# r67691: (September 4th, 2018): ADF2017.114 released

### r65713: Improved stability of geometry optimizations for molecules containing linear angles.

Simple linear tri-atomic molecules should still be optimized in Cartesian coordinates.

# r65628: (May 16th, 2018): ADF2017.113 released

This subrelease of ADF2017 contains bugfixes, and the linux/intelmpi version ships with IntelMPI 2018 update 2 to solve issues when running on multiple nodes.

### r65425: ADF: fixed a bug for open shell excitations with RS LibXC functionals

A bug was fixed for RS LibXC functionals, like LibXC CAM-B3LYP, in case one performs an unrestricted excitation calculation. The excitation energies were incorrect.

### r65403: Band bugfix: Becke grid issue for small unit cells

For small unit cells, the MPV-fuzzy-cells partition function was not using a large enough super-cell. As a consequence, some integration points could get an incorrect integration weight. Now, a more robust algorithm is used for determining the super-cell used in the fuzzy-cells partitioning.

### r65397: TD-DFT+TB now honors the tolerance set in the Excitations block

Previously the convergence threshold on the residual for TD-DFT+TB calculations with the iterative solver was always 1.0e-9. It now uses the Excitations%Tolerance keyword to determine the convergence threshold. This also means that the default changed from 1.0e-9 to 1.0e-6, which should be sufficient for absorption spectra.

### r65386: bugfix adfprep: importing a substructure from a .cry file failed

For example adfprep -t BAND-SP -m \$ADFHOME/atomicdata/Molecules/Crystals/Cubic/CsCl.cry#MgTl imported CsCl instead of MgTl.

### r65330: bugfix ADFview when chaninging bond orders.

Changing the bondorder of a bond in ADFview (for visual purposes only) lead to the error invalid command name "Notebook::MainPage"

### r65273: ADF: fixed a bug for X2C icw DEPENDENCY

Fixed a bug for relativistic X2C calculations in case of linear dependence of basis functions. This bug may appear if dependent eigenvectors are removed from the valence space. Note that the key dependency is internally used in case of (meta-)hybrids, with a rather large value for determining whether basis functions are dependent.

### r65020: Fixed a stack overflow problem in the “green” module.

This problem affects only the Windows and Mac versions.

### r64950: BAND: PEDA-NOCV gives warning (due to fractional occupations in fragment).

The warning would be. Not all NOCV orbitals pair! The results might be useless! Usually this is caused by having fragments with fractional occupations. The program now checks for this, and stops if this is the case.

### r64844: DFTB: Fixed orbital plotting issue for certain elements

The orbitals of elements where the energetically lowest valence shell are not s-orbitals (e.g. Ag with [Kr] 4d^10 5s^1) were not plotted correctly for DFTB calculations with the QUASINANO parameter sets. This has been fixed.

# r64726: (April 4th, 2018): ADF2017.112 released

This subrelease of ADF2017 contains bugfixes, and is also compiled with ifort 2017.0.5 to solve issues when running on Skylake SP CPUs. On linux this unfortunately raises the GLIBC requirement to 2.11 or higher, which means the compute engines no longer run on CentOS 5.

### r64726: Fix conversion of rydberg units

This was used by the QE GUI, when converting QE result from the QE output files to the rkf format.

### r64665: Band: changed the way the spin symmetry is broken for the spin-orbit case: use mix=1 at the first SCF cycle.

For instance Iron would fail to become magnetic. A larger SCF%vsplit also would have helped.

### r64621: Reaxff: GCMC: Fix inserting polyatomic molecules, improve output, handle optimization failures.

Polyatomic molecules were previously always inserted in the same orientation due to a bug. This fix makes the orientation random as expected. The information about attempted GCMC moves printed to the standard output has been made easier to understand. Additionally, GCMC now rejects a move if the trial geometry optimization fails. Although such moves were almost always rejected anyway due to their very high potential energy, sometimes they were randomly accepted, bringing the system into a highly unfavorable geometry and causing all subsequent GCMC moves to fail.

