Changelog bug-fixes Amsterdam Modeling Suite 2019-3
|r93121||(April 19th, 2021): AMS2019.307 released
This unscheduled subrelease for AMS2019.3 was made because of the fixed PBEc bug.
|r93119||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.
|r90329||(January 4th, 2021): AMS2019.306 released
This is the last scheduled subrelease for AMS2019.3
|r89916||ADF: Fix rotational entropy value after frequency scan (it was incorrect due to using a lower symmetry).|
|r89776||ADF: fix quadrupole intensities XAS in unrestricted calculations.|
|r89619||GUI: Fix issues with automatic OpenGL check and software fallback on Windows.|
|r89467||ADFjobs: Fix remote file retrieval on Windows when SCM_NO_RSYNC is set.|
|r89466||ADFspectra: Better error handling when determining rotational equivalent atoms for NMR.|
|r88181||ADFtrain: fix handling of ReaxFF ffield header flags like ereaxff, acks2 and lgDispersion.|
|r88018||(October 6th, 2020): AMS2019.305 released|
|r87916||Band: XC_orb term wrong in PEDA analysis when using the libxc key.|
|r87902||Band: Prevent stop with ERROR: KMatrixIOModule: inconsistent dim, during a band structure calculation.|
ADF: bug fix: sometimes too few frequencies were calculated for linear molecules.
In case of linear molecules of more than 2 atoms, the number of frequencies calculated with symmetric displacements was incorrect in case of hybrids or range-separated hybrids.
ADF: bug fix: bond energy in case of numerical frequncies may be incorrect.
In some cases, instead of the bond energy of the molecule at the central geometry, the bond energy may be given of the molecule at the last geometry step.
|r87814||ADFInput, avoid hang when creating a slab from bulk material (slicing tool).|
|r87672||Write atoms to a CUB file in the input order instead of the internal ADF one.|
|r87527||Reduce the amount of disk space used by ADF analytical frequencies code.|
ReaxFF: fix a few bugs related to force-field fitting.
Multiple geometries with the same name in a geo file will now trigger an error message. Only one of these geometries will be used to calculate the cost function contribution.
The RUTYPE NORMAL RUN key will turn off any previously specified CELL OPT flag in the same geo file. Before, a NO CELL OPT key was required to achieve this.
|r87276||ADFview on MacOS: color mapping for cutplanes and colored contour lines had color issues after interacting, looking like random colors|
|r87218||ADF: Save some scratch space by not saving AOResponse restart data to TAPE13 files belonging to MPI slave processes.|
|r87155||Band: Error StrainDerElstat: nOper different at perturbed geo, could happen when disabling symmetry.|
|r87147||Band: growing memory usage when using the NewResponse option (when COApproach is true, the default).|
Reaxff: Fixed a bug in readbgf that resets some settings to default values after reading the geo file.
This, for example, results in “RUTYPE CELL OPT 0” being written to the fort.90 instead of a non-zero value from the geo file.
|r86771||Band: Crash on windows in PrepareBas for calculation on large systems without frozen core.|
|r86722||ADFtrain: Fix error when saving force fields with lgDispersion enabled.|
AMS-ReaxFF: Bugfix potentially lost interactions in highly skewed cells.
Non-reactive runs could lose some of the bonds crossing PBC boundaries when the unit cell was significantly non-orthorhombic. Similarly, reactive runs in such lattices could lose some long-range nonbonded interactions (near the cutoff distance), but the effect on results should be small.
|r86682||AMS MD: Fix a bug that caused the CalcPressure input option to be ignored.|
AMS-ReaxFF: Bugfix crashes in non-reactive mode.
