When using the builder and selecting a molecule via the file select icon, the builder would crash.
For some space groups (between 149 and 194) and some wyckoff positions the generated postions were incorrect.The affected groups are 149, 151, 153, 156, 160, 162-166, 177-183, 186-188, 191-194. At the same time the special handling of space groups with the Trigonal-R lattice was reverted (introduced at revision r33879).
To use the feature, set dumpcb to 1 in the control file. To get a list of bonds
at iteration 100 after a calculation, use the following commands:
adfreport reaxout.kf "History%strong bond atoms 100#6d##2"
adfreport reaxout.kf "History%strong bond energy 100#12.5f##1"
The first command will get you atom pairs for each bond and the 2nd command will get you the corresponding energies, in kcal/mol. Which bonds are listed is controlled by the cutof3 parameter in the control file, which specifies the minimum bond order and is equal to 0.3.
Some of them could fail when given without atoms present.
Fixed a bug in srtbon2 used with non-zero iconne in the control file. This bug would result in incorrect bond order for pi-bonds.
Due to a typo this force field file did not show up in the force field menu in ReaxFF GUI.
Sometimes ADFlevels could end up hanging when selecting a level.
Interaction lines were sometimes incorrectly drawn (sometimes not taking the extended width of degenerate levels into account).
Now it should again be possible to watch the progress of a reaxff calculation in ADFmovie.
Now the RUNKF result file is copyied back by the .job script, no longer the .run script.
In such a case, the QMMM will be initialized from scratch.
Sometimes multiple jobs started running, hopefully fixed
If the integration option was used, the produced files would have overlapping names (for integragtion 2.0 and 2.5 etc)
The XC potential menu has been fixed: it is different for ADF and BAND. Only the BAND options were shown, thus the hybrids did not show up for ADF calculations.
The GUI only supports restarts from the KF result file
The lattice vectors history (only in adfmovie) where incorrectly read (transposed).
In rare cases periodic systems were opened in non-periodic mode
Fix a bug in unrestricted collinear spin-orbit coupled calculations of the g-tensor in case of metaGGA's and meta-hybrids. Note that still no GIAO's are used in this case.
To use the feature, set dumphb to 1 in the control file. To get a list of hydrogen bonds
at iteration 100 after a calculation, use the following commands:
adfreport reaxout.kf "History%strong Hbond atoms XHO 100#6d##3"
adfreport reaxout.kf "History%strong Hbond energy 100#12.5f##1"
The first command will get you atom triples for each hydrogen bond and the 2nd command will get you the corresponding energies, in kcal/mol. In each atom triple, the 1st atoms is the acceptor, 2nd - the H and the 3rd atom is the atom bonded to H.
Occasionally the D3 dispersion correction may yield a bumpy potential energy surface. In such cases lowering the k3 parameter will smoothen the surface. The input key is Dispersion%k3. The default is 4, the advised alternative is 3. This effectively defines a new functional.
The command line arguments passed to ffmpeg have been changed to make them compatible with the latest ffmpeg. Also, on windwos the default movie type is now .avi, on other platforms .mp4. Note that you can specify the file name with an extension, and that the extension actually determines the file format to be produced (via ffmpeg).
If EDITOR is set in the environment, that editor will be used. Otherwise gedit is the default. If gedit is not found, vi inside an xterm window is the default.
When you were viewing a movie with a graph, and then used the Open command to open a different file to view, the graph data was not properly reset. As a result the graph shown was incorrect. This has been fixed.
A minor new feature: when using the Open command in ADFmovie, now all recognized files are listed.
In some GUI modules pressing the ESC key caused an error message.
Users who downloaded a r34238 version should update to r34245 or later. The change introduced with r34235 causes rxkf files to be corrupted.
Another fixed bug is related to non-reactive part of reaxff. Without the fix the number of molecules can still change with time. Note however, that even though the bonds identified at the first step of simulation are kept throughout the non-reactive steps, their bond orders do change over time. This means that analysis of molecules based on the updated bond orders may result in molecules different from the 1st step.
Some basis sets have empty numerical orbitals. In combination with spin-unrestricted calculations the error message InitialPMat: qqq=0 would appear.
The MOPAC2012 executable was not included in the distribution by mistake
The PDB trajectories are an experimental feature for now, feedback and suggestions welcome.
