A team of Harvard University researchers has been allocated time on the Trestles supercomputer at the San Diego Supercomputer Center (SDSC) at the University of California, San Diego, to perform computational calculations with the goal of creating the next generation of organic solar cells as an inexpensive and efficient source of energy.
The allocation is a key part of the team's efforts to conduct larger, data-intensive computations related to its Clean Energy Project (CEP), which combines the group's computational chemistry expertise with the large, distributed computing power of IBM's World Community Grid (WCG).
Specifically, the CEP combines theory, computation, experiments, and grid computing by harvesting idle computing time from donors around the world using the WCG to perform ab initio computational quantum chemistry calculations on a large number of candidate molecules that could potentially form the next generation of solar cells. The complete CEP database will soon be made publicly available to the scientific community.
Despite the success of the CEP – more than 6 million molecular motifs of potential interest have been characterised and thousands of new molecules are being added to its database every day – the program's research of larger, more complex datasets has been limited because the majority of WCG compute resources consist of home or office PCs and are on public networks, which create issues such as hardware heterogeneity, data transfer speeds, and tailoring of computing times according to the needs and interests of donors.
Enter SDSC's Trestles system, a resource for modest-scale researchers who need to be as computationally productive as possible. Alán Aspuru-Guzik, an associate professor with Harvard University's Department of Chemistry and Chemical Biology and head of the CEP initiative, was allocated more than 1.36 million service units, or core-hours, on Trestles through the National Science Foundation's (NSF) Extreme Science and Engineering Discovery Environment, or XSEDE program, to perform these high-volume computations.
''Trestles allowed us to perform calculations on larger molecular systems that are difficult to calculate elsewhere,'' said Aspuru-Guzik. ''We were able to perform more complex calculations of systems with more than 300 electrons, which are currently impossible to run on smaller systems. As well as large molecules, the computing power of Trestles let us gather most interesting and promising candidate molecules at a higher level of theory, resulting in a much improved molecular characterization of those systems of interest.''