MCTDH: Changes of the Code

• Changes in Version 8.4
• Changes in Version 8.3
• Changes in Version 8.2
• Changes in Version 8.1
• Changes in Version 8.0
• Changes in Version 7.1
• Changes in Version 7.0
• Changes in Version 6.4
• Changes in Version 6.3
• Changes in Version 6.2
• Changes in Version 6.1

Back to Main Menu

What's New

Version 8 Release 4

Revision 9, (November 2013) (Public: Nov 22, 2013)

• ---- Version 8.4.9 released. (Nov 22, 2013) ----
• The HTML-docu for Improved-Relaxation is updated. See Run-Section.
  (HDM 11/13)
• Keyword realphi added to Run-Section. Useful in RDAV-runs when FFT or PLeg is used. Also keyword realpsi added to Init_WF-Section.
  (HDM 11/13)
• Continuation lines are now possible in the Hamiltonian tableau.
  (HDM 06/13)
• Wigner matrices implemented.
  (PST 06/13)
• KLeg/PLeg operators, jpjm = j_+ * j_- and jmjp = j_- * j_ are implemented.
  (HDM 04/13)
• The embedded copies of BLAS and LAPACK are updated.
  (FO 03/13)
• The vpot generation of POTFIT has been MPI-parallelized.
  (FO 02/13)
• Parallel builds (compile -j) now work without problems.
  (FO 12/12)

Revision 8, (November 2012) (Public: Nov 28, 2012)

• ---- Version 8.4.8 released. (Nov 28, 2012) ----
• MPI communication is optimized in various subroutines. The keyword summf2 is removed as summf2 is now default. The old standard is established through the keywords no-summf2 or summf1.
  (MS 10/12)
• The keyword cutalong is added to the run-section of potfit. It requests to perform a cut along a specified DOF of the PES (both exact and fitted).
  (DPR 08/12)
• A problem with flux is fixed. flux84 did not work correctly, when the option -es was used and a full projection was made.
  (HDM 07/12)
• Option -lin added to autospec. If set, the box-filter g0 is replaced by a linear filter g(t) = 1 - t/T. The filters g1 and g2 are then replaced by linear combinations of the linear filter with cos or cos^2 filters, respectively. In plspec these filters are accessible via the option -g3, -g4, or -g5.
  (HDM 07/12)
• New keyword print-npot added to operator-section. Most of the natpot-lines are now suppressed in the op.log file, print-npop restores the previous behavior.
  (HDM 06/12)
• Some tuning for the Runge-Kutta integrators to achieve better step-size predictions. This may require a slightly better accuracy setting.
  (FO 03/12)
• New analyse program npotminmax implemented. It is similar to vminmax but works directly with natpot files.
  (MS 03/12)
• Option -E added to showd1d. It shows the population of all grid-point over time. This may be used to plot electronic populations of single-set runs.
  (HDM 01/12)
• Modified statistical analysis so that weighted trajectories can be used. Documentation of statistical analysis is updated.
  (MS 02/12)
• A problem with plstate is fixed. This includes a modification of rdcheck.F
  (HDM 11/11)

Revision 7, (November 2011) (Public: November 03, 2011)

• ---- Version 8.4.7 released. (November 03, 2011) ----
• Sections 6.6 and 6.8 of the guide (how to install new functions/surfaces) as well as the related files user1d.F and usersrf.F are improved.
  (YCC and HDM 11/11)
• Keyword twall and option -twall added. This allows to stop a run after some elapsed wall-time (real time). Useful for parallelized runs.
  (HDM 10/11)
• Several new keywords are added to POTFIT. Some of them are needed or useful for MGPG runs only.
  • Command pes-file (Syntax: pes = pes-file{PATH}). Potfit is requested to read a pes file located at PATH.
  • Command cutnpot A previously existing natpot file is read and the number of natural potentials is reduced to the values given in the NATPOT-SECTION in the potfit input file.
  • Command rdnpot A previously existing natpot file is read. Used mainly in conjunction with cutnpot.
  • Command sp-energy = n₁...n_f Potfit is requested to perform a single point energy calculation for a given set of DVR indices n₁...n_f. The outcome is the PES energy and, if a natpot file is available, the approximated potfitted energy.
  • Command dvronly The program generates the DVR files (ASCII and binary) and then stops.
  • Command nat-weights The program generates the 'nat-weights' ASCII file which includes the natural weights.
  • Command no-dcoeff The computation of the contracted coefficients (i.e. D tensor) is avoided.
  • Command nofiterr The calculation of errors is disabled. Useful in the case of large systems.
  • Command printrho The program generates the 'rho-ascii' ASCII file which includes the density matrix elements.
  • Command wrallevec The density matrix for ALL modes (Note: including modc) is computed and diagonalized. Then, the resulting natural potentials are written to evec (ASCII) and evecb (binary) files.
  • MGPF Command fullcoarse A full representation calculation (i.e. m=N_coarse) is performed. Natural potentials for each particle are written to the binary files cfull<mode>. Note that the D tensor is not computed.
  • MGPF Command onlydcoeff. Performs a computation of contracted coefficients. The D-tensor is written to the binary file fdcoeff. Note that no natural potentials are generated.
  • MGPF Command finemode. The natural potentials for a single mode are computed and are written to the binary file fmod<mode>.
  • MGPF Command ext-natpot Read natural potentials and contracted coefficients from the external files ext-natpot and fdcoeff, respectively, and build a correctly formated natpot file.
  (DPR 10/11)
• • dvrweights.f modified to avoid overflow when very long grids are used.
  • install_mctdh adjusted to allow longer path length.
  
  (HDM 09/11)
• Formats for sys-time changed in timing.F to avoid '***' output.
  (HDM 08/11)
• Bug fixed in flux. Appeared when both options -O and -es are used.
  (HDM 08/11)
• Cluster-expansion related python tools. These include:
  • statistical analysis ('clusterstat' and 'clusterana')
  • sampling of clusters on a DVR ('clusterbuild')
  • tool to test PES files generated by MCTDH ('chkpes').
  
  (MS 06/11)
• LMR (left/middle/right) operators implemented. This allows for using operators of the form:
  d/dx * v(x,y) * d/dy
  where (x,y) are in a combined mode.
  (HDM 04/11)
• Fix library routine dtxxyz. The first gridpoint had been forgotten.
  Previous calculations with operator terms of htmtype=4 should be checked.
  (FO 02/11)

Revision 6, (January 2011) (Public: January 07, 2011)

• ---- Version 8.4.6 released. (January 07, 2011) ----
• Derivative form of flux evaluation (-d option) added to flux84.
  (FO 01/11)
• vminmax: added options to run over selected grid-points (trajectory, -t option) and to compare PES with ASCII file (-c option).
  (MS 11/10)
• The file "input" of POTFIT is made read-only (similar to MCTDH).
  (HDM 11/10)
• Block-Davidson when using "rlxemin=.." performs now better. The treatment of already converged states was not optimal.
  (HDM 11/10)
• Memory allocation modified to allow to allocate more memory for analyse routines and potfit.
  (HDM 11/10)
• • mplementation of multi-target operators for optcntrl.
  • Example inputs added to inputs/optcntrl/.
  • Keyword "usepack" added for Hamiltonian-Sections.
  
