Massively Parallel Computation Using Graphics Processors with Application to Optimal Experimentation in Dynamic Control

The rapid growth in the performance of graphics hardware, coupled with recent improvements in its programmability has lead to its adoption in many non-graphics applications, including a wide variety of scientific computing fields. At the same time, a number of important dynamic optimal policy problems in economics are athirst of computing power to help overcome dual curses of complexity and dimensionality. We investigate if computational economics may benefit from new tools on a case study of imperfect information dynamic programming problem with learning and experimentation trade-off, that is, a choice between controlling the policy target and learning system parameters. Specifically, we use a model of active learning and control of a linear autoregression with the unknown slope that appeared in a variety of macroeconomic policy and other contexts. The endogeneity of posterior beliefs makes the problem difficult in that the value function need not be convex and the policy function need not be continuous. This complication makes the problem a suitable target for massively-parallel computation using graphics processors (GPUs). Our findings are cautiously optimistic in that the new tools let us easily achieve a factor of 15 performance gain relative to an implementation targeting single-core processors. Further gains up to a factor of 26 are also achievable but lie behind a learning and experimentation barrier of their own. Drawing upon experience with CUDA programming architecture and GPUs provides general lessons on how to best exploit future trends in parallel computation in economics.     

Blasting Open a Choice Space: Learning Inflectional Morphology for NLP

This article discusses the various aspects of designing a system for eliciting knowledge about language from informants. For each design aspect, various options for implementation are presented, along with their pros, cons, and repercussions for other parts of the knowledge elicitation system. A running example throughout the text is taken from the paradigmatic morphology elicitation module of a system called Boas, which elicits knowledge to support a machine translation system. The main point of the article is an argument about the necessity to analyze the design choice space for complex natural language processing (NLP) systems early, comprehensively, and overtly.     

Parameterizing and Eliciting Text Elements across Languages for Use in Natural Language Processing Systems

This paper analyzes the structure and meaning of text elements cross-linguistically and discusses how that information can be elicited from people in a way that is directly useful for NLP applications. We describe a recently developed computer-based linguistic knowledge elicitation system that initiates a new paradigm of knowledge acquisition methodologies for NLP. In particular, we describe the natural language phenomena the system seeks to cover, the approach to knowledge elicitation and its rationale, the elicitation modules themselves, and broader implications of this work.     

Trifluoromethylation of [60]- and [70]Fullerene in the Ionization Chamber of a Mass Spectrometer

Abstract

Both [60]- and [70]fullerene react with either the Scherer radical (perfluorodiisopropylethylmethyl, C₉F₁₉), a mixture of branched perfluorononenes (C₉F₁₈, the product of hexafluoropropene trimerization), or β-fluorosulphatotetrafluoroethyldiheptafluoroisopropylmethyl radical (C₉F₁₈OSO₂F) in the ionization chamber of a mass spectrometer to give the positive parent ions of trifluoromethylation products, the reaction being accompanied by hydrogen addition. The reaction occurs at least partly on the walls of the ionization chamber, by a radical mechanism employing CF₃ radicals formed from the radical reactants both thermally and under electron impact; only the latter route occurs with the perfluorononenes.     

Laser operation and computation of thermal stress in endpumped 1.1 at.% Nd-doped yttrium aluminium garnet rods with sub-millimetre diameters

Laser operation of a fibre-like 1.1 at.% Nd-doped yttrium aluminium garnet rod with a diameter of 0.34 mm and a length of 4.7 mm has been demonstrated. The fibre was end pumped with 0.5 W of continuous-wave radiation from a laser diode emitting at 808 nm. The measured laser thresholds are in a reasonable agreement with theoretical estimations. The distributions of temperature and stress in the crystal fibre were calculated with finite-element simulations showing that the fibre can be pumped up to a power of 40–50 W without fracture caused by thermal stress. The advantages and disadvantages of the proposed laser rod design are discussed.     

Hydrophobicity of Thin Films of Compounds of Low‐Electronegativity Metals

Hydrophobic materials capable of withstanding harsh conditions are required for various applications. Here, we show that oxides and nitrides of various low‐electronegativity metals are hydrophobic hard ceramics. We attribute their hydrophobicity to low Lewis acidity of the low‐electronegativity cations implying a low ability of the cations on the surface to form coordinate bonds with water oxygen anions. Furthermore, we observe a systematically stronger hydrophobic behavior of nitrides compared with the corresponding oxides, which we attribute to nitrogen being a poorer Lewis base than oxygen due to a reduced number of lone pairs of electrons, implying a lower ability of nitrogen anions on the surface to form hydrogen bonds with water hydrogen cations. Most of the oxides and nitrides investigated exhibit high values of hardness. Therefore, oxides and nitrides of low‐electronegativity metals should find application as hydrophobic materials in harsh conditions.     

Domain Wall Motion Across Various Grain Boundaries in Ferroelectric Thin Films

Domain wall movement at and near engineered 10°, 15°, and 24° tilt and 10° and 30° twist grain boundaries was measured by band excitation piezoresponse force microscopy for Pb(Zr,Ti)O₃ films with Zr/Ti ratio of 45/55 and 52/48. A minimum in nonlinear response was observed at the grain boundary for the highest angle twist and tilt grain boundaries, while a maximum in nonlinear response was observed at the 10° tilt grain boundaries. The observed nonlinear response was correlated with the domain configurations imaged in cross section by transmission electron microscopy.     

Complex Formation of Divalent Cations with Carboxylic Acid Resins as a Factor Determining Different Stationary States of Their Swelling

Three samples of carboxylic acid gel-type cross-linked polyelectrolytes were examined by registering the extent and kinetics of their volume change depending on the temperature, concentration, and nature of external electrolyte solutions, as well as the initial state (ionic form, water content) of the bead. It is found that in some systems the swelling of the bead (as a result of lowering the temperature) proceeds much more slowly than the preceding shrinking (as a result of the increase of temperature). Moreover, the initial volume of the bead is not restored, and a new stationary state of swelling is established. By means of optical dilatometry, it is found that in the systems in which the differences in the stationary states of swelling are found, the beads display plasto-elastic properties, whereas in the absence of such differences the beads show elastic properties. It is found that different stationary states of swelling are also established when the beads of copolymer of methacrylic acid cross-linked with divinylbenzene in Na⁺ form in the dry state and in the NaCl solution-swollen state are placed into a solution of a divalent cation salt. The difference in the stationary states of swelling reaches 12% in 4 N NiCl₂ solution. The substantial differences in the swelling behavior and viscoelastic properties of the resins in their Na⁺, Mg²⁺, Ca²⁺, and Ni²⁺ forms are interpreted in terms of formation of different kinds of complexes of functional groups of the polymer with Ca²⁺ and Ni²⁺ ions.