Changelog bug-fixes Amsterdam Modeling Suite 2021
| Revision | Description |
|---|---|
| r103044 | AMS: Added a check for normal modes in symmetric displacements in combination with periodicity (which is not supported). |
| r102736 | ADF: bug fix single point gradients dDsC dispersion correction. |
| r102703 | AMS: Fix a few minor bugs in the shared array locking. |
| r102598 |
AMS: Work around a bug in Cluster MKL on AMD Zen that may cause AMS to stop with an “Error exit from DistributedMatrix%SVD” message.
IRC: save engine rkf files for the final points on the IRC path; enable calculation of optional properties for the initial IRC point. |
| r102383 | AMS: Add a possibility to configure the KF page size (=file size increment) via the SCM_KF_PAGESIZE environment variable (which is in kB). |
| r102374 | AMS: fix a bug in IRC that prevented it from computing optional properties (such various ADF chanrges) at converged reaction path points. |
| r102161 | AMSjobs: avoid crash when the modification time of a job cannot be determined. |
| r102137 | Only compute transition dipole moments between excited states if asked for. |
| r102050 | BAND: bug fix for band NewResponse when using the ActiveXYZ option with the X component disabled. . ERROR: kfOpenSection: Attempt to open nonexistent section IndDensVec1Re IndDensVec1Re. |
| r101852 | ADF: Fixed a bug that made it impossible to restart a partially done analytical frequencies calculation from TAPE13. |
| r101126 | Windows: Replace the GetNumCores utility with a new version that should work better under Windows 11. This update should solve issues with AMS crashing inside the Intel-MPI library during initialization. |
| r101004 | ADF: fixed a string overflow when using restart orbitals from a file with a long path name. |
| r100769 | ADF+NBO6: Fixed a bug for writing Fock matrix to file in case of hybrids. |
| r100583 | AMS: Fixed a bug in generation of delocalized coordinates for a highly compressed periodic linear chain. |
| r100041 | AMSinput: Fix checking the correct license target for the Mopac pre-optimizer. |
| r99925 | AMSpackages: bugfix expired modules should no longer be presented as licensed. |
| r99899 | DENSF: bugfix writing cube files for multi-components values such as gradients and spinors. Before the fix, the x component was used when writing values for x, y, and z components. |
| r99803 | ADF: bugfix input reading and documentation of custom solvent SM12. |
| r99751 | BAND: bugfix for serializing the TAPE10. The last line in the log file would be “saving tape 10”, and the program would crash. This would happen for instance with a QTAIM (bader) calculation. |
| r99722 | Densf: bugfix in MO gradient calculation. |
| r99350 | AMS2021.106 released (December 6th, 2021) |
| r99329 | AMSinput fixes for occupations. |
| r99169 | The TestJob now also checks some floating license details, and produces a more detailed license check output. |
| r99129 | AMSpackages: Fix a warning message. |
| r99106 | Fixed a bug in a common library function dealing with KF files that would not trigger a parallel program (for example, nmr) to exit gracefully if a mandatory KF file was missing. |
| r99093 | AMSinput fix crashing but with settings occupations (the bug was introduced in r98924). |
| r99076 | AMSjobs TestJob: also generate detaile license.log |
| r98925 | AMS2021.105 released (November 29th, 2021) |
| r98924 | AMSinput: Spin and Occupations panel limitted to show only 5 virutals (for speed). |
| r98918 | AMSpackages: Workaround potential issues with parsing of license information. |
| r98904 | AMS pipe on Windows: Work around an Intel Fortran bug that would cause AMS pipe to hang when receiving data larger than 64kB. |
| r98898 | AMSPackages: Improve error message description in GUI. |
| r98828 | DFTB: Add a check for dipoles in System%ElectrostaticEmbedding in combination with periodicity. |
| r98702 | BAND: Prevent (harmless) warnings about wrong arguments in call to zgemm in case of Non-Collinear calculations. Does not affect the results, as this problem was about zero-sized arrays. Th warning looked likeIntel MKL ERROR: Parameter 10 was incorrect on entry to ZGEMM |
