Release notes

1.1.0

• Pinned numpy < 2

• Fixed an issue with sampling concentrations.

1.0.6

• Now requires the c++14 compiler flag.

• Fixed updating current energy for metadynamics. See !590.

1.0.5

• Added supercell_which_contains_sphere() and cell_wall_distances().

• SGCObserver now only tracks the singlet averages if observe_singlets=True. The default observer which is created by the SGCMonteCarlo can be controlled by the SGCMonteCarlo observe_singlets keyword.

• reset_eci in get_thermodynamic_quantities() now defaults to False. This keyword is likely to be removed in the future, see #321.

1.0.4

• Added CleaseCacheCalculator, as a primitive cache calculator object with no cache validation.

• Performance improvements to updating correlation functions.

• Performance improvements to calculating the translation matrix, so the first calculation of the clusters should be faster.

• Performance improvements to the LowestEnergyStructure. Correlation functions are also no longer tracked by default, but can be enabled with the track_cf key.

• The default SGC observer in SGCMonteCarlo should now be reset automatically upon changing the temperature.

1.0.3

• Getting thermodynamic quantities in the SGC MC now also retrieves averages from observers.

• Added interactive option to plot_eci()

• Added get_cluster_corresponding_to_cf_name().

• Minor performance improvements to SGC MC.

• Added set_normalization() for adjusting what elements to normalize by. Default is to normalize by everything.

1.0.2

• insert_structure() returns both the initial and final ID if both an initial and final structure was inserted.

• Fixes a bug with writing the Clease calculator to a DB row.

1.0.1

• Added the ignore_sizes keyword to plot_eci()

• Changing the maximum cluster diameter will now clear any cached clusters, and requires a new build.

• Calling observers in canonical MC can now be disabled with the call_observers keyword for performing burn-in, without executing observers.

1.0.0

• 21 June 2022 - CLEASE is no longer considered beta.

• Evaluate can now properly support fitting with custom LinearRegression schemes, even if they don’t support alpha cross-validation.

• Evaluate now required explicit calls to fit(). Calls to get_eci() and get_eci_dict() can no longer implicitly do fitting. This un-does a change introduced in version 0.11.6.

• Added the current_accept_rate property, and export the current accept rate in the thermodynamic quantities dictionary under the accept_rate key.

• Removed a series of deprecated things:

  • Removed the clease.concentration module.

  • Removed the clease.new_struct module.

  • Removed old regression imports. Regression classes must now be imported from the clease.regression module.

  • Removed the clease.structure_generator module.

  • Removed the max_cluster_size settings argument.

0.11.6

• Some small performance optimizations.

• Added a warn_on_skip parameter to the insert_structure() method.

• Evaluate should now correctly remember if it doesn’t need to re-fit the ECI’s (see the new fit() and fit_required() methods).

• Introduced load_eci() for loading stored ECI values, which is convenient for subsequent plotting.

• Added get_attempt_freq() which allows for more flexible customization of the attempt frequencies.

• Added the interactive keyword to plot_fit().

• Added an experimental parallelization feature. See Parallelization.

• Added clease info to the CLI to display some information about the installation.

0.11.5

• Fixed a bug with interactive plotting and convex hulls.

• Added the max_sphere_dia_in_cell() for calculating sphere diameters within the given cell boundaries.

• Changing the temperature of the Montecarlo object will now reset the internal energy averagers. Also, BaseMC now requires a temperature, and the temperature property has been renamed temperature. The old T attribute name is still accessible for backwards compatibility.

  For more information, see #302.

• Added iter_reconfigure_db_entries().

0.11.4

• Fixed an issue where attach_calculator would incorrectly try to snap the atoms onto a grid.

• Typo in the axis labeling in plot_fit.

0.11.3

• MCStep and SystemChange instances are now savable to json via the jsonio module.

• Fixed a bug which prevented the primitive to have more than 255 atoms.

• The primitive cell is now always wrapped in the settings object.

• Changing db_name will check if the primitive exists in the new DB, and write it if it’s missing.

• size and supercell_factor are now stored and managed by the TemplateAtoms object.

• Made some adjustments to the compilation process.

• Removed the include_background_atoms setter in the settings object. This value must now be set explicitly in the constructor for consistency reasons. For more information, see #292.

• Fixes a bug with the ConcentrationObserver.

