Abstract. Exploiting data sparsity in dense matrices is an algorithmic bridge between architectures that are increasingly memory-austere on a per-core basis and extreme-scale applications. The Hierarchical matrix Computations on Manycore Architectures (HiCMA) library tackles this challenging problem by achieving significant reductions in time to solution and memory footprint, while preserving a specified accuracy requirement of the application. HiCMA provides a high-performance implementation on distributed-memory systems of one of the most widely used matrix factorization in large-scale scientific applications, i.e., the Cholesky factorization. It employs the tile low-rank data format to compress the dense data-sparse off-diagonal tiles of the matrix. It then decomposes the matrix computations into interdependent tasks and relies on the dynamic runtime system StarPU for asynchronous out-of-order scheduling, while allowing high user-productivity. Performance comparisons and memory footprint on matrix dimensions up to eleven million show a performance gain and memory saving of more than an order of magnitude for both metrics on thousands of cores, against state-of-the-art open-source and vendor optimized numerical libraries. This represents an important milestone in enabling large-scale matrix computations toward solving big data problems in geospatial statistics for climate/weather forecasting applications.

In this work, the particular and combined influence of nonparabolicity and phonon scattering on the device characteristics of a triple-gate silicon nanowire is investigated. In addition, different approximations of the retarded self-energy for electron-phonon scattering are analyzed in terms of the electrostatics, current and computational cost.

http://www.gianlucafiori.org/articles/SSE09.pd

We study the properties of ultracold Fermi gases in a three-dimensional optical lattice when crossing a Feshbach resonance. By using a zero-temperature formalism, we show that three-body processes are enhanced in a lattice system in comparison to the continuum case. This poses one possible explanation for the short molecule lifetimes found when decreasing the magnetic field across a Feshbach resonance. Effects of finite temperatures on the molecule formation rates are also discussed by computing the fraction of double-occupied sites. Our results show that current experiments are performed at temperatures considerably higher than expected: lower temperatures are required for fermionic systems to be used to simulate quantum Hamiltonians. In addition, by relating the double occupancy of the lattice to the temperature, we provide a means for thermometry in fermionic lattice systems, previously not accessible experimentally. The effects of ramping a filled lowest band across a Feshbach resonance when increasing the magnetic field are also discussed: fermions are lifted into higher bands due to entanglement of Bloch states, in good agreement with recent experiments.Comment: 9 pages, 7 figure

We present an experimental study of the effective dissipation relevant in the thermally activated supercurrent decay of extremely underdamped Josephson junctions. Data referring to the supercurrent decay of Nb/AlO x /Nb Josephson junctions are compared with the Kramers theory. Our measurements allow us to obtain the ''effective'' resistance to be used in the resistively shunted junction model that results to be the subgap resistance due to the presence of thermally activated quasiparticles. The extremely low dissipation level obtained at low temperatures renders our result quite interesting in view of experiments in the quantum limit.