Non-reactive runs would crash when the number of the atoms in the system changed (such as in combination with the molecule gun) or in combination with multilevel parallelism (such as when calculating the elastic tensor.
|r86606||DFTB: Erroneous behavior for some elements, like Ba, when using the GFN1-xTB model, due to a too short function range.|
|r86531||Band: special basis sets set by adfinput ignored. Now the ByAtomType Path subkey is no longer ignored by band.|
|r86141||AMSinput: add warning when saving an MD run with DFTB with multiple K points: no bond info, and thus no molcule info, will be calculated.|
|r86082||Make sure the floating license is released for crashed and killed DFTB and BAND calculations.|
|r86035||Improve prediction of the next energy during quasi-Newton optimization with constraints.|
|r86033||adfreport: fixed a bug that would cause the program to crash when processing many (over 60) files at once.|
|r85941||GUI: running remote jobs left tail processes on the remote machine for showing the logfile, cleanup code fixed.|
|r85848||BAND: When using the explicit Occupation options, the occupations would be wrong in the GUI.|
|r85685||ADFinput: Avoid endless warning popups when saving a job with the Run Script window open.|
|r85679||ADFinput: Builder could add an incorrect number of molecules due to round-off errors (one too many or few).|
ADF: bug fix workaround Grimme4 dispersion for (trans-)actinides.
Grimme4 dispersion is not available for (trans-)actinides (Z>86). Workaround in ADF is to use the parameters for the lighter 6th row homologs (Z = Z – 32) instead However, this workaround was incorrectly implemented for elements with Z=87-94.
|r85660||COSMO-RS GUI: Allow space in name QSPR COSMO result file.|
|r85509||ADFinput: Fix polymer builder performance issue and ‘Orient link’ ‘UFF Scan’ option.|
|r85473||ADFinput: use of sumdiff constraint generated SumDist instead of SumDiff constraint in run script.|
|r85329||Reaxff: Fixed format when writing restraints to the result geo file.|
|r85314||AMS NEB: Fixed a small glitch related to NEB restart fix in r84969.|
|r85191||ADFmovie: When opening a GCMC result file, do not disable the Show Accepted Geometries menu (so user can select to show all geometries).|
|r85061||AMS: The InitialHessian option in GeometryOptimization and IRC can now accept a ADF t21 file as input.|
|r84969||AMS NEB: Fix a bug that would cause problems restarting an NEB calculation.|
|r84927||BAND: Fix a bug that would caused noise on the gradients, especially when using many cores, hampering geometry convergence.|
|r84703||AMS CVHD: Fix a bug that would cause problems restarting CVHD MD.|
|r84685||ADFmovie: The labels for the X and Y axes for 2D PES graphs were switched, and show marker in 2D PES graph corresponding to current frame.|
|r84674||AMS driver: The Quasi-Newton optimizer now uses Cartesian coordinates for linear systems with >2 atoms by default.|
|r84625||ADFjobs: Fix cloudmachines error. This can be manually fixed by removing the .scm_gui/CloudMachines file found in your userdir, e.g $HOME/.scm_gui/.|
|r84609||ADFspectra: Moving through frequency peaks with arrow keys would crash after highest frequency when imaginary freqs were present.|
|r84573||ADFinput: Slicer did show slice plane incorrectly. The fix has no effect on the actual slab generated.|
|r84542||ADFmovie: Fix the initial MD Molecules window position when there is not enough room.|
|r84404||ADF: Fix ADF input order versus internal order for Vibrations section on TAPE21.|
|r84340||ADFmovie: Fixed the Graph->Lattice Vectors menu command for systems with dim < 3.|
|r84303||ADF: Fixed a bug in SumDist and DifDist constraints, which should make them more stable.|
|r84219||ADFinput DIMQM: MDC-m charges were used if MDC-D charges were requested, and Hirshfeld charges used for DIMQM could be in wrong order.|
(May 11th, 2020): AMS2019.304 released
This version includes numerous bug fixes, as well as a new software rendering option for the GUI to replace the OpenGL1 fallback mode. For more info see https://www.scm.com/doc/Installation/Remote_GUI.html#ams2019-304-and-vtk7
|r84174||AMS MD: Fix a bug related to combining CVHD and RemoveMolecules.|
|r84139||ADFinput: fix NEGF jobs for ADF (they did not run because of incorrect fragment blocks).|
|r83937||ADF: allow for B97-D with different Grimme dispersion.|
|r83768||ADFspectra: Fix issue with switching modes on Windows, introduced in r83131.|
|r83570||GUI: Fix issue with hanging modules, observed on some macOS machines, related to installed fonts.|
Windows: Fixed an issue in figuring out the number of cores in the machine.
