A re-ordered valence band in GaN/AlGaN quantum wells with respect to GaN epilayers has been found as a result of the observation of an enhanced g-factor in magneto-luminescence spectra in fields up to 55 T. This has been caused by a reversal of the states in the strained AlGaN barriers thus giving different barrier heights for the different quantum well hole states. From k.p calculations in the quasi-cubic approximation, a change in the valence-band ordering will account for the observed values for the g-factors. We have also observed the well-width dependence of the in-plane extent of the excitonic wavefunction from which we infer an increase in the exciton binding energy with the reduction of the well width in general agreement with theoretical calculations of Bigenwald et al (phys. stat. sol. (b) 216, 371 (1999)) that uses a variational approach in the envelope function formalism that includes the effect of the electric field in the wells.
https://arxiv.org/abs/cond-mat/0101069
We calculate the expected Mid-IR molecular hydrogen line emission from the first objects in the universe. Due to their low masses, the stellar feedback from massive stars is able to blowaway their gas content and collect it into a cooling shell where H_2 rapidly forms and IR roto-vibrational (as for example the restframe 2.12 micron) lines carry away a large fraction (up to 10%) of the explosion energy. The fluxes from these sources are in the range 10^{-21}-10^{-17} erg/s/cm^2. The highest number counts are expected in the 20 micron band, where about 10^5 sources/deg^2 are predicted at the limiting flux of 3 \times 10^{-18} erg/s/cm^2. Among the planned observational facilities, we find that the best detection perspectives are offered by NGST, which should be able to reveal about 200 first objects in one hour observation time at its limiting flux in the above band. Therefore, Mid-IR instruments appear to represent perfect tools to trace star formation and stellar feedback in the high (z>5) redshift universe.
https://arxiv.org/abs/astro-ph/0005461
A number of diverse bulk properties of the zincblende and wurtzite III-V nitrides AlN, GaN, and InN, are predicted from first principles within density functional theory using the plane-wave ultrasoft pseudopotential method, within both the LDA (local density) and GGA (generalized gradient) approximations to the exchange-correlation functional. Besides structure and cohesion, we study formation enthalpies (a key ingredient in predicting defect solubilities and surface stability), spontaneous polarizations and piezoelectric constants (central parameters for nanostructure modeling), and elastic constants. Our study bears out the relative merits of the two density functional approaches in describing diverse properties of the III-V nitrides (and of the parent species N$_2$, Al, Ga, and In), and leads us to conclude that the GGA approximation, associated with high-accuracy techniques such as multiprojector ultrasoft pseudopotentials or modern all-electron methods, is to be preferred in the study of III-V nitrides.
https://arxiv.org/abs/cond-mat/0011434
It might appear that natural language processing should improve the accuracy of information retrieval systems, by making available a more detailed analysis of queries and documents. Although past results appear to show that this is not so, if the focus is shifted to short phrases rather than full documents, the situation becomes somewhat different. The ANVIL system uses a natural language technique to obtain high accuracy retrieval of images which have been annotated with a descriptive textual caption. The natural language techniques also allow additional contextual information to be derived from the relation between the query and the caption, which can help users to understand the overall collection of retrieval results. The techniques have been successfully used in a information retrieval system which forms both a testbed for research and the basis of a commercial system.
自然语言处理似乎应该通过对查询和文档进行更详细的分析来提高信息检索系统的准确性。尽管过去的结果似乎表明情况并非如此，但如果把重点转移到短语而不是完整的文件上，情况会有所不同。 ANVIL系统使用自然语言技术来获得用描述性文本标题注释的图像的高精度检索。自然语言技术还允许从查询和标题之间的关系中获得附加的上下文信息，这可以帮助用户理解检索结果的总体集合。这些技术已经成功地用于一个信息检索系统，它既是研究的试验台，又是商业系统的基础。
https://arxiv.org/abs/cs/0011028
We present the results of a wide-field survey for extremely red objects (EROs hereafter), the widest so far, based on Ks and R band imaging. The survey covers 701 arcmin^2 and it is 85% complete to Ks\leq18.8 over the whole area and to Ks\leq19.2 over 447.5 arcmin^2. Thanks to the wide field covered, a complete sample of about 400 EROs with R-Ks\geq5 was selected. The distribution of the EROs on the sky is strongly inhomogeneous, being characterized by overdensities and large voids. We detect at the 8 sigma level a strong clustering signal of the EROs which is about an order of magnitude larger than the clustering of K-selected field galaxies in the same magnitude range. A smooth trend of increasing clustering amplitude with the R-Ks color is observed. These results are strong evidence that the largest fraction of EROs is composed of high-z ellipticals, of which we detect for the first time the z\simgt1 large scale structure clustering signal. We show how the surface density variations of the ERO population found in our survey can explain the highly discrepant results obtained so far on the density of z > 1 ellipticals, and we briefly discuss the main implications of our results for the evolution of elliptical galaxies. The number counts and the colors of the K-selected field galaxies are also presented and briefly discussed.
