An essential update for all users of ATK 10.8.x has been released. Users who are still using version 2008.10 are strongly encourarged to upgrade too!

For details on ATK 10.8, see the release notes for version 10.8.0. Below we will only list the updates compared to that version.

This release does not contains any new functionality compared to 10.8.1, but users of that version should upgrade since there are several bug fixes in this release specifically for problems with 10.8.1.

You can obtain the new version from the Download page. (NOTE: There is now also a GLIBC 2.3 version of ATK, for Red Hat/CentOS 4.8 and similar Linux distributions.)

What's new in 10.8.2 compared to 10.8.0

• Molecular eigenstates for bulk and 2-probe (a.k.a MPSH eigenstates)
  • In 10.8.0 only the eigenvalues could be computed, now you can also visualize the states in 3D.
• Contour integration
  • Completely redesigned algorithm, leading to greatly improved convergence at finite bias.
• Forces can now be visualized in VNL
• Show movies of relaxations in VNL
• More advanced result view of DOS, with projections (was earlier a Custom Builder, is now an integrated feature)
• Import VASP geometry files directly
  • Just drop the POSCAR on any VNL instrument
• Calculation and visualization of Bloch states
• Native 64-bit version for Windows
  
• Fixed a critical MPI bug which led to crashes when running parallel calculation with a finite bias (and possibly poor convergence for calculations in general)
• The current for unpolarized calculations was missing the spin-degeneracy factor 2.
• Minor bugs:
  • improved error messages for unit cell mistakes in the two-probe setup
  • improved nlprint function for molecular spectra
  • fixed a visualization bug which sometimes gave wrong lighting
  • improved GPAW support
  • used-defined band structures are now plotted with correct labels, even if the labels are longer than 1 character

Known issues

• Multigrid uses more memory on Windows 64 than on Linux (sometimes a lot more); there is not much we can do about it, as the problem lies in the library we use, and not in our code.

Some images to further whet your appetite!

Also, check out this movie of an MD simulation in ATK! That specific functionality is coming in version 11.2, but you can already today visualize a geometry optimization trajectors in the same way.