Sharp estimates for eigenvalues of integral operators generated by dot product kernels on the sphere

We obtain explicit formulas for the eigenvalues of integral operators generated by continuous dot product kernels defined on the sphere via the usual gamma function. Using them, we present both, a procedure to describe sharp bounds for the eigenvalues and their asymptotic behavior near 0. We illustrate our results with examples, among them the integral operator generated by a Gaussian kernel. Finally, we sketch complex versions of our results to cover the cases when the sphere sits in a Hermitian space.

Calculating the tetrahedral bond angle using spherical polars and the dot product

A proof using the methane tetrahedroid bond angle can be obtained by using spherical polar coordinates to calculate the Cartesian coordinates of the hydrogen atoms, then using the dot product of vectors.

Spectral Clustering with Jensen-type kernels and their multi-point extensions

Motivated by multi-distribution divergences, which originate in information theory, we propose a notion of `multipoint' kernels, and study their applications. We study a class of kernels based on Jensen type divergences and show that these can be extended to measure similarity among multiple points. We study tensor flattening methods and develop a multi-point (kernel) spectral clustering (MSC) method. We further emphasize on a special case of the proposed kernels, which is a multi-point extension of the linear (dot-product) kernel and show the existence of cubic time tensor flattening algorithm in this case. Finally, we illustrate the usefulness of our contributions using standard data sets and image segmentation tasks.

Hip power analysis in individuals with transfemoral amputation: A different strategy different from stabilisation during gait stance

Introduction and Objectives Joint moments and joint powers during gait are widely used to determine the effects of rehabilitation programs as well as prosthetic fitting. Following the definition of power (dot product of joint moment and joint angular velocity) it has been previously proposed to analyse the 3D angle between both vectors, αMw. Basically, joint power is maximised when both vectors are parallel and cancelled when both vectors are orthogonal. In other words, αMw < 60° reveals a propulsion configuration (more than 50% of the moment contribute to positive power) while αMw > 120° reveals a resistance configuration (more than 50% of the moment contribute to negative power). A stabilisation configuration (less than 50% of the moment contribute to power) corresponds to 60° < αMw < 120°. Previous studies demonstrated that hip joints of able-bodied adults (AB) are mainly in a stabilisation configuration (αMw about 90°) during the stance phase of gait. [1, 2] Individuals with transfemoral amputation (TFA) need to maximise joint power at the hip while controlling the prosthetic knee during stance. Therefore, we tested the hypothesis that TFAs should adopt a strategy that is different from a continuous stabilisation. The objective of this study was to compute joint power and αMw for TFA and to compare them with AB. Methods Three trials of walking at self-selected speed were analysed for 8 TFAs (7 males and 1 female, 46±10 years old, 1.78±0.08 m 82±13 kg) and 8 ABs (males, 25±3 years old, 1.75±0.04, m 67±6 kg). The joint moments are computed from a motion analysis system (Qualisys, Goteborg, Sweden) and a multi-axial transducer (JR3, Woodland, USA) mounted above the prosthetic knee for TFAs and from a motion analysis system (Motion Analysis, Santa Rosa, USA) and force plates (Bertec, Columbus, USA) for ABs. The TFAs were fitted with an OPRA (Integrum, AB, Gothengurg, Sweden) osseointegrated implant system and their prosthetic designs include pneumatic, hydraulic and microprocessor knees. Previous studies showed that the inverse dynamics computed from the multi-axial transducer is the proper method considering the absorption at the foot and resistance at the knee. Results The peak of positive power at loading response (H1) was earlier and lower for TFA compared to AB. Although the joint power is lower, the 3D angle between joint moment and joint angular velocity, αMw, reveals an obvious propulsion configuration (mean αMw about 20°) for TFA compared to a stabilisation configuration (mean αMw about 70°) for AB. The peaks of negative power at midstance (H2) and of positive power at preswing / initial swing (H3) occurred later, lower and longer for TFA compared to AB. Again, the joint powers are lower for TFA but, in this case, αMw is almost comparable (with a time lag), demonstrating a stabilisation (almost a resistance for TFA, mean αMw about 120°) and a propulsion configuration, respectively. The swing phase is not analysed in the present study. Conclusion The analysis of hip joint power may indicate that TFAs demonstrated less propulsion and resistance than ABs during the stance phase of gait. This is true from a quantitative point of view. On the contrary, the 3D angle between joint moment and joint angular velocity, αMw, reveals that TFAs have a remarkable propulsion strategy at loading response and almost a resistance strategy at midstance while ABs adopted a stabilisation strategy. The propulsion configuration, with αMw close to 0°, seems to aim at maximising the positive joint power. The configuration close to resistance, with αMw far from 180°, might aim at unlocking the prosthetic knee before swing while minimising the negative power. This analysis of both joint power and 3D angle between the joint moment and the joint angular velocity provides complementary insights into the gait strategies of TFA that can be used to support evidence-based rehabilitation and fitting of prosthetic components.

