icet

3.2 verified Sat May 09 auth: no python

icet is a Python library for constructing cluster expansions, providing a pythonic approach to modeling alloy thermodynamics and ordering. Current version is 3.2, with a new release roughly every few months. It requires Python >=3.9 and is actively maintained.

pip install icet

Common errors

error ImportError: cannot import name 'ClusterExpansion' from 'icet' ↓

cause In icet 3.x, the class was renamed to 'ClusterExpansionSettings'.

fix

Use 'from icet import ClusterExpansionSettings'

error ValueError: Number of clusters does not match number of structures ↓

cause Cluster space and structure list are inconsistent. Check if structures have the same number of atoms and are compatible with the primitive.

fix

Ensure all structures are derived from the same primitive and have consistent atom numbers.

Warnings

breaking In icet 3.x, the 'ClusterExpansion' class is renamed to 'ClusterExpansionSettings'. Old imports will fail. ↓

fix Replace 'from icet import ClusterExpansion' with 'from icet import ClusterExpansionSettings'.

deprecated The 'ClusterSpace' constructor keyword 'cutoffs' is deprecated in favor of 'cutoff' (singular) in icet 3.2. ↓

fix Use 'cutoff' instead of 'cutoffs' when creating ClusterSpace.

gotcha Cluster expansion fitting requires normalized training data. Forgetting to normalize energies per atom can lead to incorrect results. ↓

fix Ensure training energies are divided by the number of atoms in each structure.

Imports

ClusterExpansionSettings
wrong
```
from icet.core import ClusterExpansionSettings
```
correct
```
from icet import ClusterExpansionSettings
```
In icet 3.x, the recommended import is from the top-level 'icet' package, not the 'core' submodule.
ClusterSpace
wrong
```
from icet.cluster_space import ClusterSpace
```
correct
```
from icet import ClusterSpace
```
Importing from submodules is not guaranteed to work in future versions.

Quickstart

Basic usage example: create a cluster space for Cu-Au and initialize cluster expansion settings.

from ase.build import bulk
from icet import ClusterExpansionSettings, ClusterSpace

# Create a primitive structure
prim = bulk('Cu', 'fcc', a=3.6)

# Define cluster space with pair interactions up to cutoff 5.0
cs = ClusterSpace(prim, cutoffs=[5.0], chemical_symbols=['Cu', 'Au'])

# Create settings object
settings = ClusterExpansionSettings(cs)

print(settings)