https://arxiv.org/abs/astro-ph/0005581
An associative memory has been discussed of neural networks consisting of spiking N (=100) Hodgkin-Huxley (HH) neurons with time-delayed couplings, which memorize P patterns in their synaptic weights. In addition to excitatory synapses whose strengths are modified after the Willshaw-type learning rule with the 0/1 code for quiescent/active states, the network includes uniform inhibitory synapses which are introduced to reduce cross-talk noises. Our simulations of the HH neuron network for the noise-free state have shown to yield a fairly good performance with the storage capacity of $\alpha_c = P_{\rm max}/N \sim 0.4 - 2.4$ for the low neuron activity of $f \sim 0.04-0.10$. This storage capacity of our temporal-code network is comparable to that of the rate-code model with the Willshaw-type synapses. Our HH neuron network is realized not to be vulnerable to the distribution of time delays in couplings. The variability of interspace interval (ISI) of output spike trains in the process of retrieving stored patterns is also discussed.
https://arxiv.org/abs/cond-mat/0007198
We have analysed VSOP (VLBI Space Observatory Programme) data at 5 GHz and ground-based VLBI (Very Long Baseline Interferometry) data at 15 GHz for the GHz-Peaked Spectrum (GPS) radio galaxy 2021+614. Its morphology is consistent with it being a compact symmetric source extending over 30 h^-1 pc (H_0 = 100 h km s^-1 Mpc^-1, q_0 = 0.5). From a comparison with earlier observations we have detected an increase in the separation and a decrease in the size of the two most prominent components. We determine the projected speed with which these two components recede from each other to be 0.12+/-0.02 h^-1 c. Given the projected separation of the two components of 16.1 h^-1 pc, the infered kinematic age is 440+/-80 years, measured in the source reference frame. These results provide additional support for the contention that compact symmetric radio objects are young and the precursors of the classical FR I or FR II radio sources. The sizes of individual components appear to contract with time which is not consistent with the self-similar evolution model for peaked spectrum sources. In order to overcome problems related to the estimation of uncertainties for separation measurements between source components, we have developed and applied a method that compares two uv-data sets obtained at different epochs. This method parametrizes the most important structural change, the increase in separation between components, by rescaling the u and v axis of the amplitude interference pattern. It provides best-fit values for the parameters and uses a bootstrap method to estimate the errors in the parameters.
https://arxiv.org/abs/astro-ph/0007068
[Abriged] Astronomical Wide Field Imaging performed with new large format CCD detectors poses data reduction problems of unprecedented scale which are difficult to deal with traditional interactive tools. We present here NExt (Neural Extractor): a new Neural Network (NN) based package capable to detect objects and to perform both deblending and star/galaxy classification in an automatic way. Traditionally, in astronomical images, objects are first discriminated from the noisy background by searching for sets of connected pixels having brightnesses above a given threshold and then they are classified as stars or as galaxies through diagnostic diagrams having variables choosen accordingly to the astronomer’s taste and experience. In the extraction step, assuming that images are well sampled, NExt requires only the simplest a priori definition of “what an object is” (id est, it keeps all structures composed by more than one pixels) and performs the detection via an unsupervised NN approaching detection as a clustering problem which has been thoroughly studied in the artificial intelligence literature. In order to obtain an objective and reliable classification, instead of using an arbitrarily defined set of features, we use a NN to select the most significant features among the large number of measured ones, and then we use their selected features to perform the classification task. In order to optimise the performances of the system we implemented and tested several different models of NN. The comparison of the NExt performances with those of the best detection and classification package known to the authors (SExtractor) shows that NExt is at least as effective as the best traditional packages.
