Fig. 5: Applying predicted formation energies to stability predictions in a sparse chemical space.
From: A critical examination of compound stability predictions from machine-learned formation energies
Re-training each model on all of MP minus 267 quaternary compounds in the Li–Mn–TM–O chemical space (TM ∈ {Ti, V, Cr, Fe, Co, Ni, Cu}) and obtaining ΔHd using the predicted ΔHf for each of the excluded compounds (ΔHd,pred) and comparing with stabilities available in MP, ΔHd,MP. FP = false positive, TP = true positive, TN = true negative, FN = false negative.
