The key to energy independence from petroleum, coal and other fossil fuels, could be tiny materials called nanostructures. At approximately 100,000 times finer than human hair, these structures maybe microscopic individually, but in groups of thousands, they could revolutionize solar-cell design by providing a cost-efficient resource for harvesting solar-energy. To theoretically understand and simulate the energy harnessing potential of nanostructures, a team of researchers in the Berkeley Lab’s Computational Research Division (CRD) developed the Linear Scaling Three Dimensional Fragment (LS3DF) method. The computer algorithms in this method use a “divide-and-conquer” technique to efficiently gain insights into how nanostructures function in systems with 10,000 or more atoms. “By incorporating the correct chemical formulas into efficient computer programs, scientists can learn a lot about the structures and properties of molecules and solids.… I like to think of computers as chemistry’s ‘third leg.’ In most cases, computer simulations complement information obtained by chemical experiments, but in some cases it can predict unobserved phenomena,” says Dr. Lin-Wang Wang, a CRD computational material scientist and leader of the LS3DF project. The developers of LS3DF are finalists in the Association for Computing Machinery’s (ACM) Gordon Bell Prize Competition, which recognizes outstanding achievement in high-performance computing applications. The winners will be announced on November 20, 2008 at the SC08 Conference in Austin, Tex. According to Wang, traditional methods for calculating the energy potential of nanostructure systems containing 10,000 or more atoms can be very time consuming and resource intensive. Because these techniques calculate the entire structure as a whole system, the compute time, disk space and memory required to determine the energy potential of these structures grows to the third power of the system’s size. That means calculating a 1000-atom system will be a thousand times more expensive than calculating a 100-atom system. He notes that LS3DF offers a more efficient way for calculating energy potential because it is based on the observation that the total energy of a large nanostructure system can be broken down into small pieces, and each piece can be calculated separately. Wang refers to this technique as “divide-and-conquer.” The total energy of the large system has two components: electrostatic energy and quantum mechanical energy. To determine the structure’s total quantum mechanical energy, the LS3DF method breaks the entire structure into small fragments, applies its algorithm to each individual piece, and then combines the results of the pieces to get a total for the whole system. Scientists say that under the traditional density functional theory methods, the quantum mechanical energy calculation typically requires the most compute time and resources. By breaking up the big problem into small pieces, LS3DF can solve it a lot more quickly and efficiently, making the computational cost proportional to the total number of the atoms in the system. Meanwhile, the electrostatic energy of large-scale nanostructure systems is not as resource intensive to solve. Scientists calculate this classical energy by looking at the whole system, which may contain tens of thousands of atoms. This problem is solved separately from the quantum mechanic energy. In the end, both energy results are combined to get the structure’s total energy potential. When team members tested the LS3DF method on supercomputers at the Department of Energy’s (DOE) National Energy Scientific Research Center (NERSC) in Oakland, Calif, National Center for Computational Sciences (NCCS) at Oak Ridge National Laboratory in Oak Ridge, Tenn., and Argonne Leadership Computing Facility in Argonne, Ill., they found that the LS3DF method can work hundreds to thousands of times faster than traditional density functional theory calculations for systems with tens of thousands of atoms, and yielded essentially the same results. “The core of LS3DF is a novel patching scheme that cancels out the artificial boundary effects caused by dividing the system into smaller fragments,” says Wang. “This cancellation is what gets us the same results as the traditional method.” Because LS3DF scales almost perfectly with the number of compute cores, it is the first electronic structure code that runs efficiently on computer systems with tens to hundreds of thousands of cores. On 17,280 cores of the dual-core Cray XT4 (Franklin) at NERSC, LS3DF achieved 32 Tflop/s or 32% of the peak floating-point performance of the machine. On 30,720 cores of the quad-core Cray XT4 (Jaguar) at NCCS, LS3DF reached 60 Tflop/s or 23% of the theoretical peak. In a later run on the IBM BlueGene/P system (Intrepid) at Argonne, the code achieved 107.5 Tflop/s on 131,072 cores, or 24.2% of peak. Energy Independence from Fossil Fuels Scientists agree that a fundamental understanding of nanostructure behaviors and properties could provide a solution for curbing our dependence on petroleum, coal, and other fossil fuels. According to Wang, nanostructure systems are cheaper to produce than the crystal thin films used in current solar cell designs, and offer the same material purity. In addition, nanostructures are extremely versatile. They can act as electrodes to carry electric currents, or active materials that absorb sunlight and convert it to electricity. One type of nanostructure, called quantum dots, actually changes color with size. Scientists say this color, or band gap, affects the type of light that the structure absorbs, which will be very useful for designing solar-cells. “We still don't quite understand how the electron moves around in a nanostructure, and how such properties depend on the size, geometry, composition, and surface passivations … Understanding such dependence will allow us to design nanostructures for desired applications, and LS3DF can help us to understand and predict these properties with computers,” says Wang. Other authors on the Gorden Bell paper include the Berkeley Lab’s David H. Bailey, Zhao, Byounghak, Zhengji Lee, Juan Meza, Hongzhang Shan, and Erich Strohmaier.