Bug-fixes and improvements to the latest major release, with the latest changes listed first. The revision number corresponds to that of the bug-fixed binaries.
|r93137||AMS2020.103 released (April 15th, 2021)|
|r93118||ADF: Bug fix analytical frequencies open shell PBEc
The bug affects the analytical frequency calculation of spin-polarized systems with PBE, OPBE, RPBE, revPBE, and S12g. This bug was present in all ADF releases starting from ADF2006. Analytical frequencies using LibXC are not affected.
Workaround is to use numerical frequencies.
|r93101||COSMO-RS: fixed saving multipleform compound with only conformers.|
|r93086||AMSjobs: closing AMSjobs could give an error during cleanup when checking the mtime of certain jobs.|
|r93075||Plumed on Linux: on builds with MKL2021+, we use a plumed that was also compiled with MKL2021.|
|r93007||AMSmovie: fix angle graph for results other than .rkf and .t21.|
|r92948||Reaxff Classic: add a shortb control flag that allows ignoring short bonds, similar to the System%AllowCloseAtoms setting in AMS.|
|r92910||COSMO-RS: fixed several numerical issues related to small numbers.|
|r92905||Reaxff: fix a bug finding random initial guesses with FFOptimizer.|
|r92888||Prepare AMS2020 for switching to Intel MKL 2021.1.1|
|r92695||AMSjobs: fix previous bug fix regarding job monitoring|
|r92657||AMSjobs: detecting jobs that were no longer running (and thus are missing) were not properly detected.|
|r92506||AMStrain: importing via SDF file failed if pointing to engine files, now matching AMS result file is used.|
|r92479||AMS: Automatic atomtyping in the ForceField engine would fail if antechamber created new atomtypes.|
|r92394||AMSjobs: moving a remote job to a directory that exists locally but not remetly would fail on the remote side.|
|r92393||AMS: Fix the path length for the last point in the ams.rkf file after IRC scan.|
|r92361.||BAND: bug with EnforcedSpinPolarization. Would give wrong occupations when one of the occupations should be zero.
For instance a Hydrogen atom with enforced polarization of one would give occupations 1//1 rather than the intended 1//0.>
|r92342||AMSjobs: fix crash when trying to rename a job without selecting one job first.|
|r92339||AMSinput: improve reading of old Quild setup (system read, switched to Hybrid engine but no options read).|
|r92308||DFTB: possible segfault after message "reducing k-group size to the nr. of k-points used during the SCC".|
|r92291||AMSinput: avoid pre-optimizer failure due to AMS properties requested in current module.|
|r92266||AMS IRC: Fix small bugs in the output of IRC scan.
Switch forward and backward barriers in the IRC summary so they are more chemically intuitive (energy change on the path backwards equals to negative of the barrier of the forward reaction and vice versa). Fix values for arc lengths on the rkf file and the final path length in the printed output.
|r92245||ADF and BAND: Again workaround precision issues LibXC REVSCAN and REVSCAN0 functional.|
|r92225||AMSjobs: avoid potential crash when deleting files in a directory in which hidden files are present.|
|92222||AMSinput: fix menu coloring for some panels like the Expert ReaxFF panel.|
|92213||BAND/amsview: plotting of orbitals could fail with a techical "section name too large" error.
For a bulk system with the unrestricted key only the first 99 oribtals could be plotted.
|92203||AMS: Fix a bug that can cause symmetry be distorted during geometry optimization with delocalized coordinates.
The fix does not apply to periodic systems when optimizing lattice.