The problem was introduced in r82879 and it affects the number of parallel processes spawned for local Windows jobs.
|r83528||ADFjobs: Fix remote jobs with parentheses in the path.|
|r83523||DFTB: improved performance of serial calculations for large systems|
|r83487||ADF: fix unrestricted calculation of PM localized MOs.|
Windows: upgrade Putty to v0.73 (64-bit).
The new version Putty version is a 64-bit binary. The default RSA key size has been increased to 2048 which should help with “TIMEOUT” errors when running remote jobs. If you are still using a 1023-bit key you may have to create a new, 2048-bit one.
Fixed retrocompatibility issue for band.rkf results files.
Now it is possible to visualize in ADFView Band results files created with AMS2019.1 using AMS2019.3.
|r83255||ADFmovie: Fix handling non-AMS result files that contain systems with different periodicities.|
|r83177||ADFmovie: Fix map atoms into unit cell for non-AMS/ADF result types.|
|r83131||ADFspectra: Fix warning message when hovering over the View->Atom Details (and related) menu.|
|r83026||ADFinput: CV(n)-DFT requires an all electron basis set|
|r83024||ADFinput: CD requires NOSYM|
SCM menu behaviour changed
Now using the SCM menu will try to open the currently open file first, if possible, instead of trying the main job. This is as it was before r82616, except for fixing opening of explicitly selected files in ADFjobs.
|r82879||Corrected an issue in detecting the number of physical cores for Windows machines with more than 64 virtual processors.|
|r82763||ADFinput: unit labels for repeated fields (like Molecule Gun) did not correctly show after redisplaying the page|
SCM menu: uniform behaviour
In all modules except ADFjobs using the SCM menu will act the same as opening the selected module with the current job.
In ADFjobs with a job selected behaviour is the same, when a file has been explicitly selected that file will be opened.
|r82561||ADFjobs SCM menu: when a file is selected in ADFjobs, always open that file (if supported by the module started)|
|r82559||ADFmovie: Change default movie export to mp4 with x264 encoding.|
|r82504||ADFmovie: Fix displaying SDF files with different amount of atoms.|
|r82441||NMR and CPL: Remove the scratch RIHartreeFock file at the end of a calculation.|
|r82369||ADFmovie: Fix error when displaying a PES scan, introduced by r80628.|
GUI: Removed the OpenGL1 fallback in favour of providing OpenGL libraries that use software rendering.
On Windows, usage of the provided libraries is decided automatically at runtime.
On Linux, the environment variable SCM_OPENGL_SOFTWARE needs to be set in order to use the provided libraries.
The old SCM_OPENGL1_FALLBACK environment variable is no longer used.
|r82337||GUI: Fix issues with certain modules (mainly ADFspectra not loading spectra) when used over SSH.|
|r82164||DFTB/band: prevent issue — cannot linearize extreme dispersion — . This was not implemented for 1D and 2D periodic systems. It prevents very high bands picking up small occupations, which coulc happen with very small basis sets.|
|r82157||GUI: Fixed several issues (e.g. pre-optimizer, field visualization) related to TMPDIR permissions on Windows.|
Fixed a license issue for PES point characterization with BAND
The AMS driver attempted to use the DFTB engine for the calculation of the preconditioning Hessian and guess modes, which failed for users without a DFTB license. It now falls back to UFF in case DFTB is not available.
Bugfix error histograms for ReaxFF parameter fitting.
A bug in the analyze_errors script used in the ReaxFF fitting tutorial caused error histograms (error_bonds.csv, errors_angle.csv, etc.) to be shifted left by one bin. The count for the leftmost bin was lost, so the sum of all counts did not match the number of corresponding entries in the training set.