  (MS 09/10)
• Option -max added to rdacoeff. rdacoeff can print the number of coefficients with absolute value between thr and max.
  (HDM 09/10)
• Some max-parameters in include files increased.
  maxkoe, maxhtm, maxhop: 2000 → 9600; maxrdf: 16 → 64 ; maxpes: 24 → 92 ; maxham, maxmuld: 32 → 64.
  (HDM 09/10)

Revision 5, (June 2010) (Public: June 25, 2010)

• ---- Version 8.4.5 released. (June 25, 2010) ----
• flux84 and wfproj84 can now handle wavefunctions for identical particles (id keyword).
  flux84 now writes an iwtt file (integral over wtt).
  (HDM 06/10)
• Block improved relaxation for complex Hermitian Hamiltonians is MPI parallel.
  (MRB 06/10)
• The keyword split-rst is now recognized in geninwf runs as well.
  (HDM 05/10)
• There is a new keyword reflex (Run-Section) which enables to make use of symmetry in e.g. malonaldehyde.
  (HDM 05/10)
• autoblock works now for fermions as well. (Bug fixed).
  fast now works with continuation. (Bug fixed).
  (HDM 04/10)
• The file input is made read-only. This makes it less likely that this file is erroneously edited.
  (HDM 05/10)
• Block improved relaxation is now possible for complex Hermitian Hamiltonians as well (DAV rather than RDAV or rrDAV).
  (HDM 03/10)
• The mctdh executables are now under $MCTDH_DIR/bin/binary/<platform> and no longer under $MCTDH_DIR/bin/<platform>.
  (HDM 03/10)
• The routine chnpot can now handle combined modes with more than 2 DOFs.
  (MS 02/10)
• The C-DAV routine (complex davidson for resonances) is MPI-parallel.
  (MRB 01/10)

Revision 4, Patchlevel 2 (December 2009) (Public: December16, 2009)

• ---- Version 8.4.4.2 released. (December 16, 2009) ----
• Bug removed in Block-Improved-Relaxation which occurred when using MPI.
  The check_system script is updated. It now checks additionally for shared memory and NUMA libraries.
  (MRB 12/09)

Revision 4, Patchlevel 1 (December 2009) (Public: December09, 2009)

• ---- Version 8.4.4.1 released. (December 09, 2009) ----
• New operators implemented: Rf, hKEh, hFRh, hdqh, hdqRh, dqR, dq2R.
  Bug fixed in project.F, which was introduced in version 8.4.4. This bug let calculations with identical particles fail.
  (HDM 12/09)
• RRDAV (improved relaxation for real operators) can now treat (multi-set) electronic states.
  Elk_test can check parallel MCTDH now.
  (MRB 12/09)
• The gcc 4.4.1 now compiles the package correctly.
  (MRB 10/09)

Revision 4 (September 2009) (Public: September 11, 2009)

• ---- Version 8.4.4 released. (September 11, 2009) ----
• NUMA has been introduced to MCTDH.
  (MRB 09/09)
• A chapter "Installing the MCTDH package" is aded to guide Appendix E.
  compile.cnf and platform.cnf scripts updated. A installation on Mac OS X (Darwin) should now be painless.
  The read-inwf feature can now handle (complex) weight factors which multiply the A-vectors of different electronic states.
  New analysis routine edengen. (Thanks to Emmerich Tempfli and Sascha Zoellner).
  Showpot now knows the option -evec which writes the natural potential terms (eigenvectors of rho) to an ASCII file.
  (HDM 08/09)
• Conversion constants (energy, length, mass) exchanged with more accurate ones. See on-line docu "Input Documentation / Units" or excecute "mhelp -s unit".
  (HDM 07/09)
• Ported optcntrl script to python
  New version of optcntrl documented in tutorial and on-line docu.
  (MS 05/09)
• The Distributed Memory Parallelization using the Message Passing Interface (MPI) has been introduced to MCTDH.
  (MRB 04/09)
• Bug in initkleg fixed (IMPORTANT!!). A phase factor was missing for negative k.
  New KLeg operators added: sJp, sJm, sJpk, sJmk.
  (HDM 03/09)

Revision 3 (December 2008) (Public: December 4, 2008)

• ---- Version 8.4.3 released. (December 4, 2008) ----
• The routine adpop can now run in a so called quick modus and can make use of Monte-Carlo integration.
  (GAW 11/08)
• A parallel version of the DSYEV routine of LAPACK is now available. This one is used in the getdavmat and b/rdavstep routines.
  (MRB 09/08)
• New keyword "block-A" for Init_WF-Section. It allows to read A-vectors form a foreign restart files. (For block improved relaxation).
  (HDM 08/08)
• Build-up of the pre-conditioner for improved relaxation is parallel now. Timing of parallel routines improved. Keywords "no-calcha" and "no-funka2" removed and replaced by "no-funka".
  (MRB 07/08)
• New analyse routines "wfproj" and "rdacoeff". wfproj computes expectation values of a projector. The projector must be defined by an SPF, P=|phi><phi| . There may be a product of projectors.
  rdacoeff reads the coefficients (A-vector) and prints them to screen.
  (HDM 06/08)
• Partial (a)symmetrization of the A-vector implemented. This allows to treat mixtures of bosonic/fermionic systems. New script "check_system" added to install directory.
  (HDM 06/08)
• New keywords in potfit. "read-cw" allows to read the correlated weights form file and "write-cw" writes them to file.
  (OV 05/08)
• New keyword "block-SPF" for Init_WF-Section. It allows to read SPFs form a foreign restart file. (For block improved relaxation).
  (HDM 05/08)
• The compile script now knows the -j option which enables parallel compilation.
  (HDM 05/08)

Revision 2 (December 2007) (Public: December 14, 2007)

• ---- Version 8.4.2 released. (December 20, 2007) ----
• A description of the block improved relaxation feature is given in the tutorial section of the guide. (HDM 12/07)
• An multi-packet run can now also be performed in single-set representation. Improved relaxation can now be done in block form. (This uses the multi-packet single-set feature). A block relaxation is quite useful if not a particular state is sought, but, say, the 10 lowest states. The new keyword autoblock automatically generates an appropriate set of initial vectors, and the keyword split-rst let the program split the restart file, which contains all eigenstates formally on different (single-set) electronic states, into individual restart files. Currently, block relaxation works only for real hermitian Hamiltonians (i.e. only for RDAV or RRDAV). Block relaxation can treat systems with electronic states of both variants, single-set and multi-set.
  (HDM 12/07)
• Several bug-fixes in operator generation (genoper). These fixes apply to the summation of operators in general and to KLeg operators in particular: See the changelog file for details.
  (FO 11/07)
• New keyword iteration-factor allows to speed-up the convergence of potfit iterations. Intermediate timing information is written to output file when performing a diagonalisation run. Bug fixes in runinwf.F and logop.F . (HDM 11/07)
• The structure of the parallelization is changed, now the threads are created at the beginning of the program and are destroyed when mctdh has finished. New keywords for steering the parallelization (see html-docu).
  The array hlsym is in use again.
  (MRB 09/07)
• Potfit can now do continuation.
  (HDM 09/07)