For the space groups 146, 148, 155, 160, 161, 166, and 167 the wrong lattice vectors were used.
This non-default use of unscaled ZORA spinor energies produsced incorrect results for spinorbit coupled NMR icw hybrids.
Fixed a bug that (on some platforms) could cause ignoring of some of the COSMO surface points.
The energy function during geometry optimization contained the MM energy twice. This has been fixed now. The bug existed in all ADF versions.
Any calculation done with the NMR single-dipole method with more early band 2012 versions should be considered wrong. For NMR supercell method there is now an option that will make the result independent of lattice translations of the NMR atom, so that the user does not need to worry whether or not the probed nmr atom is central in the unit cell.
The DFTB.org parameters need to be enabled explicitly in the license file. Currently by default they are not.
Now the GUI will warn users of the problem.
When making a slection by dragging a rectangle (with left mouse and shift key), doing this again will now extend the selection with the new items (as it did in ADF2010).
In some cif file the translational part of the operators is given as a floating point number rather than as a fraction. This was not properly recognized.
When using the SSB-D funtional this incorrect warning appeared--- WARNING: Unknown GGA used. Dispersion added with factor 1.0
The t21 files now are opened in ADFview, not in the KFbrowser, when giving them as cmd line argument to adfjobs.
This should also fix opening by drag-and-drop on top of ADFLaunch.
The ...-water force fields have been renamed to ...-H2O, some force fields have additional elements (based on the references shown). The old NiCH force field contained also parameters for other elements, but the origin of these parameters is not known. So it has been moved to undocumented, and the new NiCH force field has only the Ni C H atoms, matching the references.
The GridBasesAIM keyword did not work when there were no frozen core electrons. The AIMBondPaths keyword did not work when the number of critical points exceeded 128.
The lattice vectors were wrong when not all angles were 90 degrees.
ADFview would present an error about ::t21::nProperties
Fix the "can't read "data( C,ribbon)": no such element" in array error
During a geometry optimization with constraints the diagonalization of the Hessian could fail in a low level routine DSTEGR2A.
It actually might make sense to have spin polarization 0 when specifying occupations.
Selecting None did not do anything, and some consistency checks have been added
Bug fix in writing Fock matrix on AOs for beta spin.
For periodic systems you can translate the system with respect to the cell. This was broken.
Some CIF files have multiple DATA elements. The importer searches for the first one that contains atoms.
If you first make a crystal with bonds across unit cells (using the BANDinput mode), next switch to ADFinput, the resulting ADF job script contains a malformed GUIBONDS key. This causes ADF to crash. A work-around is to delete the GUIBONDS block key from the run script.
Fixed a bug in the bond energy analysis (EDA) for the Pauli term in case of metaGGA's and metahybrids, which could result in large numbers. The final bond energy was correct.
After making it non-empty the user was unable to make a field empty again.
In some situations deleting atoms in a periodic system could break the GUI (non-existent bonds were deleted).
Errors in the following forcefield files have been fixed: AB, CHO, ZnHO. ReaxFF users who run calculations with these forcefields should update their ADF distribution. The errors in AB and CHO forcefields mainly deal with hydrogen bonds, which is probably not that important in the systems for which these forcefields are targeted. The errors in the ZnHO file were more severe so one should repeat their calculation using this forcefield.
Instead of stopping of ADF, ADF will now print warning messages like: "WARNING: COSMO delley program error ...", and ignore problematic surface points in the COSMO Delley surface construction.
In rare cases, a click on a bond would select a different bond.
Use GuessBonds if you are opening an old file that still gives troubles.
Note bonds are not used in the calculations (they are for display only)
Now the lattice vectors are correctly pasted for example when copying a structure out of a BAND-logfile.
For periodic systems, when reading a logfile of a calculation where the lattice was optimized: only the lattice vectors for the last iteration were used.
The bug was introduced in r32444
If you have a periodic structure first with bonds between cells, and then switch to ADF, you could not save the run script. This has been fixed/
The input file Cm.4f for the auxiliary program dirac, which calculates the frozen core potential for the basis set Cm.4f has been fixed.
The first frame would contain the wrong atom positions. This has been fixed, and now also the gradient background is retained.
Another bug fixed is where switching could cause drawing problems when the selection was not empty.
