search

ziyuanrao11 / Machine-learning-enabled-high-entropy-alloy-discovery

This is the code for the paper 'Machine learning-enabled high-entropy alloy discovery'
57 stars 11 forks source link

Variable specification and code repetition #3

Open chuchuanp opened 12 months ago

chuchuanp commented 12 months ago

in file https://github.com/ziyuanrao11/Machine-learning-enabled-high-entropy-alloy-discovery/blob/main/TERM/PCA%20and%20AffinityPropagation.py, line 31 plt.scatter(c_trans[:,0],c_trans[:,1], s = 10, c='black', marker='^',zorder=2) i confused what variable 'c_trans' stands for, and the file https://github.com/ziyuanrao11/Machine-learning-enabled-high-entropy-alloy-discovery/blob/main/TERM/Ensemble%20model_stage_2.py seems to be the same as https://github.com/ziyuanrao11/Machine-learning-enabled-high-entropy-alloy-discovery/blob/main/TERM/Ultility%20functions.py. could you please update the file? thank you!

chuchuanp commented 11 months ago

The reply from code author is below: In the ensemble model stage1, we use input features with 17 dimensions (composition+atomic features). In the stage 2, we used input with 20 dimensions (composition+atomic features+calphad+DFT). The structure of the two ensemble model are the same, only the inputs are different. If you use the file after stage 1, WAE_x is the composition from the WAE+MCMC sampling. If you use this file after stage 2, WAE_x is the composition from the stage1. In the file (HEA-COGS/MCMC Sampling.py in the github), line 60, you can generate the dataframe, WAE_comps, which is the WAE_x composition for stage 1. The c_trans is the latent space (dimension 1 and dimension 2 you get from the WAE).