Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Milind Malshe Applicant's institution: ANL Applicant's division: CSE Project Name: MD_Potential_Oxide Project title: Synthesis Science of Functional Layered Complex Oxides Associated funding: Strategic Initiative on Materials for Energy (LDRD-2012-188-R1) Other Systems: NERSC: ~ 400000 core-hours Science: The principal objective of this project is development of new, more flexible and more accurate many-body potentials for the description of interatomic interactions in various inhomogeneous systems and use of these potentials in large scale molecular dynamics (MD) and Monte Carlo (MC) simulations of the synthesis and properties of new functional materials such, e.g., as layered complex oxides. The results of the proposed studies will aid the ongoing collaborative experimental effort at MSD, APS, CNM and CSE in defining synthesis schedules and conditions – types of elements, temperature, pressure, etc. - that will result in materials with the desired properties and functionalities. We will also explore the properties of new materials as a function of the chemical identity and number/thickness of layers, strain, and nature and concentration of defects, including the mechanisms of formation of defects (e.g., as induced by temperature or pressure). Project description: The initial effort will focus on the development of new, more robust, more flexible, and more accurate potentials for mimicking the interatomic interactions in layered complex oxide materials. The systems of interest include SrTiO3, BaTiO3, PbTiO3 (these are currently under investigation by the experimental component of this project), and others. The effort will be pursued in two directions. The first aims at adjustment (fitting of the parameters) and generalization (combining variable charges with variable polarizabilities) of the existing potentials to/for the systems of interest. The second is a principally new approach that is based on the neural networks (NN) methodology, evolutionary (specifically, genetic algorithms (GA)) techniques, and the combination of the two. A particularly beneficial aspect of the NN approach is that no a priori assumption about the functional form of the potential is made, which results in a much higher degree of flexi bility and robustness. In developing the NN and GA based potentials we will use general algorithmic tools implemented in MATLAB (which is installed on Fusion; the NN and GA toolboxes will have to be installed). The data for fitting the new parameter values (in the first approach) and training the NNs (in the second approach) will be generated using state-of-the art density functional theory (DFT). An additional algorithmic advantage of the NN and GA methodologies is that they are inherently amenable to high degree of parallelization. Work on this aspect and the related methodological developments will be an important part of the planned activities. The MD runs will be performed using the parallelized version of the LAMMPS and DL_POLY packages. The high scalability of these packages with the system size (number of atoms) is well documented. The number of atoms in our planned simulations will vary from a few tens of thousands to hundreds of thousands and possibly a few milli ons. The planned DFT computations will be carried out with the VASP, Crystal (we have recently acquired licenses for the latest versions of both), Siesta (installed recently on Fusion) and Quantum Espresso (a free electronic structure code that we will install on Fusion) packages. All of them exhibit high parallelization scalability for the type of geometry, energy and electronic structure computations needed for this project. The number of project members is 4. Project URL: http://blogs.anl.gov/major_initiatives/materials-for-energy/ Requested allocation: 800000 Q1: 150000 Q2: 150000 Q3: 250000 Q4: 250000 Justification: The allocation request is based on MD test runs of 2 ns for a thin film of SrTiO3 consisting of 20,000 atoms performed with the DL_POLY package on Carver at NERSC: Cores Wall time (s) 8 38880 16 31629 32 24843 64 16282 128 8698 256 3895 (A better parallelization scaling is achieved for this system, as evaluated over runs of 2 ns in duration, with a larger number of cores.) The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System