### r64511: Reduced the severity of problems related to the “Suspicious delocalized coordinates” warning.

The “Suspicious delocalized coordinates” warning is often caused by the presence of (almost) linear angles in the system. To circumvent problems caused by such an angle, the program replaces it by angles in two out of the three planes (XY, XZ, YZ). These additional angles result in the six (or five for linear systems) lowest singular values of the B matrix being not exactly zero. In principle, the three lowest modes should always be zero because they correspond to translations of the coordinate system. This was not the case due to a bug that has been fixed in this patch. However, even with this fix, the other three (or two in linear systems) values, the ones corresponding to rigid rotations of the system, may still be non-zero. This should be less of a problem though. The advice to users is to repeat the optimization in Cartesian coordinates if that in delocalized did not converge.

### r64500: Bugfix in ADFjobs on Windows: all calculations in a fragment analysis job should now run in parallel.

If the user does not set NSCM or leaves the options field in ADFjobs empty, the bug was causing the calculation of the fragment file(s) to run on a single core and only the main job was using all cores. This is due to the peculiarity of handling empty environment variable by MSYS.

### r64471: BugFix in ADF: SCANFREQ and RAMANRANGE can now be used with SYMMETRY NOSYM.

For SCANFREQ and RAMANRANGE, ADF reintroduced symmetry even if SYMMETRY NOSYM was defined in the input. Now, if SYMMETRY NOSYM is specified in the input, ADF will keep using NOSYM for the SCANFREQ and RAMANRANGE numerical displacements.

### r64360: ADFinput with lattice vectors that are integers would cause issues in ADFinput when using fractional coordinates.

For example fractional coordinates are all 0.0, mapping into specific cells will fail,…

### r64205: Reaxff: Fixed a bug in e-Reaxff that would sometimes result in the energy being not a number (NaN).

The NaNs were caused by a singularity in one of the eReaxff expressions for the total BO on an atom equal to zero.

### r64170: Generation of SDF files would fail with more than 999 bonds.

SDF files cannot have more than 999 atoms or bonds. If it is just the number of bonds that is too big, now the atoms are still written but the bonds are skipped.

### r64057: ADFinput for Quantum Espresso: fix selection of pseudo potential.

The first time the selection would work, but after that it would not be changed in the run script even if the users selected another option.

### r64015: Band: Too many lines. RDSETS.

This error can (still) happen when there are very many atom types. Now the STO fit functions are no longer considered if not needed, making it less likely to occur.

### r63787: Reaxff: Allow charge on an atom to be a 0.5e more positive than that of the nucleus.

Reaxff will stop if the charge is more positive than that and badeem is not set to 1 in the control file.

### r63683: Reaxff: changed e-Reaxff input handling for charges of pseudo-atoms.

During e-Reaxff calculation the pseudo-atoms, for example, corresponding to the explicit electrons, are supposed to carry a constant charge. Originally, such a pseudo-atoms was supposed to be listed in a MOLCHARGE key in the geo file. Since e-Reaxff requires exactly one MOLCHARGE key per pseudo-atom this can become cumbersome. With this change, the El, Lv, Eh, and Ho pseudo-atoms must not be included in any MOLCHARGE range and they will each be assigned a separate “molecule”.

# r63420: (January 22th, 2018): ADF2017.111 released

### r63381: Fix ADFspectra to work when no bonds have been saved to t21 (GUIBONDS not present)

This fixes error can't access "t21::bondatoms": parent namespace doesn't exist

### r63379: Fix TMPDIR for OpenMPI on mac

The default value has a path that is too long for OpenMPI to handle. Now TMPDIR is reset to /tmp.

### r62964: ADFinput fix import of mol2 files

Import could fail in some situations when substructure info is missing.