|92164||AMSinput: When importing an .ams or .adf file via File->Import Coordinates, the lattice vectors were not imported.|
|92149||GUI: Avoid potential issue with couldinfo missing when pid files are removed (which you should not do...).|
|92075||AMSinput: Fix handling of ghost atoms and other atom properties.|
|92065||GUI: fix crash in AMSmovie and AMSspectra when VROA results are present without the CID results.|
|92050||DFTB: fix hanging while saving orbitals for periodic systems with multiple k-points.|
|r92045||AMSbands: Fix using custom colors for bands colored by character.|
|r92037||AMSinput: Fix typo froze core for ADF.|
|r92033||AMSinput: Fix issue with FCF when files have special characters (quotes added).|
|r92029||GUI: Fix issues with display of Geometry Info->Distances.|
|r91998||AMSinput: COSMO-RS setup job made more portable.|
|r91957||ADF and BAND: for some of the M06 functionals find automatically values for Grimme dispersion.|
|91872||Hybrid engine: new option TweakRequestForSubEngines, with default true. Disabling it leads to more printing by subengines such as ADF and BAND, in particular charge analysis.|
|r91851||AMSinput: do not create .fixed atom types when fixing atoms during an optimization.|
|r91833||DFTB: Crash during calculation of bandstructure for large systems, due to exsessive memory usage. The same could also happen for the Effective Mass calculation, although less likely.|
|r91788||GUI: Fix atom selection of periodic images, bug introduced in r91186.|
|r91760||AMSprefs: Default engine in AMSinput ReaxFF uses AMS ReaxFF, and old ReaxFF can not be set as default.|
|r91711||AMSinput: setting up a molecule gun input, and next switching to a non-MD task, would crash on saving.|
|r91680||amsprep: include new ams modules like MLPotential and Hybrid|
|r91607||AMSinput: enable the use of FCF for properties-only calculations.|
|r91569||AMS driver: charges and atom types for the ForceField engine no longer cause extra atomic species to be created
The enormous number of atomic species caused issues when the ForceField engine was used together with ADF in a QM/MM calculation through the HybridEngine. This resulted in ADF stopping with a "too many fit sets over types" error. For some exotic cases this may change the symmetry detection behavior on AMS, since the ForceField engine's atomic charges and atom types are now ignored in determining the system symmetry.
|r91501||AMSview: avoid crash when changing the Grid when a colored isosurface has been added but with no fields specified.|
|r91476||AMSinput: when setting up a properties only ADF calculation, make sure results are selected to start from (which may not be the job itself).|
|r91473||AMSinput: fix converting a non-scanning PES constraint into a proper Geometry constraint (the target was specified twice in the input)|
|r91459||AMSjobs: make sure every job has a queue assigned. This avoid overwriting the options field for new jobs when running.|
|r91443||AMSview: fix for visualization NCI and DORI.
Visualization of NCI: use 'Isosurface:with phase' isosurface of property NCI (at 0.5). Alternative visualization of NCI: use colored isosurface with isosurface of property RDGforNCI (at 0.5) colored with NCI (range -0.5 to 0.5). The value of RDGforNCI is the same as RDG (reduced density gradient) except in regions where the density is larger than 0.05, in order to get rid of surfaces around atoms, that one would see if one is using an isosurface of RDG. Visualization of DORI: use colored isosurface with isosurface of property DORI (at 0.9) colored with DenSigned (range -0.02 to 0.02). For both NCI and DORI one may need a medium or fine grid for best visualization.
|r91402||ParAMS: Fix inadvertent case-sensitive handling of the Task input key for AMS.
The AMSWorker interface in PLAMS used by ParAMS incorrectly interpreted the value of the Task key in the AMS input, so that any value other than "geometryoptimization" led to a single point calculation.
|r91318||Windows: force using the shm fabric in Intel-MPI because using OFI introduces a memory leak.
Also remove some smaller memory leaks due to a Intel Fortran compiler bug.
|r91251||ADF: Fix a problem passing SCF convergence failure from the ADF engine back to AMS.
The problem could result in geometry optimizations and other tasks producing nonsense for difficult non-converging cases.