AMS NEB: Improve interpolation in the internal coordinates and add an option to add (valence and/or dihedral) angles to it.
Possible values of the NEB%InterpolationOption key:
1 – only distances,
2 – distances and near-linear valence angles,
3 – distances and all valence angles,
4 – distances and dihedral angles,
99 – all coordinates
|r82031||ADF via AMS was not enabled in ADFinput in some cases.|
|r82010||Densf: add a repulsive NCI region as -3 value of the NCI field.|
|r81909||(February 24th, 2020): AMS2019.303 released|
|r81700||ADFspectra: fix issue with using the menu on Windows.|
AMS-ReaxFF: Return corrected bond orders by default.
The ReaxFF engine now accepts a new input option “ReturnCorrectedBonds”. With the default setting (enabled), the bonds returned to the driver are based on the final bond orders that include valence-dependent correction terms. Disable this if you want to see the raw uncorrected bond orders instead (ReaxFF will still use the corrected bond orders internally to evaluate the potential as usual).
Although classic reaxff defaulted to uncorrected bonds, in practice all our control files always set icobo=1, so most users likely expect seeing corrected bonds (which are also more chemically intuitive). Uncorrected bond orders also make molecule analysis unreliable.
|r81511||Pipe engine, ASEPipeWorker: Fix compliance with the protocol specification.|
|r81425||KFBrowser: fix getting frequency labels from rkf style files|
|r81414||ADFmovie: Fix some issues with displaying electrons and electron holes (from eReaxff).|
Bugfix XZ and YZ components of the polarizability tensor in the TAPE21 Properties section.
Before the fix, the XZ and YZ components of the tensor were interchanged.
Bugfix in the MESA SCF acceleration method.
In rare situation the bug could result in an “All MOs have zero eigenvalue” error. With the fix, the method that produced an invalid Fock matrix will be discarded at the current step.
Updated the python stack on Windows to 3.6.9 to match MacOS
This also includes pip 19
|r81125||ADFinput: Charge key added for Mopac and DFTB when used within Quild|
|r81085||ADFinput: Fix dragging centroids.|
|r81055||ADFInput: GGA:S12g-D3 does not require integration Good for frequencies.|
|r81024||ADFinput: GGA:S12g-D3 can handle analytical frequencies.|
|r81015||ADFinput: Show Elastic wall restraint with old style ReaxFF|
|r81005||ADFinput: Fix unresponsive ADFinput GUI on certain Windows systems.|
Updated the python stack on Linux to 3.6.9 to match MacOS
This also includes pip 19
Updated python stack on MacOS to avoid notarization errors on MacOS Catalina
Updated the python stack to 3.6.9. This includes pip 19. Also removed the pyxenon package, as it gives an error during notarization.
Make changes to avoid notarization errors on MacOS Catalina
Changes are only in packaging, app structure bundle and corresponding initialization code.
|r80812||ADFINPUT: changed the Hcp crystal structure in the database (second basis vector).|
|r80775||DFTB: bugfix for very large systems. Reduce memory requirement (once more).|
|r80701||AMS NEB: Fix a segmentation fault trying to use results of an unconverged SCF.|
|r80690||DFTB: bugfix for very large systems. Reduce memory requirement.|
|r80628||GUI: Fix the initial legend position in certain plots.|
AMS: Fix a problem with long interpolation path for periodic NEB.
If you intend to move an atom across more than half the unit cell in a periodic NEB, set the NEB%InterpolateShortest option to False.