### r62956 ADF: bug fix STCONTRIB icw symmetry NOSYM

ADF could crash with an error message like: ‘Error in the transformed matrix. STContrib’, or crash with ‘Context has remaining references after call to adf_end_blacs.’, or give incorrect results.

### r62930 Reaxff: Removed the I/O bottleneck for long trajectories.

Running a long simulations that produces a very large rxkf file should become much faster with this update.

# r62318 (November 20th, 2017): ADF2017.110 released

### r61878: ADFinput: for DFTB the constraints were not included

The bug was introduced in r60718, and thus is present in the 2017.108 and 2017.109 release.

### r61814: Reaxff: An important bugfix for the r62775 below.

In r61775, the torsion angle value is incorrectly adjusted by 2*PI while the angle is given in the units of degree and not in radians as it was assumed.

### r61812: Bugfix ADFspectra for MCD spectra

Clicking in the table below the spectrum would highlight the wrong peak (off by one).

### r61775: Reaxff: small change for handling dihedral angles in the training set.

When calculating a difference between the optimized dihedral angle theta and its reference counterpart theta_0 the minimum of theta-theta_0, theta+2pi-theta_0, and theta-2pi-theta_0 is selected. This will make sure that, for example, two diherdals -pi and +pi will be considered equal.

# r61708 (October 16th, 2017): ADF2017.109 released

Using a right click on a job in ADFjobs will show the SCM menu. On linux in some cases it would lead to an error.

### r61580: ADFinput: OCCUPATIONS key now sums full shells to avoid problems with the OCCUPATIONS line being too long for ADF.

This is mainly useful for excitations from a core level.

### r61513: ADFinput: fix crash on missing chain info.

Fixes errors like can't read "data(-,ribbon)": no such element in array

### r61380: BAND: avoid crash when /tmp has no write permissions.

This should fix the error couldn't create error file for command: permission denied.

# r61155 (September 12th, 2017): ADF2017.108 released

### r61057: ADFspectra: PVDOS spectra with multiple molecules of different sizes would crash

Fixes errors like can't read "data(1,pvdos,9,1)": no such element in array

### r61028: ADFinput: fix bond display issue that could happen after using New

Fixes errors like can't read "::Mol::bid ..."

### r61017: ADFmovie: avoid problems getting molecule fraction data wirh ReaxFF

Fixes error can't read "molnumber(Iteration)" ...

### r60954: ADF: BugFix ZORA SAPA

A bug was fixed for the non-default ZORA SAPA method. In ADF2017 in case of ZORA the ZORA MAPA method is default.

### r60935: ADFinput: avoid scrolling crashes when windows no longer available.

Fixes errors like bad window path name ...

### r60624: ADFinput: Builder did not work any more

A line of debug code broke the builder for all users (except the developer …), fixed.

### r60590: Reaxff: Extend the molecule gun.

A NUMADD option has been added to the addmol.bgf file format. The NUMADD keyword has one integer argument that sets a limit on the number of molecules added.

### r60570: bugfix ADFinput: slider for geometry changes fixed.

If you would click outside the thumb in the slider for geometry adjustments, it would keep changing the value as if you kept clicking.

### 60538: Reaxff: Fix a long-standing bug: the external electric field component entering the EEM equations is now treated properly in eV/A units.

Before this fix, the field was multiplied additionally with the 1/(4*pi*epsilon_0) factor (=332.0638 kcal/mol*A/e^2), which was erroneous in two ways. First, the electric field should already contain the 1/(4*pi*epsilon_0) (=14.40 eV*A/e^2) factor when specified in eV/A. Second, the EEM system of linear equations is solved in the eV units, which means that no conversion from eV to kcal/mol should be done for the EEM right-hand side term due to the external electric field.

### 60517: Band: big fix: do not compute partial DOS and fat-bands for spin-orbit coupling

Bug Fix band: Band will not compute the partial DOS and fat-bands in case of spin-orbit coupling (it used to calculate those properties, but they were incorrect).