|r91242||ReaxFF: Fix problems in the calculation of per-atom stress tensors for symmetric structures.|
|r91226||ReaxFF: Fix a memory leak bug in the calculation of per-atom stress tensors.|
|r91186||GUI: When using the Intel Ice Lake integrated GPU, clicking on atoms or bonds can select the wrong ones. This change attempts to fix this.|
|r91184||Analysys: fix in Fourier transform in trajectory analysis tool. Before, the spectrum was shifted to the left by one frequency point.|
|r91148||GUI: fix importing lattice vectors for 3D systems from geo files. This would cause for example AMStrain to show the lattice incorrectly for 3D systems.|
|r91120||AMSview: when exporting fields to Cube files show the proper location where they are saved to the user.|
|r91046||AMS: Reduce the threshold on the initial Euclidian distance between end points for the NEB.|
|r91038||AMSspectra: interacting with offset or scaling would show all conformer curves when only average curve should be shown|
|r91028||AMSspectra: fix remembering the width of the broadening function. Opening the same spectrum again could generate a different value, depending on the units used on the X axes.|
|r91018||BAND/AMSinput: Allow the user not to store the DOS per basis function. For large super cells this may be very expensive. See storeDosPerBas in band Band Expert panel.|
|r91014||AMSinput: fix name displayed in preoptimizer (would shoow ReaxAMS instead of ReaxFF, etc)|
|r91010||AMSinput: fix packmol builder with non-orthogonal cells, removing too many atoms in the final step.|
|r90976||AMSinput: update DIM/QM GUI panel on method change. No longer ask to update the full system after a DIM/QM calculation.|
|r90965||AMS: Geometry optimizations with the quasi-Newton optimizer will be performed in Cartesian coordinates by default when a fast engine is used.
The delocalized coordinates can be very expensive compared to the engine.
|r90956||AMS driver: fix issue with input reading of the NEB%Parallel block during NEB restarts|
|r90919||AMSinput: avoid crash in Z-matrix panel when less then three atoms are present.|
|r90912||GUI: fix opening KFBrowser via the SCM menu, and AMSmovie now will not open logfiles any more for AMS jobs.|
|r90880||GUI: starting a module that was already open sometimes started a new module with the same file open. Fixed.|
|r90861||AMSspectra: fix reading old raman spectra (from .t21 files)|
|r90821||AMS2020.102 released (January 12th, 2021)|
|r90820||AMSinput: input key names for most keys were missing in the AMSinput search database.|
|r90802||DFTB: the fix r90772 introduced a bug for response DFTB when using more than one node.|
|r90790||AMSinput: fix test when using dispersion correction with heavy elements (did produce an incorrect warning message when saving the job).|
|r90782||GUI: When using the Intel Ice Lake integrated GPU, clicking on atoms or bonds can select the wrong ones. This change attempts to fix this.
An alternative solution is to use software rendering, which can be enabled by adding SCM_OPENGL_SOFTWARE to your Windows environment variables with value 1.
|r90772.||DFTB: Orbitals coefficients wrong for systems with multiple k-points when using more than one node. This affects only the plotting in amsview of the orbitals and density. It also affect QTAIM for DFTB.|
|r90744||GUI on non-MacOS: on macos Big Sur dark/light mode fix only for mac version.|
|r90738||BAND: low level kf-error about reading occup variable. Only observed for DTB-1/2 calculations.|
|r90736||GUI: on macos Big Sur the GUI was only usable (readable) in light mode.|
|r90719||AMS driver: fixed size of the Gradients variable in the ams.rkf History section for geometry optimizations with constraints using the Quasi-Newton optimizer|
|r90642||AMSoutput: make sure to open the main output file instead of CreateOutputs when started via the SCM menu (with no local file selected)|
|r90612||KFBrowser: the reported size of a variable is now the actual numer of elements (instead of the reserved size).|
|r90603||KF library: Update the number of used elements in a KF variable after writing less data than there was.
Before this change, writing n elements to a pre-existing variable of length m (n < m) did not update its used elements count if it was already larger than n. After the change it will be set to n regardless of its previous value. This should, for example, fix issues when performing a restart from an ams.rkf file from MD calculation in which some atoms were removed from the system.
|r90572||DFTB: Orbital coefficients possibly wrong when using more than one node. TDDFTB results would always be wrong when using more than one node. (A node is not a core, but rather a physical computer)|
|r90559||AMSinput: when setting up a job for the Hybrid engine, # dependencies lines where missing (if needed for a restart).|
|r90528||AMS MD: Fix fractional coordinate handling in the AddMolecule feature (molecule gun) for non-3D-periodic systems.
This affects the FractionalCoords and FractionalCoordsBox input options. The components corresponding to non-periodic dimensions are now correctly interpreted as absolute values in Ångströms as documented. This also includes a fix to performance issues with large values of NumAttempts.
|r90525||ReaxFF: Re-enable OpenMP parallelization in AMS.