Restructuring Sparse High Dimensional Data for Effective Retrieval

The task in text retrieval is to find the subset of a collection of documents relevant to a user's information request, usually expressed as a set of words. Classically, documents and queries are represented as vectors of word counts. In its simplest form, relevance is defined to be the dot product between a document and a query vector--a measure of the number of common terms. A central difficulty in text retrieval is that the presence or absence of a word is not sufficient to determine relevance to a query. Linear dimensionality reduction has been proposed as a technique for extracting underlying structure from the document collection. In some domains (such as vision) dimensionality reduction reduces computational complexity. In text retrieval it is more often used to improve retrieval performance. We propose an alternative and novel technique that produces sparse representations constructed from sets of highly-related words. Documents and queries are represented by their distance to these sets. and relevance is measured by the number of common clusters. This technique significantly improves retrieval performance, is efficient to compute and shares properties with the optimal linear projection operator and the independent components of documents.

On classical and other methods of discriminant analysis and estimation of log-odds

For two multinormal populations with equal covariance matrices the likelihood ratio discriminant function, an alternative allocation rule to the sample linear discriminant function when n1 ≠ n2 ,is studied analytically. With the assumption of a known covariance matrix its distribution is derived and the expectation of its actual and apparent error rates evaluated and compared with those of the sample linear discriminant function. This comparison indicates that the likelihood ratio allocation rule is robust to unequal sample sizes. The quadratic discriminant function is studied, its distribution reviewed and evaluation of its probabilities of misclassification discussed. For known covariance matrices the distribution of the sample quadratic discriminant function is derived. When the known covariance matrices are proportional exact expressions for the expectation of its actual and apparent error rates are obtained and evaluated. The effectiveness of the sample linear discriminant function for this case is also considered. Estimation of true log-odds for two multinormal populations with equal or unequal covariance matrices is studied. The estimative, Bayesian predictive and a kernel method are compared by evaluating their biases and mean square errors. Some algebraic expressions for these quantities are derived. With equal covariance matrices the predictive method is preferable. Where it derives this superiority is investigated by considering its performance for various levels of fixed true log-odds. It is also shown that the predictive method is sensitive to n1 ≠ n2. For unequal but proportional covariance matrices the unbiased estimative method is preferred. Product Normal kernel density estimates are used to give a kernel estimator of true log-odds. The effect of correlation in the variables with product kernels is considered. With equal covariance matrices the kernel and parametric estimators are compared by simulation. For moderately correlated variables and large dimension sizes the product kernel method is a good estimator of true log-odds.