https://arxiv.org/abs/astro-ph/0006115
We present ab-initio local density FLAPW calculations on non-reactive N-terminated [001] ordered GaN/Ag and GaN/Au interfaces and compare the results (such as metal induced gap states and Schottky barrier heights) with those obtained for GaN/Al, in order to understand the dependence of the relevant electronic properties on the deposited metal. Our results show that the density of gap states is appreciable only in the first semiconductor layer close to the interface. The decay length of the gap states in the semiconductor side is about 2.0 $\pm$ 0.1 \AA $:$ and is independent of the deposited metal, therefore being to a good extent a bulk property of GaN. Our calculated values of the Schottky barrier heights are $\Phi_{B_p}(GaN/Ag)$ = 0.87 eV and $\Phi_{B_p}(GaN/Au)$ = 1.08 eV; both values are smaller than the GaN/Al value ($\Phi_{B_p}(GaN/Al)$ = 1.51 eV) and this quite large spread of values excludes the possibility of a Fermi level pinning within the GaN band gap. Because of the low screening in GaN, the potential barrier at the junction is strongly affected by the structural arrangement of the first metal layer at the interface. This leads to quite large variations of the Schottky barrier height as a function of the metal, in contrast with the behavior of GaAs/metal interfaces.
https://arxiv.org/abs/cond-mat/0002353
We found that at least 18 Young Stellar Object candidates invisible at optical and near-infrared wavelengths, as well as ^12CO emission. They are IRAS point sources with FIR colors similar to those of usual YSOs, and the infrared spectral indices between 2.2 um and 25 um correspond to those of class I YSOs. These peculiar far-infrared objects are highly concentrated around the galactic plane and the distances are estimated to be around 1 kpc. Although their distribution is away from molecular clouds, some of them seem to be associated with large dark clouds or weak radio sources.
https://arxiv.org/abs/astro-ph/9909387
We propose a realistic device for detecting objects almost without transferring a single quantum of energy to them. The device can work with an efficiency close to 100% and relies on two detectors counting both presence and absence of the objects. Its possible usage in performing fundamental experiments as well as possible applications are discussed.
https://arxiv.org/abs/quant-ph/9906102
We show that with an efficiency exceeding 99% one can use a monolithic total-internal-reflection resonator in order to ascertain the presence of an object without transferring a quantum of energy to it. We also propose an experiment on the probabilistic meaning of the electric field that contains only a very few photons.
https://arxiv.org/abs/quant-ph/9908023
Results of a first-principles study of the Li impurity in GaN are presented. We find Li is a channel interstitial, with an onset for diffusion at T$\sim$ 600 K. Above this temperature, Li can transform to a Ga-substitutional acceptor by exothermic recombination with Ga vacancies. This process implies capture of at least one electron; therefore Li acts as an electron sink. Li$_{\rm Ga}$ is stable again interstitialcy, and has a shallow first ionization levels of 0.16 eV, and second ionization at 0.63 eV. Lattice locations and their temperature dependence are in close agreement with recent experiments.
https://arxiv.org/abs/cond-mat/9908038
The paper describes the speech to speech translation system INTARC, developed during the first phase of the Verbmobil project. The general design goals of the INTARC system architecture were time synchronous processing as well as incrementality and interactivity as a means to achieve a higher degree of robustness and scalability. Interactivity means that in addition to the bottom-up (in terms of processing levels) data flow the ability to process top-down restrictions considering the same signal segment for all processing levels. The construction of INTARC 2.0, which has been operational since fall 1996, followed an engineering approach focussing on the integration of symbolic (linguistic) and stochastic (recognition) techniques which led to a generalization of the concept of a ``one pass’’ beam search.
https://arxiv.org/abs/cs/9907021
In this paper we present an application of explanation-based learning (EBL) in the parsing module of a real-time English-Spanish machine translation system designed to translate closed captions. We discuss the efficiency/coverage trade-offs available in EBL and introduce the techniques we use to increase coverage while maintaining a high level of space and time efficiency. Our performance results indicate that this approach is effective.
在本文中，我们提出了解释学习（EBL）在实时英语 - 西班牙语机器翻译系统的解析模块中的应用，用于翻译隐藏式字幕。我们讨论EBL中可用的效率/覆盖率折衷，并介绍我们用来增加覆盖率的技术，同时保持高水平的空间和时间效率。我们的表现结果表明这种方法是有效的。
https://arxiv.org/abs/cs/9907008
A detailed investigation by one-step photoemission calculations of the GaN(0001)-(1x1) surface in comparison with recent experiments is presented in order to clarify its structural properties and electronic structure. The discussion of normal and off-normal spectra reveals through the identified surface states clear fingerprints for the applicability of a surface model proposed by Smith et al. Especially the predicted metallic bonds are confirmed. In the context of direct transitions the calculated spectra allow to determine the valence band width and to argue in favor of one of two theoretical bulk band structures. Furthermore a commonly used experimental method to fix the valence band maximum is critically tested.