All calculationd done using the ReaxFF engine in 2020.101 accidentally ran with disabled OpenMP parallelization.
|r90523||ReaxFF: Reset charge solver when running with disabled predictor and no previous calculation to restart from.
This affects calculations done through the pipe interface (AMSWorker class in PLAMS). Before this change, running multiple independent calculations through the same AMSWorker object caused minor numerical noise in the calculated charges. Setting ReaxFF.Charges.Predictor.Type to "None" now guarantees that the results will be exactly reproducible regardless of any previous calculations.
|r90510||AMSview: Fix opening non-kf files, issue was introduced in r90305.|
|r90485||License code improvement.|
|r90483||AMSmovie: Fix error with active MD replica graphs when no other graphs are present.|
|r90458||Reaxff: Fix a bug triggering ACKS2 error for the Eh pseudo-atom.|
|r90431||AMSinput: Fix polymerbuilder cap with a structure (other than H or None).|
|r90428||ADF and BAND: Workaround precision issues LibXC REVSCAN and REVSCAN0 functional.
However, this does not solve all numerical issues for these functionals calculated with LibXC.
|r90415||AMSmovie: Fix issue when reading MD properties for non-rkf result files (error message: can't read "ix").|
|r90405||AMSmovie: Fix graphs of PBC distance, angle and dihedral for lattice optimizations.|
|r90338||AMS: Print names of the molecules removed during molecular dynamics|
|r90321||AMSinput: resolve a problem with basis set info not working without the corresponding (ADF or BAND) license|
|r90309||Band: remove memory leak in NewResponse|
|r90305||AMSmovie, AMSview: display regions if defined in AMSinput, also for single point jobs|
|r90300||AMSspectra: click in empty space in excitation details would (incorrectly) be handled like clicking on the last line, opening unexpected AMSview windows.|
|r90291||Linux: improvements and a bugfix for the create_linux_icon.sh script.|
|r90253||AMSmovie: Fix x units when adding a Molecule Fractions graph to an empty graph window.|
|r90195||AMSjobs: fix issue with sorting by type for some types (like conformers, hybrid, mlpot)|
|r90186||AMSpython: Fixed incorrect TCL_LIBRARY path on Windows.|
|r90135||AMSjobs: Fix Job Requirements filter found in the View menu.|
|r90008||AMSinput: Fix performance regression of 'Select Within Radius' and 'Select Molecule' atom selection commands|
|r89994||GUI: Fix several issues with periodic bond guessing.|
|r89947||AMSinput: avoid warning popups in basis panel for ADF and BAND when no local license for ADF or BAND is available|
|r89944||AMSinput: handle opening old .adf files using the no longer supported AMS via ADF module without crashing|
|r89931||AMSmovie: Fix reading region info from result file.|
|r89924||AMSinput: Fix charge and spin checks when using FDE fragments. Fix region label for FDE.|
|r89909||AMSinput: avoid check for non-default solvent when no solvent is used.|
|r89899||ReaxFF forcefields: fix typo in CHO-radiation.ff (element O instead of element 0)|
|r89887||AMSinput: Fixed field names in Parallel details|
|r89880||AMSinput: avoid crash when saving a Hybrid job that is missing engine information|
|r89868||AMSinput: fix crash when activating ETS-NOCV before activating the Fragments option|
|r89830||AMSinput: Fix FDE uninitialized region charge error and relaxcycles key. Fix FragOccupations block generation when all electrons get paired.|
|r89826||Windows: switch to using mpiexec from the Intel-MPI legacy directory that supports the -bootstrap fork command line option.
This allows to kill running jobs easier.
|r89800||AMSmovie: fix displaying (e.g. Color Atoms By) properties for non-AMS result files.|
|r89793||AMSoutput: fix displaying UTF-8 characters, fix zero-point energy menu link.|
|r89774||ADF: fix quadrupole intensities XAS in unrestricted calculations.|
|r89752||COSMO-RS: fixed COSMO-SAC Enthalpy calculation.|
|r89746||AMSinput: fix potential issue that could block detecting clicks in the molecule editing window while the bonding indicator is showing.|
|r89738||AMS & ADF: The EngineRestart keyword can now be used to restart the ADF engine from TAPE21 and TAPE13 files written by ADF2019|
|r89718||AMSprefs: Add SCM_NO_RSYNC to the list of available environment variables, it could already be added as custom before.