The basis set dependency criterion was changed from 1e-5 to 1e-8, thus avoiding often encountered (unneeded) dependent basis errors.
There was a problem with short loops that had a string value at the same line as the key.
It was impossible to save a BAND job with the GUI.
There was a bug in ADF for parsing of Grimme DFT-D3 correction parameters: when a parameter was explicitly set to zero, ADF incorrectly replaced it with the PBE value. This has been fixed. This can be a problem for the Minnesota M06 family where the s8 parameter (PAR3 in ADF) needs to be zero. Setting it to 1e-99 is a workaround.
ADFview offered to display the WS cell even if the data for it is not available. When a users used the menu, ADFview would crash.
Unrestricted collinear spin-orbit coupled calculations of the g-tensor in case of metaGGA's and hybrids and metahybrids now uses common gauge, which makes them dependent on the origin of the molecule. Thus the GIAO formalism for the unrestricted collinear spin-orbit coupled calculations of the g-tensor is only implemented for LDA and GGA's. Note that the g-tensor calculation that perturbatively includes spin-orbit coupling does use GIAO's for hybrids (and LDA and GGA's).
Fixed a bug for unrestricted spin-orbit coupled hybrid calculations in the collinear and noncollinear approximation. This bug still exists in older versions of ADF.
Communicate the amount of memory between nodes to make sure all nodes calculate the same number of passes
The transfered files could end up in the home directory instead of the job directory.
When using the transfer coordinate feature (where you push the coordinates of some selected frame in ADFmovie to ADFinput), the bonds could be wrong in cases of periodic systems when molecules cross the unit cell boundary.
This is a significant rewrite of the code that handles periodic details, both for setting up jobs and for visualizing results. For the user the effect is that display of periodic systems is much faster, bonds between periodic cells are now visible (created by GuessBonds), and many bugs have been fixed that had to do with periodicity.
A minor detail is that for more atoms colors have been defined, and the default color has changed. If you do not like the new colors, you can change them via the Preferences (in the SCM menu).
When you have internet connection troubles while running a job, the .pid file could disappear. This should be solve (but is hard to test...).
Fixed a bug for unrestricted spin-orbit coupled hybrid calculations in the collinear and noncollinear approximation, in case the calculation was performed in parallel. Note this bug is fixed only in ADF2012 and the development snapshots. This bug still exists in older versions of ADF.
After the previous fix, tail commands could still be around after a job has finished.
The program occasionally detects a too low symmetry. With lattice vector optimizations this could lead to the error: mode break symmtry.
Slightly modify analysis of Mo→Mo contributions for TDHF, such that no negative contributions appear.
For hybrids it is possible to change the amount of Hartree-Fock exchange in the input.
XC Hybrid B3LYP HF=0.25 ENDHowever, for excited state optimizations, the HF percentage was not recognized, and an error messaage was produced: 'Error in SetHybridStrings: kernel of this hybrid functional not available'. This has been fixed.
Often the main menu did not work properly, fixed. The alternative of changing the molecule view option via the regions details was not broken.
Thus for periodic calculations with DFTB or UFF the lattice vectors were incorrect in the visualization of the results if the lattice vectors are not symmetric.
The program might stop with: check idempotent too bad. This occurred for systems with a non symmorphic space group.
The most common occurance will be if your MOPAC license is expired while you still have a valid license for other GUI modules.
A work-around is to delete the expired license targets manually from the license file (remove also the encrypted lines for these targets, otherwise your license file will be corrupt).
ADFspectra required a .t21, .runkf or .rkf extension when opening a file. This requirement has been removed, so users can now easily open a result from from an FCF calculation.
ADF may hang due to a deadlock in situations when running in parallel on multiple nodes. The deadlock occurs when some nodes have only one process of a job running while other nodes have more than one process from the same job. The case when all nodes have either multiple processes or a single process from the same job running is not affected by the bug.
A tool to read the Mopac results (tokf) was not properly included in the distribution. As a result, visualization of Mopac results did not work.
When slicing a crystal, making a slab, the resulting atoms would have their number of connectors reset to 1. This has only effects on the visual display of slabs.
Two example .run files for the use of the green module (non-self-consistent Green's function method) are added in $ADFHOME/examples/adf/green_Al.
ADF2012.01 has been released. Please see Release notes for a list of new features and improvements.