https://arxiv.org/abs/cond-mat/9905245
We describe an architecture and investigate the characteristics of distributed searching in federated digital libraries. We introduce the notion of a query mediator as a digital library service responsible for selecting among available search engines, routing queries to those search engines, and aggregating results. We examine operational data from the NCSTRL distributed digital library that reveals a number of characteristics of distributed resource discovery. These include availability and response time of indexers and the distinction between the query mediator view of these characteristics and the indexer view.
https://arxiv.org/abs/cs/9902020
MPEG is undertaking a new initiative to standardize content description of audio and video data/documents. When it is finalized in 2001, MPEG-7 is expected to provide standardized description schemes for concise and unambiguous content description of data/documents of complex media types. Meanwhile, other meta-data or description schemes, such as Dublin Core, XML, etc., are becoming popular in different application domains. In this paper, we propose the Multimedia Description Framework (MDF), which is designated to accommodate multiple description (meta-data) schemes, both MPEG-7 and non-MPEG-7, into integrated architecture. We will use examples to show how MDF description makes use of combined strength of different description schemes to enhance its expression power and flexibility. We conclude the paper with discussion of using MDF description of a movie video to search/retrieve required scene clips from the movie, on the MDF prototype system we have implemented.
https://arxiv.org/abs/cs/9902016
A planar force-constant model is developed for longitudinal phonons of wurtzite GaN and AlN propagating along the [0001] direction. The proposed model is then applied to the study of the phonon modes in hexagonal GaN/AlN superlattices in the longitudinal polarization. The confinement of the superlattice phonon mode is discussed.
https://arxiv.org/abs/cond-mat/9812293
We develop a class of new kinetic data structures for collision detection between moving convex polytopes; the performance of these structures is sensitive to the separation of the polytopes during their motion. For two convex polygons in the plane, let $D$ be the maximum diameter of the polygons, and let $s$ be the minimum distance between them during their motion. Our separation certificate changes $O(\log(D/s))$ times when the relative motion of the two polygons is a translation along a straight line or convex curve, $O(\sqrt{D/s})$ for translation along an algebraic trajectory, and $O(D/s)$ for algebraic rigid motion (translation and rotation). Each certificate update is performed in $O(\log(D/s))$ time. Variants of these data structures are also shown that exhibit \emph{hysteresis}—after a separation certificate fails, the new certificate cannot fail again until the objects have moved by some constant fraction of their current separation. We can then bound the number of events by the combinatorial size of a certain cover of the motion path by balls.
https://arxiv.org/abs/cs/9809035
The free-carrier screening of macroscopic polarization fields in wurtzite GaN/InGaN quantum wells lasers is investigated via a self-consistent tight-binding approach. We show that the high carrier concentrations found experimentally in nitride laser structures effectively screen the built-in spontaneous and piezoelectric polarization fields, thus inducing a ``field-free’’ band profile. Our results explain some heretofore puzzling experimental data on nitride lasers, such as the unusually high lasing excitation thresholds and emission blue-shifts for increasing excitation levels.
https://arxiv.org/abs/cond-mat/9809409
The Zeeman splitting of the donor spectra in cubic- and hexagonal-GaN are studied using an effective mass theory approach. Soft-core pseudopotentials were used to describe the chemical shift of the different substitutional dopants. The donor ground states calculated range from 29.5 to 33.7 meV, with typically 1 meV higher binding in the hexagonal phase. Carbon is found to produce the largest donor binding energy. The ionization levels and excited states are in excellent agreement with Hall and optical measurements, and suggest the presence of residual C in recent experiments.
https://arxiv.org/abs/cond-mat/9809219
We employ effective mass theory for degenerate hole-bands to calculate the acceptor binding energies for Be, Mg, Zn, Ca, C and Si substitutional acceptors in GaN and AlN. The calculations are performed through the 6$\times $6 Rashba-Sheka-Pikus and the Luttinger-Kohn matrix Hamiltonians for wurtzite (WZ) and zincblende (ZB) crystal phases, respectively. An analytic representation for the acceptor pseudopotential is used to introduce the specific nature of the impurity atoms. The energy shift due to polaron effects is also considered in this approach. The ionization energy estimates are in very good agreement with those reported experimentally in WZ-GaN. The binding energies for ZB-GaN acceptors are all predicted to be shallower than the corresponding impurities in the WZ phase. The binding energy dependence upon the crystal field splitting in WZ-GaN is analyzed. Ionization levels in AlN are found to have similar `shallow’ values to those in GaN, but with some important differences, which depend on the band structure parameterizations, especially the value of crystal field splitting used.