Revision 0 (April 2007) (Public: April 17, 2007)

• ---- Version 8.4.0 released. (April 17, 2007) ----
  
• Shared memory parallelization improved. Mutual exclusions (mutex) introduced to reduce memory consumption.
  (MRB 04/07)
• muld-potentials ignored time-dependence. Fixed. Keyword read1d added.
  (HDM 04/07)
• Minor bug fixes in VCHam programs and documentation updated.
  (GAW 03/07)
• Bug in potfit fixed. Potfit produced a wrong natpot when separable weights were used.
  Output of eigenvalues (in au) in potfit is changed to reduced eigenvalues (in eV**2). This is a more meaningful measure.
  (HDM 02/07)
• Summation of products of real side-diagonal operators is implemented. This facilitates exact calculations when using FT'd potentials by decreasing memory and CPU consumption.
  (FO 02/07)
• Memory allocation changed for improved relaxation. The arrays Krylov and HKrylov are allocated individually. This allows to allocate more than 32 GB for improved relaxation.
  (HDM 12/06)
• Shared memory parallelization implemented. See on-line docu and Guide for details.
  (MRB 12/06)
• The file adwkb and the script pladwkb are renamed to enerd and plenerd , respectively.
  (AP 11/06)
• Release 4 created.
  This version uses less memory, because the hpsi array is much shorter. As the hpsi array has now to be re-calculated, mctdh84 may use (slightly) more CPU time compared to mctdh83. The reduction of memory consumption is considerable, if large combined grids and large potfits are used.
  Version 84 no longer supports density operator propagation and s-mctdh (selected CI).
  (HDM 11/06)
• (All the additions to 8.3.14 are also in 8.4.0)

Version 8 Release 3

Revision 14 (April 2007) (Public: April 17, 2007)

• ---- Version 8.3.14 released. (April 17, 2007) ----
  
• muld-potentials ignored time-dependence. Fixed. Keyword read1d added.
  (HDM 04/07)
• Minor bug fixes in vcham programs and documentation updated.
  (GAW 03/07)
• Bug in potfit fixed. Potfit produced a wrong natpot when separable weights were used.
  Output of eigenvalues (in au) in potfit is changed to reduced eigenvalues (in eV**2). This is a more meaningful measure.
  (HDM 02/07)
• Summation of products of real side-diagonal operators is implemented. This facilitates exact calculations when using FT'd potentials by decreasing memory and CPU consumption.
  (FO 02/07)
• A template file of the script platform.cnf.priv is added to the install directory. A private platform script is very convenient when a specific compiler is to be chosen for a specific machine (hostname).
  (HDM 11/06)
• Identical lines of the operator tableau (Hamiltonian-Section) are now summed correctly. There is no longer a need to sum them "by hand".
  (HDM 11/06)
• With the keyword "gengmat" (Run-Section) or the option "-gmat" mctdh generates a matrix-element of the G-matrix as pes-file. This is useful for checking the correctness of an implemented kinetic energy operator (by comparison with TNUM results).
  (HDM 11/06)
• Support for sub-particle operators, i.e. operator terms that do not act on the full mode, but on more than 1 DOF. Currently the implementation is only for real diagonal terms (e.g natpots). All DOFs that the term acts on must be consecutive in MCTDH ("no holes").
  (FO 10/06)
• Major changes in potfit. The program code is now slimmer, e.g. the possibility of generating FT-natpots is removed (use projection). Potfit now uses less memory and converges faster. The evaluation of the fit error may now be done less often. (See docu "niteration").
  (HDM 09/06)
• Major changes to support a new method for using Fourier-transformed potfits. The new method does not use the FT feature of potfit itself; rather the Fourier components are calculated with projection and are then potfitted separately. The shifting in K is then accomplished with the (k)shift operator. The old support for FT-natpots is removed.
  (FO 08/06)

Revision 13 (July 2006) (Public: July 27, 2006)

• ---- Version 8.3.13 released. (July 27, 2006) ----
  
• Support for installing on MS Windows via the Cygwin environment. See also our Cygwin hints.
  (FO 07/06)
  
• Major changes in VCHam. VCHAM programs MCTDH compatible. Dynamic allocation in vchfit and vcplot. Keywords have MCTDH structure.
  (GAW 07/06)
• Operators of type 0.5*(f(q)dq + dqf(q)) with f=q,sin,cos,sqrt implemented. See Appendix C of the guide.
  (OV 07/06)
• The routines for optimal control, bin/optcntrl, efield.F, and prop.F, are updated. New routine guessfld.F. Example input files can be found on $MCTDH_DIR/inputs/optcntrl/ . See also the README file there.
  (HDM 06/06)
• The routine prhosub can now compute the (diagonal) momentum representation of the particle density.
  (HDM and SZ 06/06)
• New feature: one can have more than one KLeg in a combined mode:
  (i) oper file format changed
  (ii) new operator: cjpm (see hamsym.ps)
  (iii) one can use KLegs in exact runs.
  (FO 06/06)
• crosscorr83 can now compute matrix elements. This, however, requires that the two WFs have same numbers of SPFs.
  (HDM 06/06)
• Major changes in potfit. Projection is now a program of its own. All potfit runs are now of type direct=1, and the direct keyword should no longer be given. The format of the vpot file is changed. Showpot (but not potfit) is able to read the old vpot files.
  Version number increased to 8.3.13.
  (HDM 05/06)
  
• The projection program now uses a completely new syntax and has increased functionality (additional projector functions).
  (FO 01/06)
• New compiler environments: gfortran (-32, -64), g95. Improved GCC version check.
  (FO 04/06, 07/06)
• Operator labels changed for CH3I and 1D potentials, e.g vh2 --> v:H2 . See Appendic C of the Guide for the new names. There is now a check that label and parameter names are disjoint. User defined parameters must not contain a colon (:).
  (HDM 04/06)
• New directory install/AdvancedUser/ added. This directory contains some useful non-mctdh software.
  (HDM 03/06)
• New analysis routine dengen. It generates reduced densities similar to showsys, but is not interactive and may thus run in background.
  Several bug fixes. Patchlevel increased to 1 (8.3.12.1)
  (HDM 02/06)
• ENO (Essentially Non Oscillatory) interpolation added to chnpot.
  (OV 02/06)

Revision 12 (December 2005) (Public: December 15, 2005)

• ---- Version 8.3.12 released. (Dec 15, 2005) ----
  
• New feature in potfit: projection runs
  (FO 11/05)
• psi2ex can now give the exact wavefunction in second population when setting the option -pop2.
  (OV 11/05)
• New keyword for improved relaxation, RRDAV, introduced. RRDAV is faster than RDAV but requires a "simple" operator (no KLeg etc, no muld-potentials) and cannot treat electronic states.
  (HDM 10/05)
• Analyse routine adproj added. This routine creates vpot files of the adiabatic projector matrix elements.
  (MRB 06/05)
• The VCHAM (vibronic-coupling Hamiltonian) set of programs is added to the MCTDH package.
  (GAW 07/05)
• New analyse routine twprob (Tanner/Weeks probabilities), used to calculate state-resolved reaction probabilities (so this is an alternative to "flux").
  (FO 06/05)
• New keywords symcoeff, asymcoeff, and id introduced. They are useful when treating identical particles.
  (HDM 06/05)

