Probabilistic deep learning approach for targeted hybrid organic-inorganic perovskites

We develop a probabilistic machine learning model and use it to screen for new hybrid organic-inorganic perovskites (HOIPs) with targeted electronic band gap. The data set used this work is highly diverse, containing multiple atomic structures each of 192 chemically distinct HOIP formulas. Therefore, any property prediction on given formula must be associated an irreducible "uncertainty" that comes from its unknown details. As result, dozens formulas gap falling between 1.25 1.50 eV were identified validated against suitable first-principles computations. Through demonstration we show the deep approach robust, versatile, can properly quantify uncertainty. In conclusion, standpoint described herein could widely useful very common inevitable uncertainty which rooted at incompleteness information during experiments and/or