Showing papers by "University of Trento published in 1992"

First- and second-order methods for learning: between steepest descent and Newton's method

Abstract: On-line first-order backpropagation is sufficiently fast and effective for many large-scale classification problems but for very high precision mappings, batch processing may be the method of choice. This paper reviews first- and second-order optimization methods for learning in feedforward neural networks. The viewpoint is that of optimization: many methods can be cast in the language of optimization techniques, allowing the transfer to neural nets of detailed results about computational complexity and safety procedures to ensure convergence and to avoid numerical problems. The review is not intended to deliver detailed prescriptions for the most appropriate methods in specific applications, but to illustrate the main characteristics of the different methods and their mutual relations.

Semi-implicit finite difference methods for three-dimensional shallow water flow

Abstract: A semi-implicit finite difference method for the numerical solution of three-dimensional shallow water flows is presented and discussed. The governing equations are the primitive three-dimensional turbulent mean flow equations where the pressure distribution in the vertical has been assumed to be hydrostatic. In the method of solution a minimal degree of implicitness has been adopted in such a fashion that the resulting algorithm is stable and gives a maximal computational efficiency at a minimal computational cost. At each time step the numerical method requires the solution of one large linear system which can be formally decomposed into a set of small three-diagonal systems coupled with one five-diagonal system. All these linear systems are symmetric and positive definite. Thus the existence and uniquencess of the numerical solution are assured. When only one vertical layer is specified, this method reduces as a special case to a semi-implicit scheme for solving the corresponding two-dimensional shallow water equations. The resulting two- and three-dimensional algorithm has been shown to be fast, accurate and mass-conservative and can also be applied to simulate flooding and drying of tidal mud-flats in conjunction with three-dimensional flows. Furthermore, the resulting algorithm is fully vectorizable for an efficient implementation on modern vector computers.

Aesthetic Understanding of Organizational Life

Abstract: The weak point of study of aesthetics in organizational life has been theorists' definition of the object of analysis, even before their use of methodology and techniques. This article takes a holistic approach to organizations in order to promote aesthetic awareness as a legitimate form of understanding organizational life. It is in contrast to previous approaches that treat the aesthetic dimension as one organizational theme among many. The article illustrates the elusiveness of the aesthetic dimension as an object of knowledge, and it also demonstrates the close link between the organizational aesthetic and the complex process of its deconstruction, comprehension, and communication.

Luminescence of Eu3+ ions during thermal densification of SiO2 gel

Abstract: This paper compares the luminescence spectra of Eu3+ in sol-gel derived silica samples heated at different temperatures, following the densification process from wet gel to compact silica glass Lifetimes, linewidths, and Stark splittings of the transitions were used to study the structural evolution of the gel network

High-precision determination of the temperature dependence of the fundamental energy gap in gallium arsenide.

Abstract: The photoluminescence (PL) spectra of GaAs have been measured as a function of temperature between 2 and 280 K. Measurements have been performed on a high-quality nominally undoped sample grown by molecular-beam epitaxy. At the lower temperatures the recombination of free excitons in the n=1 and 2 states is observed. Increasing the temperature, the interband recombination appears and eventually dominates the PL spectra. The spectra have been successfully fitted by a spectral-line-shape theory that considers both excitonic and band-to-band transitions. The fits demonstrate that even at the highest temperatures a well-defined narrow peak due to the n=1 exciton is observable: its energy corresponds to the energy of the maximum of the PL spectra (${\mathit{E}}_{\mathit{M}}$). Hence, by adding the exciton binding energy to ${\mathit{E}}_{\mathit{M}}$, the value of the energy gap (${\mathit{E}}_{\mathit{G}}$) at each temperature has been deduced from the spectra. This way an accurate determination of the temperature dependence of ${\mathit{E}}_{\mathit{G}}$ in GaAs is obtained; values for the parameters of the semiempirical relations describing ${\mathit{E}}_{\mathit{G}}$(T) are found and compared with the literature.

Consensus in distributed soft environments

Abstract: This paper studies the problem of formalizing consensus reaching within a set of decision makers trying to find and agree upon a mutual decision. Decision makers produce their individual rankings, using their own pet decision schemas. Thus consensus reaching relies only on the aggregation of individual decisions rather than on individual decision procedures. The aggregation of the individual rankings is supported by an advising monitor which tries to contract the decision makers into a mutual decision through soft enforcement. Convergence to consensus then depends upon the decision makers' willingness to compromise. We use a topological approach to consensus where we can measure distances between decision makers. Within the approach we can also model the trade-off between a degree of consensus and a strength of majority.