Revision 10 (May 2005) (Public: May 20, 2005)

• ---- Version 8.3.10 released. (May 20, 2005) ----
  
• There are four new files: install/user_surfaces, install/user_surfdef, opfuncs/user1d.F and opfuncs/usersrf.F . These are introduced to make it easier to include new one- or multi-dimensional potentials.
  (HDM 05/05)
• New HTML page Troubleshooting on entrance page of the HTML documentation.
  (HDM 05/05)

Revision 9 (December 2004) (Public: December 09, 2004)

• Bug fixed in genoper/htmsetup.F. Setting the keyword limit-mc is no longer necessary.
  (HDM 04/05)
• There may now be a read-inwf block in the Init_WF-Section. This allows to construct an initial WF from one or several (foreign) restart files. The number of electronic states of the WF on the restart file(s) may differ from the number of states of the WF to be propagated.
  (HDM 04/05)
• New analyse routine adpop83 to compute adiabatic state populations and one- and two-dimensional reduced densities. The script pladpop plots the densities.
  (MRB 04/05)
• New keyword id in SPF-Basis-Section allows equate modes which have identical SPFs because of symmetry.
  (HDM 03/05)
• New option -s in overlap allows to overlap selected electronic states. There are also the new options -b, -c and -t.
  (HDM 02/05)
• Format of the output-file of overlap83 is changed. This file contains now the overlap and the norms, but no longer the error measures. (compare83 changed accordingly.) The output to screen is unchanged.
  (HDM 02/05)
• ---- (patched) Version 8.3.9.1 released. (January 11, 2005) ----
• Handling of (a)sym1d, (a)sym2d, and sym3d changed. New keyword persist (argument to the sym-keywords) added.
  (HDM 01/05)
• Treatment of multi-dimensional potentials (muld-pot) improved. There are now more error checks.
  (HDM 01/05)
• ---- Version 8.3.9 released. (December 9, 2004) ----
  (HDM 12/04)

Revision 8 (June 2004) (Public: June 13, 2004)

• Modifications on Makefile and compile.cnf_* scripts. Some compilers (e.g. g77-3.4) produce wrong object files for the LAPACK routine dlamch. This is now fixed by compiling this routine without optimization. The code of funcanld.F and einbuild.F is re-organized. The pgf77 produces now correct optimized object files.
  (FO 12/04)
• New unit debye. New option -Mb for autospec83 and plspec. With this option one may plot the spectrum in absolute units (mega barn). (Thanks to G.J. Kroes for suggesting this.).
  (HDM 11/04)
• The dth1 and dth2 operations are added for KLEg and PLeg. New operators Jp, Jm, Jx, Jy implemented (for KLeg only).
  (HDM 10/04)
• New analyse routine jinpol83 implemented. This routine performs J-interpolation for scattering calculations.
  (SS and HDM 10/04)
• Symmetrization of the SPFs during the geninwf step (and for improved relaxation additionally after each orbital relaxation step) has been added. The new keywords are sym1d, asym1d, sym2d, asym2d, sym3d .
  (HDM 09/04)
• Keyword limit-mc implemented. This allows to limit the amount of memory allocated during the genoper step.
  (HDM 09/04)
• The symmetrized first derivative, sdq , implemented for sin-DVR. This is for treating asymmetric functions with periodic boundary conditions (cf. cos-DVR).
  (HDM 09/04)
• A pre-conditioner for improved relaxation (RDAV only) is implemented. The new keyword precon defines the dimension of the matrix used as pre-conditioner. The Olsen correction is applied when the new keyword olsen is given.
  (HDM 08/04)
• KLeg-DVR can be combined with other modes.
  (SS 07/04)
• New analyse program joinpsi83 implemented.
  (SS 07/04)
• ---- Version 8.3.8 released (June 13, 2004) ----

Revision 7 (March 2004) (Public: April 13, 2004)

• A cosine-DVR is implemented. It is useful for treating "gerade" functions with periodic boundary conditions.
  (HDM 06/04)
• Some small addition to the guide, e.g "Golden rules for writing operator files."
  (HDM 06/04)
• The problem of the large memory consumption of the FAST algorithm is fixed.
  (SS 06/04)
• CMF is now allowed for time-dependent operators. If the time-dependent operator is uncorrelated (i.e. if it is a mode operator), CMF should work without problems. If the time-dependent operator is correlated, be more careful and check against a VMF calculation.
  (HDM 05/04)
• The use of Fortran functions is now possible in the Parameter-Section (see on-line docu and guide).
  (HDM 05/04)
• New pl-script plrlx. (Energy vs. time in impr.relax.)
  (HDM 05/04)
• mctdh83(propwf part) and potfit83 may now allocate up to 8GB memory on 64 bit systems. (The previous limit was 2GB because of the use of integer*4 numbers. On 32 bit systems (e.g. PC), the 2GB limit still exist.)
  (HDM 05/04)
• S-MCTDH is overhauled, some bugs fixed, and documentation improved.
  New keyword converged added. This allows to stop an improved relaxation run, if the energy is converged.
  (HDM 04/04)
• ---- Version 8.3.7 released (April 13, 2003) ----
• Support for Intel compilers (version 8.0) added. Auto-detection of GCC version in compile configuration added.
  (FO 04/04)
• The new analyse routine rdgrid83 writes the DVR grid points to screen. Showd1d now knows the option -l (logarithmic plot).
  (HDM 04/04)
• New H2-H2 potential surface added. (h4bmkp.f).
  (FG 03/04)
• FAST algorithm up to the 4-th level (6-mode calculation) is implemented. Fast and Natpotcut are extended to work with several natural potentials.
  (SS 03/04)

Revision 6 (February 2004)

• The program showsys83 now knows the option -pop2 . When this option is given, showsys83 will plot the WF-density in the second representation, i.e. displays the basis set occupation rather than the DVR-grid occupation. For FFT this means, that the momentum distribution is displayed.
  (HDM 03/04)
• An initial WF can now be orthogonalised against another MCTDH WF read from a foreign restart file. Keyword: orthogonalise= path where path denotes the path of the foreign restart file. This feature is useful for improved relaxation.
  (HDM 02/04)
• The analysis routine, crosscorr can now overlap two restart files (use -R and -r options).
  (HDM 02/04)
• A set of new scripts, mdircp, mdiff, mbackup, ld, implemented. With these scripts one may set up a convenient backup sytem, useful when developing new code. mdircp allows to copy a whole MCTDH directory, and mdiff compares MCTDH directories. See also "The MCTDH backup utility" from the entrance page (Developer's Manual).
  (HDM 02/04)

Revision 4-5 (August 2003) (Public: December 23, 2003)