0.11.2

• Introduces a new TransMatrix dataclass for the translation matrix.

• Temporarily restricts ASE to <3.23, until we resolve issues with current ASE master.

• Montecarlo will no longer consider background indices in the default swap move generator, if background is ignored.

• Added a new irun() method, for iteratively running MC calculations.

• MC observers can now override observe_step() instead, which takes a MCStep object.

• Added a new MC observer: clease.montecarlo.observers.MoveObserver.

0.11.1

• Fixed a bug in the FixedIndices constraint class.

• Greatly improved speed of supercell generation - this mostly affects performance concerning large supercells.

• Improved performance of the trans matrix generation.

0.11.0

• Python 3.7+ is now required.

• Removed old deprecated functions and classes.

• Some performance improvements.

0.10.9

• Now caches the CF names if requesting every CF name. Chops off some of the computation time during a full reconfigure.

• Introduces a new FinalStructPropertyGetter, which can be used to get arbitrary properties stored as key-value pairs in the database. Use the prop keyword in the Evaluate class to use this feature.

• Added the check_db keyword to NewStructures

• Some minor optimizations

0.10.8

• Fixes an issue with the coefficients generated by the Lasso method.

• Fixes an issue with the interactive convex hull plot.

• No longer opens an extra unnecessary GUI window with interactive plots.

• Fixed a bug with the fingerprint grouping, where the relative tolerance would reduce the numerical sensitivity too much.

• Now uses the packaging package for managing version numbers and comparisons. Removes usage of the deprecated distutils version comparisons.

0.10.7

• Fixed view_clusters(), which broke in 0.10.6.

• Adds ensure_clusters_exist() and get_all_figures_as_atoms(). ensure_clusters_exist() can be used to ensure that the cluster_list and trans_matrix are constructed, but will not cause a reconstruction if they are cached.

• Fixed a deprecation warning of normalize=True in sklearn’s Lasso method.

• Added a benchmarking suite in the tests directory.

0.10.6

• Fixed a bug in the clease.convexhull.ConvexHull where multiple end-points wouldn’t always find the correct minimum energy structure for that end-point.

• Added MCEvaluator.

• The settings class should now be much faster to construct, since the construction of the translation matrix and cluster list is defered until requested.

• The built in GUI (based on Kivy) has been removed, in favor of the new Jupyter based clease-gui package.

• Deprecated the use of max_cluster_size for specifying clusters in ClusterExpansionSettings. Clusters should now be specified only though max_cluster_dia, where the size of the cluster is infered from the length of the list. The index 0 corresponds to 2-body clusters, index 1 to 3-body etc., i.e. max_cluster_dia = [5, 4, 3] would result in clusters of up to diameter 5 Å for 2-body clisters, 4 Å for 3-body and 3 Å for 4-body.

0.10.5

• Added clease.logging_utils.log_stream() and clease.logging_utils.log_stream_context() functions to simplify printing the CLEASE logs to a file. The global CLEASE logger can be retreived with clease.logging_utils.get_root_clease_logger().

0.10.4

• Fixed a bug with sorting the figures in ClusterList would cause a de-synchronization of the indices, and crashing any further usage.

• Now supports clusters of arbitrary size. Used to be limited to 2-, 3- and 4-body clusters.

0.10.3

• Added convex hull plot, clease.plot_post_process.plot_convex_hull()

• Fixed a bug in clease.structgen.NewStructures.generate_gs_structures() where passing multiple atoms objects was failing

• Structure generation of pure elements should now be using the smallest possible cell.

• Alpha and CV values are now stored in the clease.evaluate.Evaluate class after running the clease.evaluate.Evaluate.alpha_CV() function.

• Added doc as an extras_require in setup.py.

• Other minor bugfixes

0.10.2

• clease.montecarlo.SSTEBarrier renamed to clease.montecarlo.BEPBarrier

• Added release notes

• Added the clease.jsonio module. This has been applied to the clease.settings.ClusterExpansionSettings, clease.settings.Concentration and clease.basis_function.BasisFunction classes, providing them with save() and load() functions.

• Tests now automatically run in the pytest temporary directory.

• Moved new_struct and structure_generator into the structgen module. These should now be imported from here, instead.

• Fixed a bug, where the current step counter in the clease.montecarlo.Montecarlo class would not be reset upon starting a new run.