Interpolação de dados de campo potencial através da camada equivalente

O uso da técnica da camada equivalente na interpolação de dados de campo potencial permite levar em consideração que a anomalia, gravimétrica ou magnética, a ser interpolada é uma função harmônica. Entretanto, esta técnica tem aplicação computacional restrita aos levantamentos com pequeno número de dados, uma vez que ela exige a solução de um problema de mínimos quadrados com ordem igual a este número. Para viabilizar a aplicação da técnica da camada equivalente aos levantamentos com grande número de dados, nós desenvolvemos o conceito de observações equivalentes e o método EGTG, que, respectivamente, diminui a demanda em memória do computador e otimiza as avaliações dos produtos internos inerentes à solução dos problemas de mínimos quadrados. Basicamente, o conceito de observações equivalentes consiste em selecionar algumas observações, entre todas as observações originais, tais que o ajuste por mínimos quadrados, que ajusta as observações selecionadas, ajusta automaticamente (dentro de um critério de tolerância pré-estabelecido) todas as demais que não foram escolhidas. As observações selecionadas são denominadas observações equivalentes e as restantes são denominadas observações redundantes. Isto corresponde a partir o sistema linear original em dois sistemas lineares com ordens menores. O primeiro com apenas as observações equivalentes e o segundo apenas com as observações redundantes, de tal forma que a solução de mínimos quadrados, obtida a partir do primeiro sistema linear, é também a solução do segundo sistema. Este procedimento possibilita ajustar todos os dados amostrados usando apenas as observações equivalentes (e não todas as observações originais) o que reduz a quantidade de operações e a utilização de memória pelo computador. O método EGTG consiste, primeiramente, em identificar o produto interno como sendo uma integração discreta de uma integral analítica conhecida e, em seguida, em substituir a integração discreta pela avaliação do resultado da integral analítica. Este método deve ser aplicado quando a avaliação da integral analítica exigir menor quantidade de cálculos do que a exigida para computar a avaliação da integral discreta. Para determinar as observações equivalentes, nós desenvolvemos dois algoritmos iterativos denominados DOE e DOEg. O primeiro algoritmo identifica as observações equivalentes do sistema linear como um todo, enquanto que o segundo as identifica em subsistemas disjuntos do sistema linear original. Cada iteração do algoritmo DOEg consiste de uma aplicação do algoritmo DOE em uma partição do sistema linear original. Na interpolação, o algoritmo DOE fornece uma superfície interpoladora que ajusta todos os dados permitindo a interpolação na forma global. O algoritmo DOEg, por outro lado, otimiza a interpolação na forma local uma vez que ele emprega somente as observações equivalentes, em contraste com os algoritmos existentes para a interpolação local que empregam todas as observações. Os métodos de interpolação utilizando a técnica da camada equivalente e o método da mínima curvatura foram comparados quanto às suas capacidades de recuperar os valores verdadeiros da anomalia durante o processo de interpolação. Os testes utilizaram dados sintéticos (produzidos por modelos de fontes prismáticas) a partir dos quais os valores interpolados sobre a malha regular foram obtidos. Estes valores interpolados foram comparados com os valores teóricos, calculados a partir do modelo de fontes sobre a mesma malha, permitindo avaliar a eficiência do método de interpolação em recuperar os verdadeiros valores da anomalia. Em todos os testes realizados o método da camada equivalente recuperou mais fielmente o valor verdadeiro da anomalia do que o método da mínima curvatura. Particularmente em situações de sub-amostragem, o método da mínima curvatura se mostrou incapaz de recuperar o valor verdadeiro da anomalia nos lugares em que ela apresentou curvaturas mais pronunciadas. Para dados adquiridos em níveis diferentes o método da mínima curvatura apresentou o seu pior desempenho, ao contrário do método da camada equivalente que realizou, simultaneamente, a interpolação e o nivelamento. Utilizando o algoritmo DOE foi possível aplicar a técnica da camada equivalente na interpolação (na forma global) dos 3137 dados de anomalia ar-livre de parte do levantamento marinho Equant-2 e 4941 dados de anomalia magnética de campo total de parte do levantamento aeromagnético Carauari-Norte. Os números de observações equivalentes identificados em cada caso foram, respectivamente, iguais a 294 e 299. Utilizando o algoritmo DOEg nós otimizamos a interpolação (na forma local) da totalidade dos dados de ambos os levantamentos citados. Todas as interpolações realizadas não seriam possíveis sem a aplicação do conceito de observações equivalentes. A proporção entre o tempo de CPU (rodando os programas no mesmo espaço de memória) gasto pelo método da mínima curvatura e pela camada equivalente (interpolação global) foi de 1:31. Esta razão para a interpolação local foi praticamente de 1:1.

