解析者の経験によらないEXAFS解析への挑戦— Constrained Thorough Search法の開発 —

抄録

EXAFS is powerful to obtain the structure information around the X-ray absorbing atom at the atomic size level. EXAFS is usually analyzed by the curve fitting method, one of the least square methods to optimize structure parameters. However, the curve fitting method demands the experience of scientists when they analyze the EXAFS of materials with complex structures. In particular, the following two items make the analysis difficult especially for beginners ; 1. the dependence on the initial selection of the fitting parameters, 2. fitting errors caused by their correlation. We developed a new method, called as Constrained Thorough Search (CTS) method to solve these problems. The CTS method searches the whole parameter space under constrained conditions to extract the structure candidates. We describe the details of the CTS method with their results. We also applied the machine learning method to CTS results which allows us to conduct the analysis more objectively.