ADF: fix issues with ghost atoms and dummy atoms for COSMO
If radius is not specified ghost atoms will get zero COSMO radii. Fixed bug in Delley surface construction in case of zero COSMO radii, relevant for dummy atoms and ghost atoms.
|r80528||ADFinput: Fix error when using the element chooser on Windows, introduced by r79967.|
|r80234||ADF: Optimize memory usage by ETS-NOCV.|
|r80231||Fix issue that could cause ADFjobs to quit very slowly after running many jobs that have errors|
|r80156||DFTB: fixed a crash when calculating effective electron masses in parallel|
|r80089||AMS2019.302 released (December 12th, 2019)|
|r80088||ADFmovie: Show MD charges from AMS MD calculations (only when not using Fast mode).|
|r80071||ADFinput: Automatically update POTCAR references when changing POTCAR library dir.|
|r80065||ADFinput: Fix unit labels on conformers main page.|
|r80049||ADFmovie: Handle click on PES and NEB surfaces.|
|r79967||ADFtrain: Fix rxffutil options on Windows.|
|r79914||ADFmovie: improve the tab order in the RDF panel.|
|r79890||ADFinput: DFTB does not handle fractional charges, add a check for this.|
|r79775||ADFmovie: fix ‘accepted geometries only’ for Linear Transit ADF results.|
|r79758||ADFmovie: fix atom label positions.|
|r79683||ADFspectra: fix opening .sdf file with conformer spectra which has been moved from its original location.|
|r79673||ADFspectra: clicks were not handled as expected (bug was introduced in r79599)|
|r79669||The SCM menu could in some cases (with subjobs from quild or conformers) open another file than intended.|
|r79659||ADF: Fix a crash at the very numerical frequencies code after thermo() call|
ADF: fix Fragments key in case a directory name is used as block header
ADF would give an error: STRINGS TOO LONG. CONCSS.
|r79615||ADFspectra: fix initial range of some plots|
|r79613||Reaxff: Skip charge equilibration for a single-atom system.|
|r79609||ADFspectra: reset graph range after chaning Boltzman/T averaging|
|r79606||ADF-GUI graphs: only show legend for curves that are visible|
|r79599||ADFspectra: make visualization of spectra of multiple conformers much faster.|
|r79587||ADFspectra: make handling multiple conformers much faster|
|r79586||ADF-GUI: make render blocking work nested|
|r79578||Reaxff: Do not remove atoms on the last MD step to avoid discrepancy between the last saved molsav and fort.90 files.|
|r79576||ADFmovie: Fix display issues with qkf/runkf or non-movie result files.|
|r79563||ADFinput: remove ghost atoms when switching to any method other than ADF or BAND|
|r79560||ADFjobs: icon for VASP jobs is now blank|
|r79554||ADF-GUI: reposition Microkinetics in the SCM menu|
|r79548||Reaxff: catch user mistakes in the restraints input|
|r79545||The default perspective for viewing is now set properly in all modules, not just in ADFinput|
Conformers filtering was broken.
The Filter by rms option for SDF files was broken (causing a crash with a pythin error about a wrong index)
|r79521||ADFmovie: Fix IRC trajectory reordering for adf results.|
Reaxff: Fixed a bug determining frozen atoms when a geo file contains multiple geometries
This bug could result in atoms specified as frozen in one geometry would also be frozen in subsequent geometries
|r79494||GUI SCM menu: show warning if nothing can be opened (instead of silently doing nothing)|
|r79479||ADFtrain: add Delete Selection menu commands with indicator that Backspace can be used for the same purpose|
|r79477||ADFtrain: save params file with hash sign in front of disabled parameter optimziation data to avoid them being optimized|
|r79461||Tutorials: add minor instructions to the generating molecules tutorial|
|r79445||DFTB: removed output spam for periodic D2 dispersion correction|
|r79439||include minor version number (301) in printed header of all programs|
|r79438||ADFmovie: Show graph markers for Replica Exchange MD.|
|r79421||ADFmovie: Fix selection issues with trajectories involving added/removed atoms.|
|r79399||Added examples/AMS to the binary distribution.|
AMS: Fixed a bug in the NEB that would ignore frozen atom constraints.
Pass the constraints to the optimizer for end images optimization. Disable interpolation in internals and initial alignment of end images if frozen atoms are present.
|r79386||Fix issues with conformer generation caused by wrong (aromatic) bonds.|
|r79304||DFTB: fixed a bug that could cause NaNs in the DFTB stress tensor for large systems.|
|r78301||AMS2019.301 released (November 11th, 2019)|