• A new analysis routine, psi2rst, is implemented. It writes a wavefunction from psi file to a restart file.
  (HDM 12/03)
• A flux operator is implemented (see guide Operator Functions). One may diagonalize this operator with eigenf to obtain flux eigenstates. Works only for sin-DVR, exp-DVR and FFT.
  (HDM 12/03)
• New keyword imp-ortho for the INTEGRATOR-SECTION implemented. The orthognormality of the SPFs is improved. Works only for the Runge-Kutta integrators rk5 and rk8.
  (FO 12/03)
• The program chnpot is overhauled and its input format is changed.
  (SS 12/03)
• New flux-projector read implemented. This allows to read an SPF from a "foreign" restart file and take it as projector. This "foreign" restart file may be build by an MCTDH-geninwf run.
  New label readsrf implemented. This allows to read a potential energy surface as energy-points from file.
  (HDM 11/03)
• Two-dimensional cpp, cmm symbols are added. See "Hamiltonian/Liouvillian Documentation"/"Available Surfaces". These 2D functions are needed for the 4-atom, J>0, kinetic energy operator in Jacobian coordinates.
  (SS 11/03)
• Makefile and compile-script changed, such other but GNU make will work again. The make used, however, must support the include directive.
  (FO 10/03)
• showsys83 can now read the restart file. This is useful e.g for plotting wavefunctions from a relaxation run, where the psi file is not needed.
  Printout to rlx_info file improved (Davidson). New keywords rlxemin, rlxemax for improved relaxation.
  "eigenf" may now be applied to complex, but hermitian, operators.
  The keyword "continuation=integrator" or the option "-ci" performs a continuation run WITH reading the Integrator-Section.
  (HDM 10/03)
• New feature meigenf (mode eigenfunctions) implemented. This allows to compute (multi-dimensional) eigenfunctions of mode operators. New script pladwkb implemented.
  (HDM 09/03)
• overlap can now compute the overlap between a wavefunction and a density operator of type II. Norm and trace are printed to the check-file (densities only).
  (CC 09/03)
• Keyword excite in the generation of the KLeg-Inwf added. New PES for H3O (WDSE) added. (See addsurf). Bug in coortrans.f fixed.
  (SS 08/03)

Revision 3 (July 2003) (Public: August 20, 2003)

• New operator functions: asleg and aslegth, associated Legendre functions of x and cos(x); reY and imY, real and imaginary parts of the spherical harmonics Y_lm(theta,phi). (See guide Tables C.1 and C.5 )
  (HDM 08/03)
• CAPs and "continuation" both now work for density operators of type II. Approximate raising/lowering operators for Morse potential are implemented.
  (CC 08/03)
• natpots may now be time-dependent. (Thanks to O. Kuehn, who noted this bug.)
  (HDM 07/03)
• New Davidson diagonaliser implemented for improved relaxation. There are actually three new "integrators" DAV, RDAV and CDAV, for the hermitian, real and complex case. New keywords: lock and quad as arguments for relaxation, tout=all , sym and check as arguments to pop, and energyorb and stdorb (Integrator-Section).
  Timing for orben, trafo and cross included.
  (HDM 07/03)
• New analyse program herma83. It checks the hermiticity error of the A-vector. Only for densities of type II.
  (CC 07/03)
• Bug fixed in the CMF integrator when using fixed update times (CMF/fix) which caused the integration of the SPFs to be only of first order instead of second.
  (FO 07/03)
• 1) Bug fixed (wrong pointer in htmprodpot) when multiplying complex potentials. CAPs will not be summed.
  2) More (useful) output is now written to the log-file for diabatic correction (keyword "correction=dia").
  3) crosspec produces now a slightly different output. The output is multiplied by 1/pi and the order of the columns is different. This makes firstly autospec and crosspec consistent and secondly allows to read the cspec.pl file by flux83 (-ed option). The option -old restores the old output format.
  (HDM 07/03)

Revision 2 (Apr 2003) (Public: June 11, 2003)

• The evaluation of 'analytic' external operators (linint.F, keword: external1d) is improved. The fitting is now done with a linear combination of quadratic functions rather than with linear interpolation. This gives much better fits.
  The treatment of complex potentials is improved. The Time functions may be complex, i.e. Exp or I may appear in the Time column.
  (HDM 07/03)
• Some small bugs fixed: einpbas.F, plqdq, plnat and plstate.
  A more severe bug concerning the exp-DVR fixed. The expectation values were incorrect, and one could not use more than one exp-DVR.
  Documentation improved (input; guide: chapter 6.5, 6.6, symtab,tex).
  To be more consistent, the order of the arguments (parameters) of the functions: sinh, cosh, tnah, switch1 and switch2 has been changed. The function cos1 has an additional parameter. One must edit the operator file, if one of those 1D-operators is used. See hamsym.ps or guide Appendix C.
  (HDM 06/03)
• Several small bugs fixed: Bug in Makefile (expect), in npotlab (rewind), in adiabwkb.F (mtr->ftr) and in heinsort.F and htmsetup.F (j_p, j_m with c_p, c_m)
  (HDM,SS,CC 05/03)
• Options of autospec83 (plspec) changed. Fourier-transform only (no energy prefactor) is now default. To invoke the energy prefactor give the option -EP.
  (HDM 05/03)
• Coefficients (different from 1.0) allowed for natpots.
  Furthermore, one may give 'I' in the coefficient column, making the coefficient imaginary. Complex coefficients are not allowed , only real or pure imaginary. However, the program may sum real and imaginary coefficients.
  (HDM 05/03)
• MCTDH can handle Fourier-Transformed (FT) natpot terms. (SS 04/03)
• Fast algorithm can work with natpotcut.
  (SS 04/03)
• Symmetry in interpolation of angular dofs in Chnpot has been added.
  (SS 04/03)
• Input file may be run through C preprocessor (use -cpp option). This allows several input files to share any subsets of their settings.
  (AM 04/03)
• Flux now also works when the CAP-DOF is within a combined mode.
  Rdcheck and crosspec now use less memory.
  Autospec now uses dynamical allocation of memory.
  (HDM 03/03)
• New method for gridpop and spop calculation for dentype I: The trace of \rho^2 is used. This is useful for traceless \rho. The keyword is quadpop.
  (CC 03/03)
• Bug fixed in dvrweights.f, a factor pi**0.25 was missing in the weights of the HO-DVR.
  Crosspec allows for automatic plotting (new option).
  Showd1d now knows the '-t' option (final time).
  New analyse routine probsq which computes the sum of the squared occupation probabilities of eigenstates.
  (HDM 02/03)
• For dentype II calculations one may read in a initial wavefunction rather than a density operator.
  (CC 02/03)

Revision 1 (Nov 2002) (Public: Jan 10, 2003)

• Some additions to the Guide.
  Showsys improved (handling of menu).
  Bug fixed in density operator propagation using FFT.
  Keyword operate_no-norm added. This command removes the normalisation of operator*psi (or of -i[operator,rho]).
  Option -mnd (make name directory) added.
  Keywords expect1 and real-only added. (HDM 01/03)
• H2O PES moved from wslfh.F to new (addsurf) file h2o.f (HDM 01/03)
• PJT2 Surface for H₂O has been added. vho potential curve for OH has been added. (SS 01/03)
• New analyse program Prhosub added. Prhosub reads pdensity file and writes reduced density of one selected DoF on primitive grid to files. (MN 12/02)
• There is a new operator keyword external1d{file}. A one-dimensional (diagonal) function may be defined by a set of points. These points are read form file file, interpolated and then treated as any other 1D-function (e.g. evaluated at the grid-points). (AM 11/02)
• New feature "FT" in Potfit has been implemented. (SS 11/02)
• Bug fixed in hnatfast. Fast algorithm was made more efficient for nmode ≥ 4. (Note: fast accelerates the multiplication natpot * A-vector). (SS 11/02)
• New operators added (KLeg/PLeg only): jp^2 and jm^2 , ( i.e. (j_+)^2, (j_-)^2 ). The symbol |& is introduced in the Hamiltonian-Section. It is equivalent to |k&l, where k and l are numbers. When using |& , k and l are determined by counting. Symbols like |&& and |&&& are similarly defined. For safety reasons, it is now forbidden to mix numbered (e.g. |1 or |4&5 ) with un-numbered (e.g. | or |& ) input within one line of the Hamiltonian-Section. (HDM 11/02)