https://arxiv.org/abs/cond-mat/9809217
The diffusion of Ga and N adatoms has been studied for the technologically relevant wurtzite (000bar1) and (0001) surfaces employing density-functional theory. Our calculations reveal a very different diffusivity for Ga and N adatoms on the equilibrium surfaces: While Ga is very mobile at typical growth temperatures, the diffusion of N is by orders of magnitudes slower. These results give a very detailed insight of how and under which growth conditions N adatoms can be stabilized and efficiently incorporated at the surface. We further find that the presence of excess N strongly increases the Ga diffusion barrier and discuss the consequences for the growth of GaN.
https://arxiv.org/abs/cond-mat/9809006
We present a new cellular data processing scheme, a hybrid of existing cellular automata (CA) and gate array architectures, which is optimized for realization at the quantum scale. For conventional computing, the CA-like external clocking avoids the time-scale problems associated with ground-state relaxation schemes. For quantum computing, the architecture constitutes a novel paradigm whereby the algorithm is embedded in spatial, as opposed to temporal, structure. The architecture can be exploited to produce highly efficient algorithms: for example, a list of length N can be searched in time of order cube root N.
https://arxiv.org/abs/cond-mat/9808243
We discuss some aspects concerning the formation and the impact of the first luminous structures in the universe (PopIIIs), with particular emphasis on their feedback effects on subsequent galaxy formation. We argue that supernovae in these objects might provide, if detected, important constraints for cosmological models. Some of these constraints come from the gravitational magnification of the supernova flux due to the intervening cosmological matter distribution which results in different predictions, for example, on rates, magnification probability, detection limits and number counts of distant supernovae. We show that NGST can tremendously contribute to such a study due to its advanced technological capabilities.
https://arxiv.org/abs/astro-ph/9807292
Ab-initio calculations have been used to study the influence of the interface morphology and, notably, of the exchange reaction on the electronic properties of Al/GaN (100) interfaces. In particular, the effects of interface structure (i.e. interfacial bond lengths, semiconductor surface polarity, and reacted intralayers) on the SBH at the Al/GaN (001) junction are specifically addressed. The electronic structure of the following atomic configurations have been investigated theoretically: (i) an abrupt, relaxed GaN/Al interface; (ii) an interface which has undergone one monolayer of exchange reaction; and interfaces with a monolayer-thick interlayer of (iii) AlN and (iv) Ga${0.5}$Al${0.5}$N. Intermixed interfaces are found to pin the interface Fermi level at a position not significantly different from that of an abrupt interface. Our calculations also show that the interface band line–up is not strongly dependent on the interface morphology changes studied. The p-type SBH is reduced by less than 0.1 eV if the GaN surface is Ga-terminated compared to the N-terminated one. Moreover, we show that both an ultrathin Ga$x$Al${1-x}$N ($x$ = 0, 0.5) intralayer and a Ga$\leftrightarrow$Al atomic swap at the interface does not significantly affect the Schottky barrier height.
https://arxiv.org/abs/cond-mat/9806377
We present the results of multi-epoch global VLBI observations of the Compact Symmetric Object (CSO), 0710+439 at 5 GHz. Analysis of data spread over 13 years shows strong evidence for an increase in the separation of the outer components at a rate of $0.251 \pm 0.029 h^{-1}c$. Dividing the overall size of 86.8 $h^{-1}$pc by this separation rate implies an estimated kinematic age of only $1100 \pm 100$ yrs. After taking into account possible temporal variations in hotspot advance speeds due to cloud collisions or hydrodynamic instabilities we argue that the upper limit to the age of 0710+439 is most likely within a factor of 2 of this estimate and certainly within a factor of 10 (i.e. $<11 000$ yrs). This result therefore strongly supports the idea that Compact Symmetric Objects are very young radio-loud sources. Furthermore the large radiative efficiency we calculate for 0710+439 is consistent with strong negative luminosity evolution as CSOs grow in size and with them evolving into classical double sources.