| r98645 | AMSpackages: Fix issues with the package mananger when editable python packages are installed by the user. |
| r98527 | AMSinput: Fix license check for running ReaxFF or MLPotential as pre-optimizer. |
| r98322 | Reaxff: Fix a bug that would lead to a segfault in a multithreaded calculation when too many errors are triggered and ignored, for example during a force-field optimization. |
| 2021-10-29 | Windows: Fixed a bug in the installer that would make it partially ignore a different location for the ADF_DATA folder set by the user. |
| r98093 | AMSspectra: Fix the unit conversion when updating the ‘Bias steps’ for NEGF current plots. |
| r98020 | Fix a bug that would cause programs of the AMS package (including the GUI) to crash with a segmentation fault when opening a corrupted rkf file, instead of returning with an error message. |
| r97937 | AMS Reaxff: Fix a bug that would cause Reaxff to stop when trying to combine molecule gun and charge constraints. |
| r97918 | AMSinput: minimum value in distance input (next to slider) reduced to 1 pm |
| r97908 | ADF: fix crash spin-polarized G0W0
However, spin-polarized G0W0 is still not implemented correctly. |
| r97906 | AMSPackages: Fix utf-8 encoding errors when installing/checking packages. |
| r97864 | AMStrain: fix import of remote SDF results |
| r97709 | AMSinput: Preoptimizer now checks the engine license before running to generate better visible error message. |
| r97610 | AMSinput: Fix ‘couldn’t open config’ when saving the runscript. |
| r97600 | Reaxff-AMS: Fix a bug in ACKS2 charge constraints (important for e-Reaxff with AMS).
With the bug, the charge is always constrained to zero instead of the value provided in the input. Users who run e-Reaxff calculations with AMS should install an updated version. The Reaxff Classic is not affected by the bug. |
| r97577 | BAND: NEGF error: “types do not match for left lead”. Could happen with a self consistent NEGF calculation on a lead with multiple atom types. |
| r97566 | BAND: Spin-flip option would do nothing for the case of Spin-Orbit coupling. |
| r97517 | Reaxff Classic: fix a bug that would ignore the optcel control flag. With the bug, the only way to optimize the unit cell was to set the RUTYPE in the geo file. |
| r97472 | Bugfix ADF: do not use A-DIIS when AccelerationMethod is set to SDIIS. |
| r97386 | AMS driver: fix an issue with the xyz-files written by the AMS driver for elements with three character symbols (e.g. Uue) |
| r97254 | AMSpackages: Update documentation, provide a python script for downloading repository copies more easily. |
| r97178 | AMSinput: Use the minimum image convention for the initial values of constraints and restraints with periodic boundary conditions. This relates to fix 96331. |
| r97169 | AMSspectra: Fix export NEGF transmission curve as XY values. Give a warning and refuse to write an empty file if no peak data is available when using export XY peaks. |
| r97106 | AMS2021.104 released (September 14th, 2021) |
| r97105 | BAND: Improved printing of the total energy (print totalenergy) for band. For heavy systems or systems with many atoms stars (***) would be printed rather than total energies. |
| r97007 | Windows: fix missing components for using AMSJobs with remote queues. This was broken in the 2021.103 release. |
| r96996 | GetLicense dialog: first line (no valid license found) has been removed as it caused confusion. |
| r96797 | AMS2021.103 released (August 26th, 2021) |
| r96783 | Python: The version of ipython was updated to 7.16.1 to solve an issue with Qt5 and matplotlib. |
| r96776 | Windows: Update and deduplicate MKL inside of the python stack, just like it is done on Linux and MacOS. This saves some space in the installer and on disk, and also avoids some conflicts when running python code from within AMS, or vice versa |
| r96745 | AMSinput: avoid coordinates check when verifying t21 for use with occupations. |
| r96740 | AMS MD: Fix molecule gun insertion failures in non-3D-PBC systems with the ContactDistance option. |
| r96704 | AMSlevels: Fix crash when viewing specific unrestricted results, introduced by r94802. |
| r96649 | AMSpackages: Fix compatibility issue with cloud licenses./td> |
| r96556 | ADF: Fix Md finite nucleus.