Revision 0 (Sep-Oct 2002)

• Three new projectors for flux are implemented. Leg,KLeg and PLeg . They project on j or on (j,m) states. (HDM 10/02)
• New analyse program Chnpot has been implemented. Chnpot allows to map the natural potentials from one grid to another (usually from a coarse to a fine grid). (SS 10/02)
• Automatic CAP (ACAP) can now be switched on after some time. This time is to be supplied as the 5th argument of ACAP. (HDM 10/02)
• New keyword added: natpotcut. This removes all natural potential terms which are smaller than natpotcut. Default: natpotcut=1.d-8 (CC 10/02)
• New potential surface tully.f added: CO on Cu. (CC 10/02)
• New potential surface added: HCN. (FG 10/02)
• The 1D operators cos, sin, tan now have two parameters. The default value for the second parameter is zero. See Built-in Symbols page 4. (FG 10/02)
• Bug fixed in operator build-up for densities (dissipative part). Small terms were deleted inconsistently. Bug fixed in showpot. (It crashed when there were more than 1024*1024 grid-points). (CC 10/02)
• Almost all potential energy surfaces (except lsth) are moved from $MCTDH_DIR/source/surfaces to the new directory addsurf. The script mklinks creates links from surfaces to addsurf. (HDM 09/02)
• Two new integrators have been added: Runge-Kutta order 5 (RK5) and order 8 (RK8) (FO 08/02)
• MCTDH now has the ability to use external BLAS and LAPACK libraries instead of the self-compiled reference code. (FO 09/02)
• The translation between Fortran and C symbols can now be configured in "compile.cnf" (via MCTDH_F2C_STYLE) (FO 09/02)
• Bugfix: When performing exact propagations with an integrator other than SIL, the format of the "steps" file is changed slightly so that it can be processed with rdsteps (FO 09/02)

Version 8 Release 2

Revision 8 (Sep 2002) (Public: Sep 12, 2002)

• Three potential energy surfaces for H₃O system: WSLFH, YZCL1 and YZCL2 have been implemented. (SS 09/02)
• New External DVR (keyword extern) has been implemented. The grid points and DVR matrices are read here from file. (SS 09/02)
• New 1D-function, exp(p*(cos(x)-p1))**r, implemented. (CC 09/02)
• Bug in rdinpf (einglib.f) fixed (HDM 09/02)
• There are four new DVR's : Lagu1 ... Lagu4. They are build from the Laguerre functions
  phi(n,x) = Sqrt( (n-1)!/(n+a-1)! ) * x^(a/2) * exp(-x/2) * L(n-1,a,x)
  with a = 1, 2, 3, 4. (HDM 08/02)
• Analyse tool norm works now for density operators typ I. (CC 07/02)
• Bug in the memory-allocation part of overlap.F fixed (density operators typ II). (CC 07/02)
• The plot programs showsys82, showd1d82, showspf82, and showrst82 depict now the wavefunction and densities with the proper ordinate scale (i.e. independent of the number of grid-points). Using the no-weights option -nw (or, for showsys82, the menu point 290) one may switch to plotting the population of the DVR-points. (HDM 07/02)

Revision 7 (Jul 2002) (Public: Jul 4, 2002)

• Documentation overhauled. Validates as HTML 4.01 Transitional W3C. (HDM 07/02)
• Timing error in propcmfada.F fixed. (time -> localtime for second half step). (FO 07/02)

Revision 6 (Mar 2002)

• flux82 can now work with left-hand sided CAPs. The last argument for flux should be left, right or both. The default is right. (HDM 06/02)
• There are new 1D functions, which may be used to build an operator: cosh, sinh, tanh, switch1 and switch2. (HDM 05/02)
• There are two new analyse routines di2ad82 and psi2ex82. psi2ex82 transforms a MCTDH wavefunction to exact format and di2ad82 transforms a WF from diabatic to adiabatic representation. To do so, the WF must be in exact format. (GAW 05/02)
• Combining degrees of freedom which were represented by FFT did not always work correctly. Bug fixed. (GAW 05/02)
• The analyse routine ortho82 now works also for density operators of type I and II. Overlap82 can now do overlaps between density operators of different type. (CC 04/02)
• There is a new analyse routine diag82. It allows to compute eigenenergies by diagonalising the matrices generated with fmat82. (SS 04/02)
• The program potfit82 can now generate combined natural potentials. (FG 03/02)
• The plot program showsys82 can now do overlay plots. I.e. it can simultaneously plot two pictures, e.g. the contour lines of a potential and the wavepacket. (GAW 03/02)
• The evaluation of the cross-correlation function, both with the analyse program crosscorr82 or "on-the-fly" using the keyword cross, now works also for density operators of type I and type II. (CC 03/02 and FO 03/02)
• The script bashpath is updated. (HDM 03/02)
• The evaluation of expectation values, both with the analyse program expect82 or "on-the-fly" using the keyword expect, now works also for density operators of type I and type II. (CC 02/02)
• The mapping algorithm (keyword map in Init_WF-Section) has been improved. It now works also when the dummy grid is only part of the grid to which the SPF is to be mapped. (HDM 02/02)
• The evaluation of coordinate expectation values is modified in case of a 2*Pi periodic grid. <sin(q)> and <cos(q)> are determined and <q> is set to arctan(<sin(q)>/<cos(q)>). Similar formulae apply for <dq> . (HDM 02/02)

Revision 5 (Feb 2002) (Public: Feb 8, 2001)

• The use of the eigenf keyword failed, when there were combined modes and more than one 1D operator in the requested Hamiltonian. Fixed. (HDM 02/02)
• The fast algorithm (keyword fast in Integrator-Section) did not work for continuation runs. Fixed (HDM 02/02)
• operate now works also for density-operators of type I and II. (CC 02/02).

Revision 4 (Jan 2002)

In addition to some small bug fixes/improvements of the code, corrections of and additions to the documentation and to the User's Guide, there are some more important bug fixes and improvements which are listed below.