Complex dynamics underlying the human electrocardiogram

Abstract: Sequences of different human cardiac rhythms terminating in ventricular fibrillation have been studied, both qualitatively and quantitively, with methods of nonlinear dynamics. The analysis has been applied to ECG epochs belonging to rhythms of increasing electrocardiographic irregularity: from sinus rhythm to prefibrillatory rhythms and then to ventricular fibrillation. The phase portraits of these rhythms have been reconstructed from the ECG recording with the time-delay technique, and their correlation dimensions have been estimated with the algorithm of Grassberger and Procaccia (1983a, b). Different cardiac rhythms exhibit different correlation dimensions that describe the corresponding degrees of complexity. The correlation dimension increases as one proceeds from sinus rhythm to fully developed ventricular fibrillation via intermediate rhythms. The fully developed ventricular fibrillation shows the highest degree of complexity. The dimensional analysis supports the existence of complex dynamics underlying different cardiac rhythms and reveals an increase in dimensional complexity corresponding to an increase in electrocardiographic irregularity. Our results indicate that nonlinear dynamics may be used to assess various dynamic states of the heart and may offer a non-invasive tool to investigate the complex dynamic phenomena occuring during arrhythmia.

Isolation of Toxic and Potentially Toxic Sesqui‐ and Monoterpenes from the Tropical Green Seaweed Caulerpa taxifolia Which Has Invaded the Region of Cap Martin and Monaco

Abstract: The green seaweed Caulerpa taxifolia (VAHL) C. AGARDH (Caulerpales), which, after its recent accidental introduction, is growing in the region of Cap Martin much more vigorously than in the tropics, is shown to contain the known sesquiterpenic toxins caulerpenyne (1) – in larger amounts than in tropical Caulerpales – and oxytoxin 1 (2). Novel, potentially toxic products isolated in small amounts from this seaweed include the sesquiterpenes taxifolial A ( = (5E)-6,10-dimethyl-2-[(E)2-oxoethylidene]undeca-5,9-dien-7- yne-1,3-diyl diacetate; 3), taxifolial B (= (1E,6E,10E)-3-[(Z)-acetoxymethylidene]-7, 11-dimethyl-12-oxododeca-1,6,10-trien-8-yne-1,4-diyl diacetate; 4), 10,11-epoxycaulerpenyne ( = (1E,6E)-3-[(Z)-acetoxymethylidene]-10,11-epoxy-7, 11-dimethyldodeca-1,6-dien-8-yne-1,4-diyl diacetate; 1:1 diastereoisomer mixture; 5), and taxifolial C ( = (2Z,6E)-3-formyl-7,11-dimethyldodeca-2,6,10-trien-8-yne-1,1, 4-triyl triacetate; 6), besides, as the first example of a monoterpene from the Caulerpales, taxifolial D ( = (2Z)-3,7-dimethylocta-2, 6-dien-4-ynal; 7).

Turbulent pressure fluctuations under hydraulic jumps

Abstract: New experimental evidence on the statistical structure of turbulent pressure fluctuations at the bottom of hydraulic jumps is brought in this paper in view of its relevance on stability of the linings in stilling basins. Maximum values and the structures of temporal and spatial correlation of the anisotropic field of fluctuating pressures are described. Pressures are measured in the zones of the jump where the uplift load produced on the slabs is maximum for a 5÷10 range of Froude numbers. The results define a novel design criterion for hydraulic engineering practice. It is concluded, upon comparison with analogous results from extensive although seemingly incomplete literature, that the systematic experimentation described herein completes the information needed to characterize the statistical structure of pressure fields past hydraulic jumps.

Code assignment for hidden terminal interference avoidance in multihop packet radio networks

Abstract: Hidden terminal interference is caused by the simultaneous transmission of two stations that cannot hear each other, but are both received by the same destination station. The authors investigate the problem of assigning orthogonal codes to stations to eliminate the hidden terminal interference and minimize the number of codes used. It is shown that this problem is computationally intractable, even for very restricted but very realistic network topologies. Optimal algorithms for code assignment in special networks, as well as both centralized and distributed suboptimal heuristic algorithms for general topologies, are presented. The results of extensive simulations to derive the average performance of the proposed heuristics on realistic network topologies are presented. >

Microscopic dynamics in liquid alkali metals.

