Event
MSE Seminar - Prof. Estela Blaisten-Barojas, George Mason University
Friday, November 7, 2008
1:00 p.m.
Rm. 2108, Chemical and Nuclear Engineering Bldg.
Annette Mateus
301 405 5207
amateus@umd.edu
"Machine Learning Classification of Zeolite Crystals and Computer Experiments of Polypyrrole"
Extracting information from data gathered in databases is a paradigm adopted only recently by the science community. On the other hand, scientists have generated data from simulations at the quantum or classical level for a long time. This talk will be divided into two parts, the first addressing the new paradigm as applied to zeolite crystals and the second describing a more traditional approach used in the study of polypyrrole.
We have developed the Zeolite Structure Predictor, which is a machine learning model trained and tested with data contained in the Inorganic Crystal Structure Database. This model uses novel topological descriptors that allow for classification of the crystals into zeolite-framework types with an almost perfect success rate. The model is an efficient tool for identifying the framework type of newly synthesized or computationally predicted zeolite crystals. The second part of the talk will describe the energetics, electron charge distribution, vibrational spectra of n-pyrrole (n=2, 4, 6, 8, 12, 15, 24) from all-electron density functional theory calculations. Comparison between the reduced and oxidized phases shows clearly the charge transfer mechanism that occurs during redox. A novel model potential was developed based on the quantum calculations, which allow for molecular dynamics simulations of polymer systems at various densities. These macroscopic systems display a preferential molecular ordering under certain thermodynamic conditions.