The so called dirac calculation for Md with a finite nucleus would fail. |
| r96537 | IRC: Fix a compatibility issue between IRC and the external electric field. |
| r96533 | AMSinput: Warn the user when trying to use the Write history “All” option. Because of the severe performance impact, this option should only be used for testing. |
| r96454 | GUI: Shortest pbc angle and dihedral are now independent of ordering (e.g. A-B-C or C-B-A). |
| r96425 | AMSspectra: Show spin info for transitions for unrestricted excitation calculations. |
| r96416 | AMSinput: Don’t force unrestricted when fragment spin is 0 instead of empty. |
| r96406 | AMSinput: Fix progation of the diagonal input in the supercell dialog, bug introduced in revision 96174. |
| r96372 | GUI: Fix display of the angle when atoms are overlapping, which is more likely to happen with periodic systems displaying the shortest pbc angle. |
| r96331 | AMS: Fix a bug in geometric constraints in combination with periodic boundary conditions. The constraints now use the minimum image convention. Before the fix they would always use atoms from the same unit cell, which would lead to wrong results for constraints crossing a unit cell boundary. |
| r96221 | GUI: Bond guessing involving metallic atoms and hydrogen would in too many cases skip those bonds. |
| r96179 | AMSjobs: Right clicking on empty space would create a popup menu that could not be dismissed. |
| r96174 | AMSinput: Bugfix for presets to make super cells in the 2D case: The sqrt(3) x sqrt(3) preset was not shown when the angle between the lattice vectors was 120 degrees, rather than 60 degrees. |
| r96145 | AMSinput: QTAIM and IQA now availble with relativity, except for atomic energies. |
| r96122 | GUI: on Linux when the new file dialog encounters an error (due to too old system libraries) it now automatically sets SCM_DEFAULT_DIALOG to 1 in the prefernces to fall back to the old file dialog. |
| r96111 | AMSinput: importing specific step from MD RKF files would import wrong lattice and bonds. |
| r96093 | AMSinput: constraints for MD offered the option to constrain regions, but this is not supported in AMS. These are now converted to atomlist constraints. |
| r96089 | AMSmovie: fix exporting 1D and 2D system trajectory to BGF format. Add check for other formats when atoms are added or removed during the trajectory, this is currently not handled. |
| r96060 | BAND: two extra fields available for :plotting laplacian[rho(valence)] and laplacian[rho(deformation/fit)]. |
| r95991 | AMSpackages: better error message for download failures. |
| r95984 | AMSpackages: provide more information for repository errors. |
| r95877 | ADF: Fix RPA correlation energy.