https://arxiv.org/abs/astro-ph/9712062
A database of radio-quiet quasars detected with ROSAT is presented containing 846 quasars seen in the All-Sky Survey and/or in pointed PSPC observations. About 70% of the objects have been detected in X-rays for the first time. We present the soft X-ray fluxes and spectra, if available. Using an optically selected subsample compiled from this database, we study the broad band properties of radio-quiet quasars with high statistical significance. We confirm that radio-quiet quasars have in general steeper soft X-ray spectra than radio-loud. The spectral differences persist to high redshifts. A spectral flattening with increasing redshift is confirmed up to redshift around 2, beyond which the spectral slopes seem to be independent of redshift. The spectral slopes of the ROSAT radio-quiet quasars at z>2.5 are consistent with those found for nearby quasars in the medium energy band (2-10keV), implying that the X-ray spectral evolution is not important. We show that there is, in a statistical sense, little or no excess absorption for most of the radio-quiet objects at z>2. The existence of a correlation between the X-ray luminosity and the luminosity at 2500A is confirmed. Individual objects show a large scatter from this correlation. The X-ray loudness alpha_ox appears to be independent of z, but the analyses indicate a slight increase of alpha_ox with optical luminosity. However, this behavior is likely not caused by physical properties inherent to the quasars but is the result of the intrinsic dispersion. Finally, we find a small fraction of objects which appear to be relatively `X-ray quiet’ compared to the bulk of the other quasars.
https://arxiv.org/abs/astro-ph/9805015
We have investigated clean and As-covered zinc-blende GaN (001) surfaces, employing first-principles total-energy calculations. For clean GaN surfaces our results reveal a novel surface structure very different from the well-established dimer structures commonly observed on polar III-V (001) surfaces: The energetically most stable surface is achieved by a Peierls distortion of the truncated (1x1) surface rather than through addition or removal of atoms. This surface exhibits a (1x4) reconstruction consisting of linear Ga tetramers. Furthermore, we find that a submonolayer of arsenic significantly lowers the surface energy indicating that As may be a good surfactant. Analyzing surface energies and band structures we identify the mechanisms which govern these unusual structures and discuss how they might affect growth properties.
https://arxiv.org/abs/cond-mat/9705111
This paper has been withdrawn by the authors until some changes are made.
https://arxiv.org/abs/physics/9802032
By redshift of 10, star formation in the first objects should have produced considerable amounts of Carbon, Nitrogen and Oxygen. The submillimeter lines of C, N and O redshift into the millimeter and centimeter bands (0.5 mm – 1.2 cm), where they may be detectable. High spectral resolution observations could potentially detect inhomogeneities in C, N and O emission, and see the first objects forming at high redshift. We calculate expected intensity fluctuations and discuss frequency and angular resolution required to detect them. For CII emission, we estimate the intensity using two independent methods: the line emission coefficient argument and the luminosity density argument. We find they are in good agreement. At 1+z \sim 10, the typical protogalaxy has a velocity dispersion of 30 km s^{-1} and angular size of 1 arcsecond. If CII is the dominant coolant, then we estimate a characteristic line strength of \sim 0.1 K km s^{-1}. We also discuss other atomic lines and estimate their signal. Observations with angular resolution of 10^{-3} can detect moderately nonlinear fluctuations of amplitude 2 \cdot 10^{-5} times the microwave background. If the intensity fluctuations are detected, they will probe matter density inhomogeneity, chemical evolution and ionization history at high redshifts.
https://arxiv.org/abs/astro-ph/9803236
Colloquial English (CE) as found in television programs or typical conversations is different than text found in technical manuals, newspapers and books. Phrases tend to be shorter and less sophisticated. In this paper, we look at some of the theoretical and implementational issues involved in translating CE. We present a fully automatic large-scale multilingual natural language processing system for translation of CE input text, as found in the commercially transmitted closed-caption television signal, into simple target sentences. Our approach is based on the Whitelock’s Shake and Bake machine translation paradigm, which relies heavily on lexical resources. The system currently translates from English to Spanish with the translation modules for Brazilian Portuguese under development.
在电视节目或典型对话中发现的口语英语（CE）与技术手册，报纸和书籍中的文本不同。短语往往更短，不太复杂。在本文中，我们看一些涉及翻译CE的理论和实施问题。我们提供了一个全自动的大型多语言自然语言处理系统，用于在商业传输的隐藏字幕电视信号中找到的用于CE输入文本的翻译成简单的目标语句。我们的方法是基于Whitelock的Shake and Bake机器翻译范式，这种范式在很大程度上依赖于词汇资源。该系统目前正在翻译从英文到西班牙文，正在开发的巴西葡萄牙语翻译模块。
https://arxiv.org/abs/cmp-lg/9706024
It is being proved that the neurochip \Totem{} is a viable solution for high quality and real time computational tasks in HEP, including event classification, triggering and signal processing. The architecture of the chip is based on a “derivative free” algorithm called Reactive Tabu Search (RTS), highly performing even for low precision weights. ISA, VME or PCI boards integrate the chip as a coprocessor in a host computer. This paper presents: 1) the state of the art and the next evolution of the design of \Totem{}; 2) its ability in the Higgs search at LHC as an example.