### 60500: Band: bug fix for hybrids

Fixed a bug that caused band to crash in case of hybrid calculations with large cells (where only 1 k-point is used)

# r60454 (July 21, 2017): ADF2017.107 released

### r60395: Bugfix in DFTB+Band: Geometry optimization with frozen atom constraints are more reliable

There was a mistake in the code which led to the rigid motions being projected out in optimizations with frozen atoms constraints. This moved the atoms which were supposed to be frozen, causing all kind of problems, e.g. failure to converge at all.

### r60363: Band problem with badly normalized charge in a phonon run.

The program would stop with: WARNING: BADLY NORMALIZED AT. CHARGE.

### r60208: ADF does not calculate VV10 dependent LibXC functionals

ADF can not calculate VV10 dependent LibXC functionals, like VV10, LC-VV10, B97M-V, WB97X-V.

### r60098: 32bit Windows GUI now OpenGL1 compatible, 64bit Windows GUI now build in 64bit mode

The 32bit Windows GUI is now compiled in OpenGL1 compatibility mode. This disables some of the performance improvements introduced into the GUI with ADF2017, but allows for using the GUI on older hardware. The 64bit Windows GUI still contains all the performance improvements (using OpenGL 3.2), and is now compiled in proper 64bit mode.

### r60049: Bugfix ADFinput: fixed constraints introduce 1 extra type per element instead of 1 per atom

BAND was running out of types. Also reset the atom type if the fix constraint is removed.

### r60042: Try to make the GUI more robust agains virus scanners blocking file deletion

Fixes error messages like: error deleting "W:/SCMTMP/user/cmd185880.err": permission denied from ADFview.

# r59763 (June 13, 2017): ADF2017.106 released

### r59736: ADFinput: saving a fragment job sometimes produced wrong job references.

The result was that the fragments did not run automatically, and that fragment files were not found.

### r59728: ADFinput: when a fragment file has been specified for a fragment, no fragment job should be created and/or run

This is relevant only when you explicitly select a fragment file to use in the fragments panel.

### r59692: BugFix: Fixed molecular dynamics calculations via ASE.

The driver script for MD via ASE was not updated after we upgraded to ASE 3.11. This effectively broke the MD functionality for Band and ADF. Fixed now.

### r59687: ADF: bugfix dDsC dispersion correction for functions it has not been parametrized for

The density dependent dispersion correction (dDsC) was giving errors for XC functionals it was not parametrized for. Now it does calculate a dispersion energy, however, using non-optimized parameters. One can set the dDsC parameters ATT0 and BTT0 with

XC
...
DISPERSION dDsC ATT0=att0 BTT0=btt0
END


### r59441: UFF: Fixed the hybrid scheme for geometry optimizations.

When the hybrid scheme is used, the nonbonded interactions are excluded at the start and get included once the system is getting close to convergence.

### r59431: Reaxff: Change again the way atom names in bgf input are interpreted.

Now element names in either in the two entries should be recognized automatically.

### r59411: Reaxff: Change the way atom names in bgf input are interpreted.

If the 1st name contains a number or an apostroph then it is ignored and the 2nd one is used. Otherwise the 2nd one is ignored, unless it is also more complicated than just one or two letters.

# r59271 (May 15, 2017): ADF2017.105 released

### r59229: GUIprefs: fix ADFjobs status window auto-refresh preference setting.

The preference value was ignored.

### r59184: ADFinput: list of frozen atoms for MOPAC calculation was not displayed after reopening saved job.

This issue had no effect on the calculation itself.

### r59181: ADFview: when both t21 and t41 are present, t21 is opened by default with the SCM menu (even if the t41 is newer).

This prevents problems after calculating a NOCV charge displacement function (which is saved on t41).

To open a .t41 file: select specifically the t41 file in ADFjobs, and use the SCM menu.

### r59168: Foray: Workaround for a python 3 issue with nfs4 filesystems

The python shutil.copy2 command fails on nfs4 filesystems, which causes the compilation from source to stop.

