Coarse-graining and data mining approaches to the prediction of structures and their dynamics
Author(s)
Curtarolo, Stefano, 1969-
DownloadFull printable version (2.559Mb)
Other Contributors
Massachusetts Institute of Technology. Dept. of Materials Science and Engineering.
Advisor
Gerbrand Ceder.
Terms of use
Metadata
Show full item recordAbstract
Predicting macroscopic properties of materials starting from an atomistic or electronic level description can be a formidable task due to the many orders of magnitude in length and time scales that need to be spanned. A characteristic of successful approaches to this problem is the systematic coarse-graining of less relevant degrees of freedom in order to obtain Hamiltonians that span larger length and time scale. Attempts to do this in the static regime (i.e. zero temperature) have already been developed, as well as thermodynamical models where all the internal degrees of freedom are removed. In this thesis, we present an approach that leads to a dynamics for thermodynamic-coarse-grained models. This allows us to obtain temperature-dependent and transport properties. The renormalization group theory is used to create new local potential models between nodes, within the approximation of local thermodynamical equilibrium. Assuming that these potentials give an averaged description of node dynamics, we calculate thermal and mechanical properties. If this method can be sufficiently generalized it may form the basis of a Multiscale Molecular Dynamics method with time and spatial coarse-graining. In the second part of the thesis, we analyze the problem of crystal structure prediction, by using quantum calculations. (cont.) This is a fundamental problem in materials research and development, and it is typically addressed with highly accurate quantum mechanical computations on a small set of candidate structures, or with empirical rules that have been extracted from a large amount of experimental information, but have limited predictive power. In this thesis, we transfer the concept of heuristic rule extraction to a large library of ab-initio calculated information, and demonstrate that this can be developed into a tool for crystal structure prediction. In addition, we analyze the ab-initio results and prediction for a large number of transition-metal binary alloys.
Description
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 2003. Includes bibliographical references (p. 245-263). This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.
Date issued
2003Department
Massachusetts Institute of Technology. Department of Materials Science and EngineeringPublisher
Massachusetts Institute of Technology
Keywords
Materials Science and Engineering.