• The keyword cross has been added to the RUN-SECTION. It is used for calculating the cross-correlation function on the fly. See on-line docs for further details. (FO 01/02.)
• There is a new keyword in the RUN-SECTION, orben. When orben is set, the so called orbital energies are computed and written to the orben file. See the on-line documentation and the User's Guide for further details. (HDM 01/02).
• The A-coeff keyword has been implemented for densities type I and type II (CC 01/02).
• It is now possible to apply several operators to the initial wavefunction, e.g.: "operate = O1, O2, O3" (HDM 12/01).
• The mctdh -o option has been re-newed. An op.log file is now printed to screen, but it is the op.log file before summation of the Hamiltonian terms is performed. (HDM 01/02).
• The operators used with the eigenf keyword no longer requires an S1&1 1D-operator for the electronic degree of freedom. See the documentation for details. (HDM 12/01).
• The maximal line-length is enlarged from 80 to 160 characters when reading input files (*.inp, *.op, and *.srf). This is helpful when writing Hamiltonian-Sections. (HDM 12/01).
• The improved relaxation, i.e relaxation to excited states, has been developed further. It is also better documented, (although not yet in final form). See on-line documentation and User's Guide, (HDM 12/01).
• The single particle functions generated by diagonalising a 1D Hamiltonian (eigenf keyword) now obey a phase convention. When diagonalising a harmonic potential, one now obtains the harmonic oscillator functions with correct signs. (HDM 11/01).
• There are two new keywords map and addmode. map is used in the Init_WF-Section and copies the SPF's of one DOF to another. (Grids must be the same, of course). addmode, used in a Hamiltonian-Section, allows to add a mode which is not specified in the SPF-Basis-Section. These keywords are useful when one wants to propagate with FFT but needs initial SPF's generated by eigenf. (HDM 10/01).

Revision 3 (Sep 2001) (Public: Sep 18, 2001)

In addition to some small bug fixes, corrections of the documentation, and other small improvements, there are some more important bug fixes (listed below) and two new analyse routines.

• There are new analyse programs: statepop (called by analysis82) (GAW 07/01) and fmat82 (HDM 08/01). fmat82 computes <psi(t1)|O|psi(t2)> , where O is an operator (can be the unit operator) in MCTDH form.
• Bug fixed in heinsort.F, the keyword analytic_pes now applies only to the system hamiltonian. (CC 07/01). Bug fixed in transmf.F, diabatic correction now possible for nstate > 1. (HDM 07/01). Energy output for exact relaxation corrected. (GAW 08/01). Bug fixed in heinsort.F, the electronic Z-matrix is now correctly duplicated for multi-packet runs. (HDM 09/01).
• Bugs in plotutils.F fixed so that showsys works with combined modes. (GAW 07/01).
• Bug in flux fixed so that flux now works again correctly when using projectors. (HDM 07/01).
• Parameters may now carry a unit when defined via the -p option. (HDM 09/01).

Revision 2 (Jul 2001) (Public: Jul 2, 2001)

In addition to several small bug fixes, there are two new features: a two-dimensional Legendre DVR (PLeg) and "relaxation" to excited states.

• There are new analyse programs: crosscorr, crosspec, and vminmax. (HDM 04/01, 06/01)
• A bug concerning the time coordinate (heinterm.F) has been fixed. (GAW 04/01)
• A bug in the generation of multi-dimensional operators has been fixed. (SS 04/01)
• The operator label dq did not work properly for FFT. Bug fixed. (HDM 05/01)
• The compile script has been changed. The available multi-dimensional potential energy surfaces are no longer automatically linked. One has to use the -i option to link a desired surface to the potfit or mctdh program. (HDM 05/01)
• There are now reasonable error messages when the input for a build-up of an operator is incorrect. (HDM 05/01)
• A new keyword, periodic, is introduced in the Init_WF-Section. This allows to place an initial gaussian properly across a periodic boundary. (HDM 05/01)
• The expectation values <q> and variances <dq> are now written to the check-file and may be plotted with the aid of the script plqdq. (CC 04/01)
• The keyword relaxation may now carry an integer argument. relaxation=n will generate the n-th eigenstate. (See "Input-Documentation/Run-section"). However, as the method is based on the Lanczos algorithm, the method may converge to a higher eigenstate as requested. relaxation and relaxation=0 both converge towards the ground state, but relaxation=0 in general is much faster. (HDM 05/01)
• There is a new two-dimensional DVR, PLeg. It is a non-separable DVR based on the spherical harmonics Y_jm . It is similar in function to the sphFBR, but it is a proper DVR. (SS 06/01)
• The dependencies on the include files as listed in the Makefile are now generated by the script mkdepend. Hopefully, they are now all correct and the -a or -A options of compile are now needed only very rarely. (HDM 06/01)
• The dummy potential energy surface mysrf was added. This makes it fairly easy to include a new surface, as long as this surface does not depend on special parameters. (HDM 06/01)
• The Elk-Test is updated and two new tests are added. (CC 06/01)

Revision 1 Patch 1 (Public: Jan 29, 2001)

The following bug-fixes and changes become effective after the patch:

• A bug in the treatment of time-dependent operators has been fixed. (GAW 01/01)
• Product of operators, like cos*dq have not always been correctly treated. Bug fixed. (HDM 01/01)
• The showpot82 program displayed a Vfit when being asked to display a Vpot, and vice versa. Bug fixed. (HDM 01/01)
• The program showpot82 now knows more options. (See showpot82 -h ). (HDM 01/01)

Revision 1 (Jan 2001) (Public: Jan 17, 2001)

The following new features have been added:

• The defaults for creating output-files have been changed. The files output, timing, speed, update and stop are now opened by default. (update is, of course, only opened when CMF integration is used.) If you do not want these files give the keyword screen, no-timing, no-speed, no-update or no-stop.(HDM 01/01)
• The handling of numbers in exponential form is improved. Previously a parameter definition like: par1 = par2*2.3d-5 failed, because the program tried to subtract 5 from the (unknown) parameter 2.3d. Numbers in exponential form are now correctly identified, however, the exponent must be indicated by a d. In particular 2.3D-5, 2.3e-5 and 2.3E-5 are invalid! (HDM 01/01)

Two bugs have been fixed:

• Hamiltonian parameters close to 1 were erroneously set to 1. In some cases that may have lead to a (slightly) wrong operator. (HDM 01/01)
• When potfit was run with the keyword direct = 1, then the separable weights of the contracted mode were not correctly treated. (HDM 01/01)

Revision 0 (Dec 2000) (Public: Dec 17, 2000)

The DMCTDH programs have now been incorporated into the MCTDH program, and propagation of either density operators or wavefunctions is achieved by using the relevant keywords.

The S-MCTDH method (Worth JCP (2000) 112:8322), which tries to reduce computational resources by including only a selection of the configurations, is now included in a provisional form. Many of the ANALYSE programs, however, are unable to deal with wavefunctions of this form.

There are a number of new ANALYSE programs including:

showsys

Allows plotting of 2D cuts of the wavefunction, 2D densities, 2D cuts of the PES. The old showpes program has been incorporated into this program.

expect

Evaluates the time-evolution of an operator expectation value.

pexpect

Evaluates the time-evolution of a particle (separable) operator expectation value.