https://arxiv.org/abs/hep-ex/9705008
In this paper, we describe a method for automatic creation of a knowledge source for text generation using information extraction over the Internet. We present a prototype system called PROFILE which uses a client-server architecture to extract noun-phrase descriptions of entities such as people, places, and organizations. The system serves two purposes: as an information extraction tool, it allows users to search for textual descriptions of entities; as a utility to generate functional descriptions (FD), it is used in a functional-unification based generation system. We present an evaluation of the approach and its applications to natural language generation and summarization.
https://arxiv.org/abs/cmp-lg/9702014
Ab initio calculations predict that Be is a shallow acceptor in GaN. Its thermal ionization energy is 0.06 eV in wurtzite GaN; the level is valence resonant in the zincblende phase. Be incorporation is severely limited by the formation of Be_3N_2. We show however that co-incorporation with reactive species can enhance the solubility. H-assisted incorporation should lead to high doping levels in MOCVD growth after post-growth annealing at about 850 K. Be-O co-incorporation produces high Be and O concentrations at MBE growth temperatures.
https://arxiv.org/abs/cond-mat/9610108
The strain induced by lattice mismatch at the interface is responsible for the different value of the band discontinuities observed recently for the AlN/GaN (AlN on GaN) and the GaN/AlN (GaN on AlN) polar (0001) interface. We present a first-principles calculation of valence band offsets, interface dipoles, strain-induced piezoelectric fields, relaxed geometric structure, and formation energies. Our results confirm the existence of a large forward-backward asymmetry for this interface.
https://arxiv.org/abs/cond-mat/9612107
Zn in GaN forms an efficient radiative center and acts as a deep acceptor which can make the crystal insulating. Four different Zn-related centers have been by now identified, leading to light emission in the range between 1.8 eV and 2.9 eV. We present a first-principles investigation total energy and electronic structure calculations for Ga-substitutional and hetero-antisite N-substitutional Zn in wurtzite GaN, using ultrasoft pseudopotentials and a conjugate-gradient total energy minimization method. Our results permit the identification of the blue-light emission center as the substitutional acceptor, while contrary to a common belief the Zn_N heteroantisite has a very high formation energy and donor behavior, which seems to exclude it as the origin of the other centers.
https://arxiv.org/abs/cond-mat/9610044
We have calculated formation energies and position of the defect levels for all native defects and for a variety of donor and acceptor impurities employing first-principles total-energy calculations. An analysis of the numerical results gives direct insight into defect concentrations and impurity solubility with respect to growth parameters (temperature, chemical potentials) and into the mechanisms limiting the doping levels in GaN. We show how compensation and passivation by native defects or impurities, solubility issues, and incorporation of dopants on other sites influence the acceptor doping levels.
https://arxiv.org/abs/mtrl-th/9609006
Quantum correlations between two particles show non-classical properties which can be used for providing secure transmission of information. We present a quantum cryptographic system, in which users store particles in quantum memories kept in a transmission center. Correlations between the particles stored by two users are created upon request by projecting their product state onto a fully entangled state. Our system allows for secure communication between any pair of users who have particles in the same center. Unlike other quantum cryptographic systems, it can work without quantum channels and is suitable for building a quantum cryptographic network. We also present a modified system with many centers.
https://arxiv.org/abs/quant-ph/9604021
We report scanning tunneling microscopy (STM) images of surfaces of GaN films and the observation of luminescence from those films induced by highly spatially localized injection of electrons or holes using STM. This combination of scanning tunneling luminescence (STL) with STM for GaN surfaces and the ability to observe both morphology and luminescence in GaN is the first step to investigate possible correlations between surface morphology and optical properties.
https://arxiv.org/abs/cond-mat/9511068
A locally iterative learning (LIL) rule is adapted to a model of the associative memory based on the evolving recurrent-type neural networks composed of growing neurons. There exist extremely different scale parameters of time, the individual learning time and the generation in evolution. This model allows us definite investigation on the interaction between learning and evolution. And the reinforcement of the robustness against the noise is also achieved in the evolutional scheme.
