What problem does this paper attempt to address?

### Problems the paper attempts to solve This paper aims to solve an urgent problem in the field of materials science: defining a metric for synthetic feasibility and predicting which new materials can be synthesized experimentally. Specifically, using machine - learning techniques and based on the existing inorganic materials database, the paper proposes a method that can predict the synthetic feasibility of materials. ### Background and challenges In contemporary materials science, being able to predict whether a material can be synthesized experimentally is an important task. This is related not only to the structure, phase state, and composition of the material, but also involves the bridge between theoretical research and experimental synthesis. Traditionally, researchers use the trial - and - error method to find material candidates, which usually combines empirical knowledge and theoretical calculations. However, due to the large differences in human experience and material selection, the discovery and synthesis process of new materials is often slow and unpredictable. Therefore, an accurate method is needed to determine the synthetic feasibility of materials to improve the efficiency and productivity of new material discovery. ### Solutions To meet this challenge, the authors use the Inorganic Crystal Structure Database (ICSD) and the Materials Project Database (MP), and propose a new crystal structure representation method - Fourier - Transform Crystal Properties (FTCP) representation, and use a deep - learning model to predict the synthetic feasibility of materials. The specific steps are as follows: 1. **Data preparation**: Obtain training and validation data sets from the ICSD and MP databases. 2. **Crystal structure representation**: Convert the crystal structure into a computer - readable form, including real - space features and reciprocal - space features. 3. **Model construction**: Build a deep - learning model that processes the crystal structure representation and outputs a Synthetic Feasibility Score (SC). 4. **Model training and validation**: Use data before 2015 for training and test on data after 2015 to evaluate the performance of the model. ### Main contributions - **High - precision prediction**: The model achieves an accuracy rate of 82.6% and a recall rate of 80.6% in predicting the synthetic feasibility of ternary crystal materials. - **Time - dependent validation**: By training and testing data from different time periods, the effectiveness of the model in predicting the synthetic feasibility of newly added materials is verified. - **Prediction of unexplored materials**: Provide a list of 100 new materials predicted to have high synthetic potential that have not been experimentally verified in the MP database. ### Conclusion By combining crystal structure representation and deep - learning techniques, this paper successfully establishes a model that can predict the synthetic feasibility of materials efficiently and accurately. This method not only helps to accelerate the discovery of new materials, but also provides strong support for future material screening and discovery.

Predicting Synthesizability using Machine Learning on Databases of Existing Inorganic Materials

Is Large Language Model All You Need to Predict the Synthesizability and Precursors of Crystal Structures?

Network analysis of synthesizable materials discovery

Explainable Synthesizability Prediction of Inorganic Crystal Polymorphs using Large Language Models

Bridging Chemical Knowledge and Machine Learning for Performance Prediction of Organic Synthesis.

Machine learning-guided synthesis of advanced inorganic materials

Predicting Miscibility in Binary Compounds: A Machine Learning and Genetic Algorithm Study

Inorganic synthesis recommendation by machine learning materials similarity from scientific literature

Explainable Synthesizability Prediction of Inorganic Crystal Structures using Large Language Models

Large-scale machine-learning-assisted exploration of the whole materials space

Machine Learning-Based Prediction of Crystal Systems and Space Groups from Inorganic Materials Compositions

What Information is Necessary and Sufficient to Predict Materials Properties using Machine Learning?

Inorganic synthesis-structure maps in zeolites with machine learning and crystallographic distances

Large Language Models for Inorganic Synthesis Predictions

Topological representations of crystalline compounds for the machine-learning prediction of materials properties

Challenges in high-throughput inorganic material prediction and autonomous synthesis

Materials discovery through machine learning formation energy

Universal machine learning aided synthesis approach of two-dimensional perovskites in a typical laboratory

Towards machine learning for microscopic mechanisms: a formula search for crystal structure stability based on atomic properties

Enhancing crystal structure prediction by combining computational and experimental data via graph networks

Materials Synthesis Insights from Scientific Literature via Text Extraction and Machine Learning