BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20141116T194500Z
DTEND:20141116T200000Z
LOCATION:288-89
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: The ability of a nanomaterial to conduct charge is essential for many nanodevice applications. Numerous studies have shown that disorder, including defects and phononic or plasmonic effects, can disrupt or block electric current in nanomaterials. An accurate theoretical account of both electron-phonon and electron-plasmon coupling in carbon nanotube- and nanoring-based structures is key in order to properly predict the performance of these new nanodevices. =0AThe parallel sparse matrix library PETSc provides tools to optimally divide memory and computational tasks required in inverting the Hamiltonian matrix across multiple compute cores or nodes. The results at different energy levels are integrated via a second layer of parallelism to obtain integrated observables. Because the inversion at each energy step dominates the run time and memory use of the code, it is important that the code scales well to realistic system sizes. Several different direct and iterative linear solvers and libraries were compared for their scalability and efficiency, including the Intel MKL shared-memory, direct dense solver, the MUltifrontal Massively Parallel direct Solver (MUMPS), and iterative sparse solvers bundled with PETSc. The NSF XSEDE resource ‘Stampede’ at the Texas Advanced Computing Center as well as regional resources in Florida (SSERCA) are used for development and benchmarking, while DOE OLCF computational resources ‘Titan’ and 'NICS/Beacon' are used for physics production runs.
SUMMARY:QRing – A Scalable Parallel Software Tool for Quantum Transport Simulations in Carbon Nanoring Devices Based on NEGF Formalism and a Parallel C++ / MPI / PETSc Algorithm
PRIORITY:3
END:VEVENT
END:VCALENDAR