One of the next great advances in materials research will be the computational, first-principles design of materials with desired target properties. The abundance of relatively inexpensive computer power, and the tremendous progress in developing robust and efficient numerical algorithms, coupled to the unparalleled interest in nanomaterials, has set the stage for materials' development through computational modeling. Microscopic, nanoscale properties can be unexpected, but are directly related to atomic arrangements, making computational modeling a natural tool to understand, predict, and design the properties of novel materials. To achieve this goal, scientists need a combination of accurate electronic-structure approaches, high-throughput data mining, and multiple-scales modeling in a computing intensive environment. With this award from the Instrumentation for Materials Research program, Scientists at MIT will acquire a 64-node (128 CPU, 192GB DDRAM, 5.8 TB disk, 783 GFlops peak performance) computational cluster which will provide a dedicated resource to perform on-demand materials' simulations for technologically-relevant problem sizes and complexities, and to be used as a computational laboratory for just-in-time realistic modeling in undergraduate and graduate computational classes. The cluster will revolutionize current capabilities at MIT, and would allow the study and characterization of nanostructures, biological molecules, and complex materials with sizes and accuracies that are not currently available. A 40-people strong computational research team will benefit from this award. The Department has redesigned the educational experience of undergraduate and graduate students in 2003. A fully revised undergraduate curriculum in Materials Science that teaches the fundamentals of quantum mechanics and of simulation and modeling has been added as core classes in the sophomore year. The new computer cluster would allow students a seamless transition between theoretical lectures and hands-on computational experiments, allowing a real-time exploration of materials using supercomputing electronic structure approaches.
One of the next great advances in materials research will be the computational, first-principles design of materials with desired target properties. The abundance of relatively inexpensive computer power, and the tremendous progress in developing robust and efficient numerical algorithms, coupled to the unparalleled interest in nanomaterials, has set the stage for materials' development through computational modeling. Microscopic, nanoscale properties can be unexpected, but are directly related to atomic arrangements, making computational modeling a natural tool to understand, predict, and design the properties of novel materials. To achieve this goal, scientists need a combination of accurate electronic-structure approaches, high-throughput data mining, and multiple-scales modeling in a computing intensive environment. With this award from the Instrumentation for Materials Research program, Scientists at MIT will acquire a 64-node computational cluster which will provide a dedicated resource to perform on-demand materials' simulations for technologically-relevant problem sizes and complexities, and to be used as a computational laboratory for just-in-time realistic modeling in undergraduate and graduate computational classes. The cluster will revolutionize current capabilities at MIT, and would allow the study and characterization of nanostructures, biological molecules, and complex materials with sizes and accuracies that are not currently available. A 40-people strong computational research team will benefit from this award. The Department has redesigned the educational experience of undergraduate and graduate students in 2003. A fully revised undergraduate curriculum in Materials Science that teaches the fundamentals of quantum mechanics and of simulation and modeling has been added as core classes in the sophomore year. The new computer cluster would allow students a seamless transition between theoretical lectures and hands-on computational experiments, allowing a real-time exploration of materials using supercomputing electronic structure approaches.