Abstract: The structural and dynamical properties of liquid alkali metals near melting are investigated by realistic computer simulations, which support the idea that several important features follow from suitable scaling criteria. The data are found to be in excellent agreement with the recent theories of single-particle dynamics in simple liquids.

Atomic and molecular hydrogen in gallium arsenide: A theoretical study.

Abstract: We present first-principles calculations of the properties of atomic and molecular hydrogen in pure bulk GaAs. Our results indicate that H penetrates into GaAs in atomic form. Inside GaAs, atomic H tends to form ${\mathrm{H}}_{2}$ molecules in tetrahedral sites, which are deep energy wells for ${\mathrm{H}}_{2}$. The ${\mathrm{H}}_{2}^{\mathrm{*}}$ defect, formed by one H in a bond-center site and one H in an adjacent tetrahedral position, has higher energy than ${\mathrm{H}}_{2}$ but lower-energy barriers for diffusion. Isolated H could be present as a metastable species. We compute the stable charge state of isolated H as a function of the Fermi energy. Our results suggest that H behaves as a negative-U defect. As a consequence, isolated H is expected to be present only as a charged species (positively charged in p-doped samples, negatively charged in undoped and n-doped samples). Our conclusions are compared with experimental results and with the results of calculations for H in other semiconductors. The main features of H in GaAs are quite similar to what has been found in Si.

Corrosion behaviour of electrogalvanized steel in sodium chloride and ammonium sulphate solutions; a study by E.I.S.

Abstract: Zinc and zinc-nickel (13% Ni) electrodeposits were passivated by dipping in chromate baths and characterized by scanning electron microscopy. The corrosion behaviour was studied using a.c. electrochemical techniques; electrochemical impedance spectroscopy (EIS) measurements were performed at open circuit and under galvanostatic control during the 24 h immersion time. In sodium chloride solution the zinc-nickel electrodeposits show a better corrosion resistance compared to the pure zinc coatings. During the immersion time, a surface nickel enrichment was observed which, together with the zinc corrosion products, acts as a barrier layer reducing the total corrosion rate. In the same solution the passivation treatment improves the corrosion resistance of the electrodeposits; nevertheless, on zinc substrates, the protection exerted by the chromate film is not, always effective during the immersion time. On the contrary the chromate coating on zinc-nickel substrates induces a remarkable and durable improvement of the corrosion resistance reducing the zinc dissolution almost completely. In the ammonium sulphate solution, the corrosion mechanism is significantly influenced by hydrogen reduction on the zinc-nickel surfaces, and by the production of a local surface acidity which is aggressive for the chromate coatings.

Effect of Mgo content changes (8 to 15%) on the devitrification of glasses obtained from porphiric sands, Mgo and Tio(in2) (4%)

Abstract: Glasses have been prepared from porphiric sands, TiO2 and MgCO3 and the effect of changing the MgO content on the isothermal devitrification has been studied.

Absolute total-cross-section measurements for intermediate-energy electron scattering on CF4, CClF3, CCl2F2, CCl3F, and CCl4.

Abstract: The total cross sections of chlorofluoromethane molecules were measured in the 75--4000-eV range. A comparison is given with existing experimental and theoretical data. The present data and previous measurements in the 15--50-eV range were fitted with a Born-like formula. This simple fit allows the cross sections to be reproduced within experimental errors. Phenomenological double Yukawa scattering potentials can be derived from the formula. A short discussion for the total-cross-section partitioning is given for ${\mathrm{CF}}_{4}$, ${\mathrm{CF}}_{2}$${\mathrm{Cl}}_{2}$, and ${\mathrm{CCl}}_{4}$.

Silicon Oxycarbide Glasses from Sol-Gel Precursors

Abstract: Silicon oxycarbide glasses have been prepared from sol-gel precursors containing not only Si-CH3, but also Si-H bonds. Three systems have been chosen containing various Si units but the same C/Si ratio. Their pyrolysis process has been mainly followed by 29Si MAS-NMR and the composition of the final glass extracted. This study shows that a suitable choice of Si-CH3 and Si-H functionnalized silicon alkoxides can lead to a strong decrease in the free carbon content and to an almost pure silicon oxycarbide phase.