Use this variable if you encounter issues (mainly on Windows) with using rsync to submit remote jobs.
|r89709||Windows: Update the IntelMPI runtime to 2019 update 9 to fix issues where a job could stay running in the background when it gets killed|
|r89707||AMSinput: total charge automatically adjusting when editing the occupations now also works for unrestricted Spin-Orbit inputs.|
|rr89703||AMSjobs try to fix the error list doesnt contain element 500, in the cleanup of the joblist when quitting.|
|r89682||Python scripting: optimize calculation of inter-molecular distances in oledtools.|
|r89670||AMSjobs: the job list mode will not keep a list of jobs in $ADFHOME (you should not save anything in that folder), and will remove jobs that have no longer a .pid file.|
|r89665||COSMO-RS: fixed volume of compound in liquid phase used in calculation.
This could have an effect, for example, on the calculated Henry coefficient, or on the calculated g/(L solution) solubility calculation.
|r89661||COSMO-RS GUI: add a saved multiple form automatically to the list of compounds.|
Introduce @include directive for including external files in the input
This replaces the INLINE keyword from ADF, which was accidentally removed when ADF was integrated into the AMS driver.
|r89644||DFTB: Symmetrize the DFTB stress tensor. This will not be done when useSymmetry in the DFTB engine input block is set to false.|
|r89623||GUI: Fix issue with automatic OpenGL detection and fallback on Windows.|
|r89600||DFTB: hanging code when storing the orbitals for periodic systems with multiple k-points, using more than one node.|
|r89568||PLAMS: Fix a typo in handling of the settings.runscript.stdout_redirect option by AMS jobs.|
|r89510||AMSinput: Import Coordinates now also shows files with .run extension.
Importing from an AMS run script often works (including lattice).
|r89490||Documentation: change references to ADFinput to AMSinput.AMS.|
|r89478||AMS: Fix order of atoms in the DifDist PESSCan coordinate.|
|r89470||AMSjobs: Fix file transfer from remote on Windows with SCM_NO_RSYNC set.|
|r89462||Band: disable analytical gradients and stress with the Hubbard option (gga+U), because they were incorrect.|
|r89452||Hybrid engine: do some sanity checking on the regions. Engine will stop when it is not happy. Use AllowSanityCheckWarnings to force the engine to continue.|
|r89435||COSMO-RS GUI: allow for making dissociated dimer and dissociated compound+solvent as one of the multiple forms of a compound.|
BAND is now able to calculate charges without calculating also all other properties
This inability lead to the engine failing in rare circumstances, e.g. when the charges were requested via the AMSWorker interface in PLAMS.
BAND: TOO MANY TRIES in didif, for some GGA functionals.
Still three remaining combinations fail: Ac-BLYP, Ac-HTBS, and Lr-OLYP. For the HTBS and OLYP functionals the RadialDefaults%RMin is made larger for light atoms.
|r89351||AMSinput: prevent the Save As question when nothing has changed (would show just by starting and quitting AMSinput).|
AMSdos: partial dos (per basis function) in AMSdos could be wrong for BAND calculations.
The partial DOS shown in AMSbandstructure was correct.
BAND stops earlier when the D4 dispersion correction is used with periodic systems
Before, BAND only stopped after the first SCF, wasting computational time before stopping with an error.
AMS driver: better logfile output for parallelized numerical derivatives
The AMS driver now prints more information on the automatic parallelization of the numerical derivatives loop to the logfile: the total number of displacements, the number of displacements skipped due to symmetry equivalence, the number of worker groups created, the number of processes per group, as well as the task scheduling strategy (static or dynamic).
|r89302||AMSinput: fix changing the angle of three atoms that were all on one line.|
|r89299||AMSinput: backwards compatibility: convert Frequencies runtype, fix default file name when saving.|
|r89295||AMSinput: coloring vector input fields (multiple numbers on one line) fixed.|
|r89291||Fix startup problem for AMSinput and amsprep when starting more than one at the same time|
|r89258||AMS2020.101 released (November 2, 2020)|