The following new features have been added:

• There is a new menu-driven interface for the analyse programs, and interactive plotting capabilities have been added to some routines. (GAW 11/00)
• Time-dependent operators can be treated. (GAW 10/00)
• The -pes option can be used in place of the genpes keyword. (HDM 10/00)
• The complete single-particle density can be output and used to evaluate the expectation value of separable operators. (HDM 10/00)
• Expectation values of a pre-defined operator can be calculated on-the-fly during a propagation. (CC 10/00)
• An operator file oper_S can be generated after a propagation using the keyword genoper = S. This can then be used to evaluate expectation values using the EXPECT program. (GAW 9/00)
• A pes file can be generated using the genpes keyword or, alternatively, by using the -pes option. This file can then be used to plot the potential energy surface. (GAW 8/00)
• An operator can be applied to the initially generated wavepacket before propagation. (GAW 8/00)
• The analyse routine flux has been enlarged substantially. The use of operators for computing averaged energy transfers and the use of projectors for determining the flux into specified quantum states is now possible. (HDM 11/00)

Other significant changes are as follows:

• A usediag/nodiag keyword has been introduced to control whether the diag flag is used when building an operator. (GAW 10/00)
• The symmetry of operators is now correctly accounted for (previously assumed that all operators are hermitian). (GAW 10/00)
• Operator log now always written to file op.log (HDM 10/00)

Various minor bugs have also been fixed and code changes made. These include:

• Bug in ABM integrator that occasionally led to very small steps. (HDM 11/00)
• Bug for FBR operators fixed. (GAW 8/00)

Version 8 Release 1

Revision 9 (Jul 2000) (Public: Jul 5, 2000)

In addition to updates in the documentation and users guide, the filter diagonalisation programs of Michael Beck have been added. The DMCTDH programs of Andreas Raab, which propagate density matrices using the MCTDH formalism, have also been added to the project. They are as yet undocumented, and not for the novice!

The following new features have also been added:

• The "normstop" option is added to stop a propagation if the norm drops below a given value. (SS 5/00)
• A "fast" algorithm for evaluating natural potential operators is added (SS 5/00)
• For SPHFBR, the operators "jz" and "jz^2" have been added (HDM 5/00)
• An exact run no longer needs an SBASIS-SECTION (HDM 4/00)
• New DVR implemented. Restricted Legendre DVR : Leg/R (HDM 4/00)
• New unit 'aJ' (= 10^-18 Joule) (MB 3/00)
• Can select only odd or only even HO functions as initial SPFs (GAW 2/00)
• Can select different SPF parameters for different sets in a multi-set calculation (GAW 2/00)
• Operator file for CO₂ added (MB 3/00)

Other significant changes are as follows:

• For SPHFBR, the operator "KE" is renamed "j^2" (HDM 5/00)
• Some ANALYSE programs have been renamed (HDM 3/00)
  • plpot -> showpot
  • rdden1d -> showd1d
  • rdspf -> showspf
  • rdrstspf -> showrst
• Path names given in input-files are now interpreted relative to the path of the input-file, and not relative to pwd. (HDM 3/00)

Various bugs have also been fixed, and minor code changes made. These include:

• Bug when relaxation + exact + BS integrator (GAW 6/00)
• Bug in power operation in LABELS-SECTION (GAW 6/00)
• Bugs in showd1d. (HDM 5/00)
• Order of summing / sorting operator terms changed. (GAW 5/00)
• Bugs in FBR routines. (GAW 5/00)
• Bug in morse operator. (GAW 5/00)
• Operators "dq" and "p" were not correctly used (MB 3/00)
• The summation of operator terms changed so that terms differing only in the coefficients are summed. (GAW 3/00)
• Orthonormalisation procedure of the spf's changed (HDM 3/00)
• Bug in generation of multi-dimensional operators (GAW 3/00)
• Bug with keyword "angst-1" (MB 3/00)
• Bug with operators "dth1" and "dth2" (MB 3/00)

Revision 6 (Feb 2000) (Public: Feb 28, 2000)

This Revision was the first distributed package. The User's Guide was started, and a user friendly installation procedure was introduced. The POTFIT program has been thoroughly overhauled by Stefan Wefing.

In addition to updates in the documentation, the following new features have been added:

• An exact propagation can start from an MCTDH wavefunction (GAW 12/99 )
• FBR integrals work for a general operator (GAW 11/99)
• "veigen" keyword to write the eigenvalues and functions of a 1D-operator to a file (GAW 11/99)
• PARAMETER-SECTION can be read from .inp file. The parameters can the be used for input. (GAW 11/99)
• Possibility of diagonalising a user-defined 1D Hamiltonian for the generation of the initial SPFs: "eigenf" (GAW 11/99)

Various bugs have also been fixed, and minor code changes made. These include:

• The generation of multi-dimensional operators changed, so that operators do not need to span entire system (partial-system operators) (GAW 12/99 )
• Changes made to summation of operator terms (GAW 11/99)
• Changes made to generation of initial wavefunction (GAW 11/99)
• Calculation of energy distribution for (a)diabatic correction changed (SS 11/99)
• Terms in the Hamiltonian with a zero coefficient are deleted (GAW 11/00)

Changes in Version 8.0 (Jan 1999)

The following changes of the MCTDH code with respect to the preceding version have been made:

The whole program was restructured and made more modular. The MCTDH main program now calls four major routines - rundvr, runoper, runinwf and runpropwf - which are quite independent from each other. They build the DVR's, the operator, the initial wave function and perform the propagation.

The following new features have been added.

• Numerically exact propagations can be performed. (GAW 9/98)
• The multi-packet propagation was implemented. (MB 10/98)
• Cross-correlation matrices can be computed. (MB 3/99)
• Lanczos diagonalisation routine implemented. (MB 5/99)
• Excited single-particle functions can be populated in the initial wavefunction. (MB 5/99)

Changes in Version 7.1 (Jan 1988)

The following changes of the MCTDH code with respect to the preceding version have been made.

• An error estimate of the auto-correlation function was introduced. (AR 2/98)
• An automatic program test has been written. (GAW 4/98)
• Higher-order auto-correlation functions were implemented. (MB 4/98)

Changes in Version 7.0

The following changes of the MCTDH code with respect to the preceding version have been made.

• A new input format using keywords has been implemented. (GAW 7/97)
• A new operator format based on keywords has been implemented. (GAW 7/97)
• Dynamical allocation is now possible. (HDM 7/97)
• The error estimate for the Lanczos-Arnoldi integrator was improved. (MB 7/97)
• An exponential DVR has been implemented. (HDM 9/97)
• An FBR representation of spherical harmonics has been established. (MCH 11/97)

Changes in Version 6.4

The following changes of the MCTDH code with respect to the preceding version have been made.

• The CMF step size can now be controlled using an error estimate of the single-particle functions only. (MB 2/97)
• In a CMF calculation the autocorrelation function can now additionally be written to file in the middle between two outputs. (MB 2/97)

Changes in Version 6.3

The following changes of the MCTDH code with respect to the preceding version have been made.

• The CMF method now allows adaptive step sizes. (MB 11/96)
• The combination of interaction picture and CAPs has been implemented. (GAW 11/96)

Changes in Version 6.2

The following major changes of the MCTDH code with respect to the preceding version have been made.

• Sine DVR has been implemented. (AJ 10/96)
• A Lanczos-Arnoldi integrator has been implemented. (MB 10/96)

Changes in Version 6.1

The following major changes of the MCTDH code with respect to the preceding version have been made.

• CDVR has been implemented. (GAW 9/96)
• Products of single-particle operator matrices and mean-field matrices are now summed up for potential terms. (MB 10/96)