Gravitons as embroidery on the weave

Abstract: We investigate the physical interpretation of the loop states that appear in the loop representation of quantum gravity. By utilizing the “weave” state, which has been recently introduced as a quantum description of the microstructure of flat space, we analyze the relation between loop states and graviton states. This relation determines a linear map M from the state-space of the nonperturbative theory (loop space) into the state-space of the linearized theory (Fock space). We present an explicit form of this map, and a preliminary investigation of its properties. The existence of such a map indicates that the full nonperturbative quantum theory includes a sector that describes the same physics as (the low energy regimes of) the linearized theory, namely gravitons on flat space.

Principles of superfluid-helium gyroscopes.

Abstract: In this paper we describe how quantum coherence of superfluid helium provides a mechanism by which very small rotations can substantially modify the flow in a toroidal container. The specific modifications to that flow are discussed. For $^{4}\mathrm{He}$, we explain how the rotationally induced flow can be detected by monitoring the apparent phase-slip critical current. The rotational resolution is limited by stochastic processes related to the nucleation of phase slips. This type of superfluid-helium gyroscope (SHEG) is an analog of the rf superconducting quantum interference device (SQUID). We also show how the large coherence length of $^{3}\mathrm{He}$ can be utilized to lead to a rotationally induced interference pattern. Changes in this pattern can permit the detection of very small rotational motion. This type of SHEG is analogous to the dc SQUID. In appendixes, electrical circuits equivalent to the SHEG are described, as are certain constraints on rotational sensitivity imposed by external measuring devices.

Indirect-energy-gap dependence on Al concentration in AlxGa1-xAs alloys.

Abstract: The energy of the indirect X exciton in AlxGa1-xAs epitaxial layers (with 0.38 < x < 0.81) has been determined by photoluminescence measurements at a sample temperature of 12 K; from these data, the dependence on the Al concentration x of the energy of the GAMMA-X gap has been obtained: E(g)X(x) = 1.988 + 0.207x + 0.055x2 eV. This relation is characterized by a very small bowing. From our data we deduce also the coordinates of the GAMMA-X crossover point: x(c) = 0.385 +/- 0.016 and E(c) = 2076 +/- 4 meV at T = 0 K and x(c) = 0.396 +/- 0.016 and E(c) = 1997 +/- 4 meV at T = 300 K.

Synthesis and characterization of siloxane-titania materials

Abstract: Diethoxydimethylsilane and titanium isopropoxide have been used for preparation, via the sol-gel process, of mixed siloxane-titania materials. Depending on the Ti/Si molar ratio, flexible to brittle gels can be obtained. The rheological properties of the sols can be adjusted in order to obtain films and transparent monoliths. Due to the high amount of hydrophobic methyl groups in these materials, thick films with an average thickness of 20 μm can be obtained. These films have been used for laser dye matrices. A characterization of the various steps of the hydrolysis-condensation process as well as of the structure of the final solid gel determined by liquid and solid state NMR and TiK edge X-ray absorption spectroscopy are presented. A structural model for the gel is proposed mainly based on siloxane chains and TiO2 based particles. Optical properties of Coumarin 4 and Rhodamin 6G incorporated in such films are discussed.

Precipitation phenomena in B4C-reinforced magnesium-based composite

Abstract: The artificial aging characteristics at 177 °C of MgAlZn (AZ80A) alloy and its composite reinforced by B4C particles were compared. The morphology and the different distributions of the hardening precipitate (Mg17Al12) were observed by optical microscopy in these two materials. The aging sequence was studied by differential scanning calorimetry. This technique coupled with X-ray diffraction analyses allowed us to evaluate both the precipitation kinetics and the amount of Mg17Al12 obtained as a function of the aging time. The mechanical characteristics were evaluated by hardness measurements. The ceramic reinforcement was found to increase the aging rate of the composite with respect to that of the unreinforced alloy. Furthermore B4C particles affect the precipitation mechanism and the precipitate-strengthening distribution. In fact the heterogenous nucleation of Mg17Al12 close to the interface between ceramic and metal matrix is greatly enhanced in the composite. As a consequence the hardness increase is slightly lower for composite than for the alloy, because of the different precipitate distributions in the matrix.