V2.0 released

We are happy to announce that icet 2.0 (Northern Lights) has been released, which contains new features, performance improvements as well as additions to the documentation.

New features

  • We implemented a version of the constituent strain formalism. This enables one to separate the long-ranged strain contributions from the short-ranged “chemical” interactions. This can be used, e.g., to describe coherent phase equilibria, see arxiv:2110.07883.
  • We have moved the fitting functionality (icet.fitting), which was shared with hiphive, to a separate Python package called trainstation. This makes it easier for us to maintain the fitting module and share it among several of our packages.
  • One can now manually merge orbits, which allows one to customize cluster spaces. This is especially useful in low-symmetry systems with very large cluster spaces, see doi: 10.1021/acs.jpcc.1c00575.
  • We added a long-range observer that computes structure factors. The observer can be used in Monte Carlo simulations or applied directly to structures of interest.

Performance improvements

  • Improved performance for Wang-Landau simulations.
  • Improved performance of Monte Carlo simulations. The simulations now scale much better than before.
1 Like

Congrats on the release! I look forward to trying out the constituent strain formalism and am very excited for the performance improvements!

I just realized that I missed the v1.4 release. Is the change log for that release posted somewhere too?

That’s great to hear. You can find the release notes on gitlab under “tags”: Tags · materials-modeling / icet · GitLab

Specifically the release notes for 2.0 and 1.4 can be found at

1 Like

Thanks for responding quickly, I think I may have been affected by the single atom structure mapping bug. I’ll have to update and try mapping again.