anfa, after processing of command line. This fixed bug 663.

2005-01-13 driver_lattice.2002.8
Moves call to lattice from molpro_aux to run_job

2005-01-13 patcher_nec.2002.8
Bug(s) fixed: 757
incorrect syntax in head command called from patcher on nec

2005-01-13 gridorb_pot.2002.8
Use drvprop direct code to calculate potential for cube file

2005-01-13 logical_aix2.2002.8
Added common/cdirect to molpro_aux.f as logical direct is used but not defined.
Fixed undefined logical contract_a_first in vfi.f.

2005-01-14 sx_updates.2002.8
Updates for SX (build and runtime).

2005-01-14 df_lccsd_defaults.2002.8
make cfit_d3ext=2, cfit_d4ext=2 the default.
Switch to cfit_d3ext=1 or cfit_d4ext=1 if file size larger than 15 GB
Compute file size and number of triples in double precision to avoid
integer overflow.

2005-01-14 sx_stat_h.2002.8
NEC compiler sensitive to location of include statement

2005-01-14 driver_lattice2.2002.8
fixes Bug occuring with print,cpu and lattice.

2005-01-15 iotest3.2002.8
Implements keywords twoonly and noseek in iobench

2005-01-14 sl_vars.2002.8
variables with prefix SL_ no longer set in restart, otherwise chaos

2005-01-16 ccv_loc3_default.2002.8
Fixes cisd, qcisd, as well as various other bugs in ccv_loc3.f.
Now ccv_loc3 and ccv_loc4 should give exactly the same results
for all opdom and check1ex cases. Previous small energy differences
with ccv_loc3 and ccv_loc4 were due to inconsistent neglect of K(Dij)(k)).
Removes call to mod_opjdom in ccdomain.f, otherwise error in ccv_loc3.
Compute lenk for all cases iopdom.gt.0 in cifile.f, otherwise inconsistency
error in ccsd_loc. Makes ccv_loc4=0 (i.e. use of ccv_loc3) the default.

2005-01-17 optgeo_dflmp2.2002.8
Fixes bug in optgeo and frequencies in handling density fitting calculations

2005-01-19 develop_sharedlib.2002.8
Bug(s) fixed: 707
If using MOLPRO_SHAREDLIB=0 then delete soft link libmolpro.a in develop directory which could be pointing to the shared library.

2005-01-19 gtg5.2002.8

2005-01-19 configure_nolapack.2002.8
Bug(s) fixed: 739
Don't put lapack into CONFIG if removed or if -nolapack flag given to configure

2005-01-19 cdollar1.2002.8
Bug(s) fixed: 717
Removing c$ comments as it clashes with openmp compiler

2005-01-19 makerc_blaslib.2002.8
Bug(s) fixed: 723
MOLPRO_BLASLIB not set correctly in bin/molpro.rc if commented out BLASLIB appears in CONFIG

2005-01-19 bug747.2002.8
Bug(s) fixed: 747
In src/molpro.f inquire whether there are open files before closing. Fix from Ross Nobes.

2005-01-19 lenstr_befehl.2002.8
Increase length of strings befehl, stand, zstand, molen to 16
Fixes some entries in commands.registry
Cosmetical change of printing in term_job

2005-01-19 basis_default.2002.8
Now allows default=basis in basis block.
Gives an error message if multi-line default basis input is attempted.