### r59151: ADF/BAND: bug fix for high integration quality

When requesting Lebedev angular integration grids of order > 59, return 59 instead of stopping

### r59144: ADF: Make it possible to specify the list of nucleii for analytical frequencies on multiple lines

It is now possible to specify the NUC sub-key multiple times within the AnalyticalFreq block. The lists will be concatenated.

# r59012(May 2, 2017): ADF2017.104 released

### r58999: MacOS: improve double click on ADF application and its documents.

Double clicking on a document (like a .adf or .t21 file) would open in the GUI, but only when the GUI was not yet running.

And it was impossible to open the GUI by multiple users (on the same mac) at the same time.

### r58995: MacOS: fixed dylib reference paths in the pythonstack and openbabel to solve MacOS signing issues.

Fixes error message: can't read "gui(DTMvectors)": no such element in array from ADFmovie.

### r58987: ADFinput: if installation of QE or LFDFT database failed, ADFinput could run into an error.

Fixes error message: can't unset "installchecked": no such variable from ADFinput.

### r58983: GUI on MacOS: remove check that DISPLAY env var is set

The current version does not use X11 on a mac. Nevertheless the GUI would refuse to start on MacOS when the DISPLAY environment variable has not been set. If XQuartz has been installed DISPLAY will be set, so then there is no problem.

### r58945: ADFinput: properly remove modifystartpotential settings when not used.

If removed (turned off) ADFinput still insisted (incorrectly) to turn on unrestricted.

### r58904: ADFjobs: update of error and warning status made more robust

It was disabled permanently when selecting the Single Point option.

### r58884: ADFinput: when saving fragment jobs, make sure they are all saved in the same folder.

In some rare cases the fragment jobs would be saved in a different location (the location of the previous job), causing issues.

### r58881: Make Builder work more reliably.

On some systems like Ubuntu the Builder (using packmol) did not work.

# r58813(April 13, 2017): ADF2017.103 released

### r58793: ADFjobs: make monitoring and handling of remote jobs more robust

This solves the issue that sometimes remote files were not automatically updated on the local machine.

Include somewhat arbitrary default COSMO radii for the heavy elements Am-Og (Z=95-118), equal to 2.1 Angstrom for all these elements, like in r58763, but now in the ADF program. This is relevant only for COSMO and if one does not specify a COSMO radius for such heavy element.

Somewhat arbitrary COSMO radii are included for the heavy elements Am-Og (Z=95-118), equal to 2.1 Angstrom for all these elements (in the ADF-GUI). Reason is that such value is better than 0.0 Angstrom. COSMO radii of 0.0 Angstrom may crash the calculation, or give NaNs (not a numbers).

### r58755: Lowered GLIBC dependencies for Linux

The compute engines and python stack now require a minimal GLIBC version of 2.4, the GUI requires GLIBC 2.7 or higher.

### r58745: ADFjobs job monitoring issue solved

ADFjobs monitoring jobs that take more then 10 minutes lead to scmjobd.exe processes getting stuck (and using one full core continuously).

### r58706: ADFjobs improve display of error status

The warning and error messages of jobs were not always shown in ADFjobs, even when warnings or errors were present. This should be fixed. This issue has no effect other then the display in ADFjobs.

### r58661: Plasmons in Molecules

In [J. Phys. Chem. C 117 (2013) 1863], a scaling approach was used to identify plasmons in molecules. This is now documented in the ADF manual.

### r58634: ADFjobs job monitoring might cause duplicate dependency lines in pid file

Fixed, did not cause any problems.

### r58623: ADFjobs monitoring of jobs started via dependencies (like fragment jobs) fixed.

The jobs and dependencies did run fine, but the status (of the jobs started automatically via dependencies) in ADFjobs would not update.

Note that this fix requires your remote ADF installation to be updated as well (this fix is actually on the remote site)!