The RPA correlation energy was calculated a factor 2 too large. |
| r95870 | AMSinput: remove the option to delete the only system block in AMS expert keys. |
| r95860 | BAND: Improvements for non-collinear calculations. For big systems the memory footprint is reduced, avoiding a crash. Furthermore a change has been made to the SCF procedure improving the convergence. Still converging a non collinear calculation is often difficult. |
| r95857 | AMSinput: do not add the coords key for the molecule gun when a fractional coords box has been defined. |
| r95854 | AMSbands: the BZ would be invisible due to wrong camera positioning (for example in the phonons tutorial). |
| r95840 | AMSinput: molecule sink should accept * as formula, fixed. |
| r95787 | AMS: Fixed a bug in the molecule gun that could result in NaN velocities if EnergySigma was too large. |
| 95783 | AMSpackages: Bugfix. package manager can now find licenses even when AMSBIN is not in path. |
| r95768 | ReaxFF documentation: mention force-bias Monte Carlo support. |
| r95737 | All: improve messages produced in case of license issues, add license debug mode |
| r95692 | AMSpackages: Bugfix. No longer produce an unhandled exception when an installation is attempted without a valid license. |
| 95621 | BAND: Error “Fragments cannot be assigned by a simple translation” when visualizing from amsview. |
| r95566 | DFTB: Bugfix. The finite electronic temperature (300K) should also be the default for periodic calculations usig more than one k-point. This improves the SCC convergence for such systems. |
| r95509 | AMSpackages: Bugfix. When updating a package, new dependencies will no longer be skipped. |
| r95464 | In case of license errors print out the #user line as found in the license file. |
| r95424 | BAND: Added expert keys SCF%PrintAllOccupiedBands and SCFPrintAllVirtualBands. Affects only the text output where the ranges of the bands are printed. |
| r95409 | AMS: Add a VaryTrustRadius input option to the GeometryOptimization%Quasi-Newton%Step input block. Setting this option will override the default behavior, which is to allow the TR to change during energy minimization but not during a transition state search. |
| r95387 | ADF: QM/FQ fix: include screening of interaction between MM atoms and QM density. This fix is to avoid unstable results in case numerical integration points are accidentilly close to MM atoms. |
| r95383 | AMSmovie: Avoid segfault on MacOS with NVIDIA drivers. |
| r95337 | AMSpackages: Use of “~” to refer to home directory inside configuration files now works as intended. |
| r95332 | Doc: Mention SCM_SHAR_NONUMA in the installation manual. See below for a description. |
| r95222 | BAND: Bug fixes for NewResponse. The results were not the same as for OldResponse, even if converged with k-space integration. The main effect of the bugfixes is that the susceptibilty is roughly twice as large. However, it is possible that the spectrum is changed in a less trivial way. The dielectric function is also affected as it depends on the susceptibility. As the induced density is larger, the self-consistent procedure for each frequency tends to take more iterations. |
| r95211 | AMS MD: Fix bugs in the ContactDistance option of the AddMolecules feature (molecule gun) leading to atom clashes and explosions. |
| r95202 | Documentation: Added documentation for the “update geometry” dialog in AMSinput. |
| r95194 | AMSinput: remove debug output for 2D cell conversion. |
| r95086 | AMSinput: fix the check for available Hessian when opening an input that has previously been run. |
| r95074 | AMSpreferences: the default pre-optimizer should be ForceField instead of UFF (though both do the same so the change is cosmetic). |
| r95042 | AMS2021.102 released (June 8th, 2021) |
| r95023 | AMSspectra: fix NMR atom selection with multiple results open. |
| r94996 | Provide more documentation for installing Quantum ESPRESSO and the COSMO-RS Database |
| r94972 | AMSspectra: fix NMR splitting bug introduced in r94876. |
| r94963 | AMSpackages now presents the user with documentation upon connection errors. |
| r94939 | AMSinput: fix update issue (like coordinates not updating) when opening a .ams file for a system with 1 atom only. |
| r94926 | AMSinput: undo previous fix in r94919, causing issues with Edit-Clear on linux. |
| r94919 | AMSinput: selection in region panel text was not cleared when clicking somewhere else, leading to visual confusion. |
| r94901 | AMSmovie: fix screen update issue when starting without file |
| r94896 | AMSinput: fix deleting atoms with Atom Details panel open. |
