A Scalable General Purpose System for Large-Scale Graph Processing

Graph analytics is an important and computationally demanding class of data analytics. It is essential to balance scalability, ease-of-use and high performance in large scale graph analytics. As such, it is necessary to hide the complexity of parallelism, data distribution and memory locality behind an abstract interface. The aim of this work is to build a scalable graph analytics framework that does not demand significant parallel programming experience based on NUMA-awareness.
The realization of such a system faces two key problems:
(i)~how to develop a scale-free parallel programming framework that scales efficiently across NUMA domains; (ii)~how to efficiently apply graph partitioning in order to create separate and largely independent work items that can be distributed among threads.

Joint Analysis of Multiple Algorithms and Performance Measures

There has been an increasing interest in the development of new methods using Pareto optimality to deal with multi-objective criteria (for example, accuracy and time complexity). Once one has developed an approach to a problem of interest, the problem is then how to compare it with the state of art. In machine learning, algorithms are typically evaluated by comparing their performance on different data sets by means of statistical tests. Standard tests used for this purpose are able to consider jointly neither performance measures nor multiple competitors at once. The aim of this paper is to resolve these issues by developing statistical procedures that are able to account for multiple competing measures at the same time and to compare multiple algorithms altogether. In particular, we develop two tests: a frequentist procedure based on the generalized likelihood-ratio test and a Bayesian procedure based on a multinomial-Dirichlet conjugate model. We further extend them by discovering conditional independences among measures to reduce the number of parameters of such models, as usually the number of studied cases is very reduced in such comparisons. Data from a comparison among general purpose classifiers is used to show a practical application of our tests.

Energy levels and radiative rates for transitions in Cr-like Co IV and Ni V

We report calculations of energy levels and radiative rates (A-values) for transitions in Cr-like Co IV and Ni V. The quasi-relativistic Hartree-Fock (QRHF) code is adopted for calculating the data although grasp (general-purpose relativistic atomic structure package) and flexible atomic code (fac) have also been employed for comparison purposes. No radiative rates are available in the literature to compare with our results, but our calculated energies are in close agreement with those compiled by NIST for a majority of the levels. However, there are discrepancies for a few levels of up to 3%. The A-values are listed for all significantly contributing E1, E2 and M1 transitions, and the corresponding lifetimes reported, although unfortunately no previous theoretical or experimental results exist to compare with our data.

Radiative rates for E1, E2, M1, and M2 transitions in S-like to F-like tungsten ions (W LIX to W LXVI)

Calculations of energy levels, radiative rates and lifetimes are reported for eight ions of tungsten, i.e. S-like (W LIX) to F-like (W LXVI). A large number of levels have been considered for each ion and extensive configuration interaction has been included among a range of configurations. For the calculations, the general-purpose relativistic atomic structure package (. grasp) has been adopted, and radiative rates (as well as oscillator strengths and line strengths) are listed for all E1, E2, M1, and M2 transitions of the ions. Comparisons have been made with earlier available experimental and theoretical energies, although these are limited to only a few levels for most ions. Therefore for additional accuracy assessments, particularly for energy levels, analogous calculations have been performed with the flexible atomic code (. fac).

Radiative rates for E1, E2, M1, and M2 transitions in Br-like ions with 43≤Z≤50

Energies and lifetimes are reported for the eight Br-like ions with 43≤Z≤50, namely Tc IX, Ru X, Rh XI, Pd XII, Ag XIII, Cd XIV, In XV, and Sn XVI. Results are listed for the lowest 375 levels, which mostly belong to the 4s²4p⁵, 4s²4p⁴4ℓ, 4s4p⁶,4s²4p⁴5ℓ, 4s²4p³4d², 4s4p⁵4ℓ, and 4s4p⁵5ℓ configurations. Extensive configuration interaction among 39 configurations (generating 3990 levels) has been considered and the general-purpose relativistic atomic structure package (grasp) has been adopted for the calculations. Radiative rates are listed for all E1, E2, M1, and M2 transitions involving the lowest 375 levels. Previous experimental and theoretical energies are available for only a few levels of three, namely Ru X, Rh XI and Pd XII. Differences with the measured energies are up to 4% but the present results are an improvement (by up to 0.3 Ryd) in comparison to other recently reported theoretical data. Similarly for radiative rates and lifetimes, prior results are limited to those involving only 31 levels of the 4s²4p⁵, 4s²4p⁴4d, and 4s4p⁶ configurations for the last four ions. Moreover, there are generally no discrepancies with our results, although the larger calculations reported here differ by up to two orders of magnitude for a few transitions.

Radiative rates for E1, E2, M1, and M2 transitions in F-like ions with 37≤Z≤53

Calculations of energy levels, radiative rates and lifetimes are reported for 17 F-like ions with 37≤Z≤53. For brevity, results are only presented among the lowest 113 levels of the 2s²2p⁵, 2s2p⁶, 2s²2p⁴3ℓ, 2s2p⁵3ℓ, and 2p⁶3ℓ configurations, although the calculations have been performed for up to 501 levels in each ion. The general-purpose relativistic atomic structure package (grasp) has been adopted for the calculations, and radiative rates (along with oscillator strengths and line strengths) are listed for all E1, E2, M1, and M2 transitions of the ions. Comparisons are made with earlier available experimental and theoretical energies, although these are limited to only a few levels for most ions. Therefore for additional accuracy assessments, particularly for energy levels, analogous calculations have been performed with the Flexible Atomic Code (fac), for up to 72 259 levels. Limited previous results are available for radiative rates for comparison purposes, and no large discrepancy is observed for any transition and/or ion.