About the determination of key of a musical excerpt

Knowledge of the key of a musical passage is a pre-requisite for all the analyses that require functional labelling. In the past, people from either a musical or AI background have tended to solve the problem by means of implementing a computerized version of musical analysis. Previous attempts are discussed and then attention is focused on a non-analytical solution first reported by J.A.Gabura. A practical way to carry it out is discussed as well as its limitations in relation to examples. References are made to the MusicXML format as needed. © Springer-Verlag Berlin Heidelberg 2006.

Practical Stability and Vector-Lyapunov Functions for Impulsive Differential Equations with "Supremum"

Stability of nonlinear impulsive differential equations with "supremum" is studied. A special type of stability, combining two different measures and a dot product on a cone, is defined. Perturbing cone-valued piecewise continuous Lyapunov functions have been applied. Method of Razumikhin as well as comparison method for scalar impulsive ordinary differential equations have been employed.

Tryst with a molecular and supramolecular oddity: an anti-Furst-Plattner epoxide opening product as a designer additive for accessing an elusive polymorph

Additive induced polymorphism of a conformationally locked tetraacetate 3 in presence of its diastereomer 4 is described. The ester 3 was specially crafted on a trans-decalin backbone to relegate the O-H center dot center dot center dot O H-bond donors to the molecular interior and have the peripheral H-bond acceptors in 1,3-syndiaxial relationship. The supramolecular assembly of 3 was destined to evolve along two mutually exclusive pathways, namely one, which employs intermolecular O-H center dot center dot center dot O H-bonds (pathway 1) and the other that sacrifices these for intramolecular O-H center dot center dot center dot O H-bonds and settles for a crystal packing dictated by weak intermolecular interactions alone (pathway 2). Exploiting the similarity between the self-assemblies of 4 and the two recently reported dimorphs of 3, the ester 3 has been stimulated to follow the elusive non-hierarchical pathway 2 through preferential inhibition of pathway 1. Interestingly, the inhibitor 4 was obtained serendipitously en route 3 via an apparent breakdown of Furst-Plattner rule.

Preparation of the layered double hydroxide (LDH) LiAl2(OH)(7)center dot 2H(2)O, by gel to crystallite conversion and a hydrothermal method, and its conversion to lithium aluminates

A layered double hydroxide (LDH) with chemical composition LiAl2(OH)(7) . 2H(2)O was prepared via a wet chemical route of gel to crystallite (G-C) conversion at 80 degrees C involving the reaction of hydrated alumina gel, Al2O3.yH(2)O (80 < y < 120) with LiOH (Li2O/Al2O3 greater than or equal to 0.5) in presence of hydrophilic solvents such as ethanol under refluxing conditions. The hydrothermal synthesis was carried out using the same reactants by heating to less than or equal to 140 degrees C in a Teflon-lined autoclave under autogenerated pressure (less than or equal to 20 MPa). Transmission electron microscopy showed needle-shaped aggregates of size 0.04-0.1 mu m for the gel to crystallite conversion product, whereas the hydrothermal products consisted of individual lamellar crystallites of size 0.2-0.5 mu m with hexagonal morphology. The LDH prepared through the gel to crystallite conversion could be converted into LiAl(OH)(4) . H2O or LiAl(OH)(3)NO3 . H2O by imbibition of LiOH or LiNO3, respectively, under hydrothermal conditions. Thermal decomposition of LDH above 1400 degrees C gave rise to LiAl5O8 accompanied by the evaporation of Li2O. LiAl(OH)(4) . H2O and LiAl(OH)(3)NO3 . H2O decomposed in the temperature range 400-1000 degrees C to alpha- or beta-LiAlO2. The compositional dependence of the product, the intermediate phases formed during the heat treatment and the possible reactions involved are described in detail.