| r94880 | Densf: fix conversion of cub files. |
| r94876 | AMSspectra: fix handling of multiple NMR spectra. |
| r94866 | AMSjobs: prepare tool now works when spaces are in the path to the source job(s). |
| r94862 | AMSspectra: avoid crash when opening AMSspectra after conversion to inverted units in the previous run. |
| r94855 | Spreadsheet export: fixed an issue where the formula character limit was exceeded for large jobs. |
| r94802 | AMSlevels: improved memory consumption which allows larger systems and basis sets to be viewed. |
| r94786 | AMSinput: centroids attached to two atoms did not drag correctly. |
| r94783 | AMSinput: after using the Builder with a non-periodic program, make sure to remove periodic images. |
| r94775 | AMSinput: input option to select molecule source (in Coordinates panel) now is disabled when not used. |
| r94766 | AMSinput: changing modules will not change lattice vectors (it may change the periodicity if needed). |
| r94746 | AMSinput: when updating geometry after an optimization, only update the lattice for periodic systems. |
| r94728 | AMSinput: when many irreps are present, the occupations panel needs scrolling. Align left the Show Levels button, and list irreps to make user aware of this. |
| r94726 | BAND: two uninialized variable issues. This may affect NewResponse and Voronoi spin polarization charge analysis. The issues wer most likely harmless however. |
| r94721 | AMSinput: disable some properties that MLPotential could not handle, like bond orders. |
| r94716 | AMSview: avoid crash when changing grid type with non-specified fields present. |
| r94712 | AMSview: avoid crash when changing grid type with non-specified fields present. |
| r94695 | AMSspectra: fix sorting of the VCD details table when selecting an entry and fix scale stored vectors when there are no vectors stored yet. |
| r94683 | BAND: DOS was wrong (huge) when using a finite electronic temperature. |
| r94675 | AMSspectra: fix crash that could sometimes happen when adding multiple curves to AMSspectra |
| r94657 | PLAMS: fix RKF trajectory reader for non-periodic and 1D/2D periodic systems. |
| r94653 | Analysis: fix autocorrelation velocity vector with non-existent indices 4 5 6 did not raise error. |
| r94648 | AMSmovie: fix reading energies for a running MD calculation when skipping through frames. |
| r94611 | AMSpackages: fix AMSpackages gui being named “Python” on MacOS, reliability improvements. |
| r94549 | AMSinput: fix molecule name after undo of delete molecule when having multiple molecules |
| r94539 | AMSinput: fix converting 2D cells to primitive or conventional. |
| r94538 | AMS: Added an SCM_SHAR_NONUMA option to disable splitting machine into NUMA nodes.
Set the SCM_SHAR_NONUMA environment variable to a non-empty string to disable NUMA. This option may be useful to speed up large calculation with a hybrid XC functional (which normally uses a lot of shared memory) on a dual-socket machine with not so many cores (say less than 32-64) and limited RAM. |
| r94530 | AMSmovie: Fix crash on certain machines or wrong graph background when using software rendering, caused by an anti-aliasing issue. |
| r94527 | AMSjobs: Fix potential errors during joblist cleanup of remote jobs at exit. |
| r94518 | AMSinput: POLTDDFT in ADF had issue when number of frequencies was changed from the default. |
| r94512 | AMS: Fix a problem causing errors writing to logfile on Windows. Fix the size of the shared memory used by AMS in multi-socket Linux configurations using NUMA (use the NUMA node’s memory size instead of the machine’s). |
| r94490 | BAND: Improvements for DOS. The total DOS is now always calculated and saved, the PDOS by default not. Also the option IntegrateDeltaE=no works, to use the exact delta function for the DOS, rather than the histogram-style (integrated over the energy intervals). |
| r94482 | BAND: Error “Fragments cannot be assigned by a simple translation”. Could also happen when visualizing from amsview. |
| r94455 | AMSpython: fixed a bug where the user’s python virtual environment would sometimes not find the Python packages included with AMS. |
| r94442 | Fix a bug in the periodic HF exchange. |
| r94438 | AMSpackages: Bugfix for package manager for MacOS GUI
The package manager GUI on MacOS may produce an error message and freeze when installing packages. This can occur if it is started from GUI windows that were opened directly from the MacOS dock. This bug has been fixed in AMS2021.102. |
| r94406 |
AMS2021.101 released (May 18th, 2021)
List of known issues and bugs in AMS2021.101
|