Read me

The codes in this repository are an additional materials for:
Unsupervised clustering for identifying spatial inhomogeneity on local electronic structures
Hideaki Iwasawa, Tetsuro Ueno, Takahiko Masui, Setsuko Tajima
npj Quantum Materials 7, 24 (2022).
doi.org/10.1038/s41535-021-00407-5
Correspondence should be addressed to H.I. (iwasawa.hideaki@qst.go.jp)

• Codes are available as Jupyter Notebook (*.ipynb).
• Brief instructions are given in below.

Contents in this repository

• Data : Two kinds of spatially-resolved ARPES mapping data
• Part1 : Data Loading and pre-processing
• Part2 : k-means clustering
  • (2-1) Application
  • (2-2) Evaluation
• Part3 : Fuzzy-c-means clustering
  • (3-1) Application
  • (3-2) Evaluation
• Part4 : Principal Component Analysis

Requried libraries

• Load data
  • nexusformat: https://pypi.org/project/nexusformat/
• Standard data handling and visualization
  • numpy: https://pypi.org/project/numpy/
  • matplotlib: https://pypi.org/project/matplotlib/
• K-means clustering
  • scikit-learn: https://pypi.org/project/scikit-learn/
  • gap_statistic: https://pypi.org/project/gap-stat/
• Fuzzy-c-means clustering
  • skfuzzy: https://pypi.org/project/scikit-fuzzy/

Instructions

• Check Input and Output Path
  Default settings
  
  • Input path is placed in the same directry as the code files (jupyter notebooks).
  • Output file will be stored in an "out" folder, which will be creacted in the same directry as the code files.
• Always run Part1 first because Part2~4 require pre-processed dataset.
  
• The below figure shows the flow of clustering analysis with typical time required for executing each analysis.
  (analysis time will depend on you machine environment and parameter settings)
  

License

Copyright (c) 2021 Hideaki Iwasawa
This "arpes-clustering" respository is released under the MIT license.