Efficient algorithms for learning kernels from multiple similarity matrices with general convex loss functions

In this paper we consider the problem of learning an n × n kernel matrix from m(1) similarity matrices under general convex loss. Past research have extensively studied the m = 1 case and have derived several algorithms which require sophisticated techniques like ACCP, SOCP, etc. The existing algorithms do not apply if one uses arbitrary losses and often can not handle m > 1 case. We present several provably convergent iterative algorithms, where each iteration requires either an SVM or a Multiple Kernel Learning (MKL) solver for m > 1 case. One of the major contributions of the paper is to extend the well knownMirror Descent(MD) framework to handle Cartesian product of psd matrices. This novel extension leads to an algorithm, called EMKL, which solves the problem in O(m2 log n 2) iterations; in each iteration one solves an MKL involving m kernels and m eigen-decomposition of n × n matrices. By suitably defining a restriction on the objective function, a faster version of EMKL is proposed, called REKL,which avoids the eigen-decomposition. An alternative to both EMKL and REKL is also suggested which requires only an SVMsolver. Experimental results on real world protein data set involving several similarity matrices illustrate the efficacy of the proposed algorithms.

Boxicity and cubicity of product graphs

The boxicity (cubicity) of a graph G is the minimum natural number k such that G can be represented as an intersection graph of axis-parallel rectangular boxes (axis-parallel unit cubes) in R-k. In this article, we give estimates on the boxicity and the cubicity of Cartesian, strong and direct products of graphs in terms of invariants of the component graphs. In particular, we study the growth, as a function of d, of the boxicity and the cubicity of the dth power of a graph with respect to the three products. Among others, we show a surprising result that the boxicity and the cubicity of the dth Cartesian power of any given finite graph is, respectively, in O(log d/ log log d) and circle dot(d/ log d). On the other hand, we show that there cannot exist any sublinear bound on the growth of the boxicity of powers of a general graph with respect to strong and direct products. (C) 2015 Elsevier Ltd. All rights reserved.

Colliding-pulse modelocked quantum dot lasers

Colliding pulse modelocking is demonstrated for the first time in quantum dot lasers. Using 3.9 mm-long devices with a 245 pm-long central absorber, 7 ps pulses at a repetition rate of 20 GHz is obtained. For Gaussian pulses a time-bandwidth product close to the Fourier transform limit is determined. These results confirm the potential of quantum dot lasers for high repetition rate harmonic modelocking.

Investigating the Mechanism and Electrode Kinetics of the Oxygen vertical bar Superoxide (O-2 vertical bar O-2(center dot-)) Couple in Various Room-Temperature Ionic Liquids at Gold and Platinum Electrodes in the Temperature Range 298-318 K

The reduction of oxygen was studied over a range of temperatures (298-318 K) in n-hexyltriethylammonium bis(trifluoromethanesulfonyl)imide, [N-6,N-2,N-2,N-2][NTf2], and 1-butyl-2,3-methylimidazolium bis(trifluoromethanesulfonyl)imide, [C(4)dmim][NTf2] on both gold and platinum microdisk electrodes, and the mechanism and electrode kinetics of the reaction investigated. Three different models were used to simulate the CVs, based on a simple electron transfer ('E'), an electron transfer coupled with a reversible homogeneous chemical step ('ECrev') and an electron transfer followed by adsorption of the reduction product ('EC(ads)'), and where appropriate, best fit parameters deduced, including the heterogeneous rate constant, formal electrode potential, transfer coefficient, and homogeneous rate constants for the ECrev mechanism, and adsorption/desorption rate constants for the EC(ads) mechanism. It was concluded from the good simulation fits on gold that a simple E process operates for the reduction of oxygen in [N-6,N-2,N-2,N-2][NTf2], and an ECrev process for [C(4)dmim][NTf2], with the chemical step involving the reversible formation of the O-2(center dot-)center dot center dot center dot [C(4)dmim](+) ion-pair. The E mechanism was found to loosely describe the reduction of oxygen in [N-6,N-2,N-2,N-2][NTf2] on platinum as the simulation fits were reasonable although not perfect, especially for the reverse wave. The electrochemical kinetics are slower on Pt, and observed broadening of the oxidation peak is likely due to the adsorption of superoxide on the electrode surface in a process more complex than simple Langmuirian. In [C(4)dmim][NTf2] the O-2(center dot-) predominantly ion-pairs with the solvent rather than adsorbs on the surface, and an ECrev quantitatively describes the reduction of oxygen on Pt also.