Dataset: Real-space visualization of a defect-mediated charge density wave transition

Authors: James L Hart¹, Haining Pan ², Saif Siddique ¹, Noah Schnitzer ¹, Krishnanand Mallayya ², Shiyu Xu¹, Lena F Kourkoutis^3,4, Eun-ah Kim^2,5, Judy J Cha ¹

Author affiliations:
¹:Department of Materials Science and Engineering, Cornell University, United States
²:Department of Physics, Cornell University, United States
³:School of Applied and Engineering Physics, Cornell University, United States
⁴:Kavli Institute at Cornell for Nanoscale Science, Cornell University, United States
⁵:Department of Physics, Ewha Womans University, South Korea

We are providing the raw data and the machine learning model used to produce the results shown in Fig. 4 and Fig. S1. Here is a list of the functionality of each file (all individual files are included in ‘code.zip’): 

• q_Txy.py: Detects peaks (kx, ky) for each point at (T, x, y). T = temperature 
• q_stat.py: Returns the statistics of the peaks 
• GMM.py: Applies a Gaussian mixture model to cluster the peaks 
• 4DSTEM.ipynb: An interactive Jupyter notebook file that generates the figure in the paper from the raw data.  

Given that data analysis from the raw data can be time-consuming, we have also provided the post-analyzed files: 

• q_Txy_count_pts_outlier_sample_2_auto_bragg.pickle: This file, output from q_Txy.py, contains the peaks in the momentum space for each (T, x, y). 
• count_pts_outlier_auto.pickle: This file, output from GMM.py, contains the cluster information, such as which point of (T, x, y) belongs to which cluster given the number of clusters k. 

 Lastly, the raw data is included as data.hdf5.