https://arxiv.org/abs/adap-org/9507008
If the dark halo of our galaxy is made of compact objects as massive as M = 10^6 solar masses, their detection by means of ordinary microlensing searches would take a very long time as the characteristic time scale of such a lensing event, t_0, is approximately 200 years. Fortunately, the very high magnification events of the numerous faint stars, which are normally well below the detection threshold, have short duration peaks with a characteristic time scale of only t_0 / A_{max}, where A_{max} » 1 is the peak magnification factor. The two almost equally bright images are separated by approximately 2’’ (M/10^6 solar masses )^{1/2}, and they rotate very rapidly around the lens with the relative proper motion enhanced by a factor 2 A_{max}. The same events will offer an opportunity to study spectroscopically stars which are normally far too faint to be reached.
https://arxiv.org/abs/astro-ph/9505016
This paper describes an implementation based on a recent model in the psycholinguistic literature. We define a parsing operation which allows the reanalysis of dependencies within an incremental and monotonic processing architecture, and discuss search strategies for its application in a head-initial language (English) and a head-final language (Japanese).
https://arxiv.org/abs/cmp-lg/9502035
In this paper, we investigate the associative memory in recurrent neural networks, based on the model of evolving neural networks proposed by Nolfi, Miglino and Parisi. Experimentally developed network has highly asymmetric synaptic weights and dilute connections, quite different from those of the Hopfield model. Some results on the effect of learning efficiency on the evolution are also presented.
https://arxiv.org/abs/adap-org/9411003
Spoken language applications in natural dialogue settings place serious requirements on the choice of processing architecture. Especially under adverse phonetic and acoustic conditions parsing procedures have to be developed which do not only analyse the incoming speech in a time-synchroneous and incremental manner, but which are able to schedule their resources according to the varying conditions of the recognition process. Depending on the actual degree of local ambiguity the parser has to select among the available constraints in order to narrow down the search space with as little effort as possible. A parsing approach based on constraint satisfaction techniques is discussed. It provides important characteristics of the desired real-time behaviour and attempts to mimic some of the attention focussing capabilities of the human speech comprehension mechanism.
https://arxiv.org/abs/cmp-lg/9409008
We discuss implementation issues of MARIE-1, a mostly symbolic parser fully implemented, and MARIE-2, a more statistical parser partially implemented. They address a corpus of 100,000 picture captions. We argue that the mixed approach of MARIE-2 should be better for this corpus because its algorithms (not data) are simpler.
我们讨论完全实现的主要符号分析器MARIE-1和部分实现的更多统计分析器MARIE-2的实现问题。他们处理了100,000个图片标题的语料库。我们认为MARIE-2的混合方法应该更适合这个语料库，因为它的算法（而不是数据）更简单。
https://arxiv.org/abs/cmp-lg/9408008
We discuss a possible method for detection of dark companions of galactic objects of stellar mass. Such binary systems are likely to occur in the galactic disk and possibly also in the halo. The high incidence of binary and higher-multiplicity systems in the solar neighborhood, if indicative of the galactic disk at large, implies that current searches for the gravitational microlensing signature of massive compact objects in our galaxy would yield a significant fraction of binary systems. Our calculations suggest that 40% of the lightcurves that will be obtained in such searches may be sufficiently perturbed to reveal, if sufficiently well-sampled, the presence of a compact dark companion of Jovian mass orbiting a primary. The likelihood of occurrence of perturbed lightcurves must also be taken into account by systematic search programs, to improve their event detection efficiency. The statistics of perturbed lensing events, if carefully interpreted, could yield estimates of the incidence of binary systems with low mass ratios, including that of systems with planets.
https://arxiv.org/abs/astro-ph/9407030
The structural and electronic properties of cubic GaN are studied within the local density approximation by the full-potential linear muffin-tin orbitals method. The Ga $3d$ electrons are treated as band states, and no shape approximation is made to the potential and charge density. The influence of $d$ electrons on the band structure, charge density, and bonding properties is analyzed. It is found that due to the energy resonance of the Ga 3$d$ states with nitrogen 2$s$ states, the cation $d$ bands are not inert, and features unusual for a III-V compound are found in the lower part of the valence band and in the valence charge density and density of states. To clarify the influence of the Ga $d$ states on the cohesive properties, additional full and frozen–overlapped-core calculations were performed for GaN, cubic ZnS, GaAs, and Si. The results show, in addition to the known importance of non-linear core-valence exchange-correlation corrections, that an explicit description of closed-shell repulsion effects is necessary to obtain accurate results for GaN and similar systems. In summary, GaN appears to be somewhat exceptional among the III-V compounds and reminiscent of II-VI materials, in that its band structure and cohesive properties are sensitive to a proper treatment of the cation $d$ bands, as a result of the presence of the latter in the valence band range.
https://arxiv.org/abs/cond-mat/9208022