Showing papers by "National University of Defense Technology published in 2002"

Efficient reinforcement learning using recursive least-squares methods

Abstract: The recursive least-squares (RLS) algorithm is one of the most well-known algorithms used in adaptive filtering, system identification and adaptive control. Its popularity is mainly due to its fast convergence speed, which is considered to be optimal in practice. In this paper, RLS methods are used to solve reinforcement learning problems, where two new reinforcement learning algorithms using linear value function approximators are proposed and analyzed. The two algorithms are called RLS-TD(λ) and Fast-AHC (Fast Adaptive Heuristic Critic), respectively. RLS-TD(λ) can be viewed as the extension of RLS-TD(0) from λ =0 to general 0≤ λ ≤1, so it is a multi-step temporal-difference (TD) learning algorithm using RLS methods. The convergence with probability one and the limit of convergence of RLS-TD(λ) are proved for ergodic Markov chains. Compared to the existing LS-TD(λ) algorithm, RLS-TD(λ) has advantages in computation and is more suitable for online learning. The effectiveness of RLS-TD(λ) is analyzed and verified by learning prediction experiments of Markov chains with a wide range of parameter settings. The Fast-AHC algorithm is derived by applying the proposed RLS-TD(λ) algorithm in the critic network of the adaptive heuristic critic method. Unlike conventional AHC algorithm, Fast-AHC makes use of RLS methods to improve the learning-prediction efficiency in the critic. Learning control experiments of the cart-pole balancing and the acrobot swing-up problems are conducted to compare the data efficiency of Fast-AHC with conventional AHC. From the experimental results, it is shown that the data efficiency of learning control can also be improved by using RLS methods in the learning-prediction process of the critic. The performance of Fast-AHC is also compared with that of the AHC method using LS-TD(λ). Furthermore, it is demonstrated in the experiments that different initial values of the variance matrix in RLS-TD(λ) are required to get better performance not only in learning prediction but also in learning control. The experimental results are analyzed based on the existing theoretical work on the transient phase of forgetting factor RLS methods.

Evaluating the Mechanical Behavior of Arterial Tissue using Digital Image Correlation

Abstract: In this study, digital image correlation (DIC) was adopted to examine the mechanical behavior of arterial tissue from bovine aorta. Rectangular sections comprised of the intimal and medial layers were excised from the descending aorta and loaded in displacement control uniaxial tension up to 40 percent elongation. Specimens of silicon rubber sheet were also prepared and served as a benchmark material in the application of DIC for the evaluation of large strains; the elastomer was loaded to 50 percent elongation. The arterial specimens exhibited a non-linear hyperelastic stress-strain response and the stiffness increased with percent elongation. Using a bilinear model to describe the uniaxial behavior, the average minor and major elastic modulii were 192±8 KPa and 912±40 KPa, respectively. Poisson's ratio of the arterial sections increased with the magnitude of axial strain; the average Poisson's ratio was 0.17±0.02. Although the correlation coefficient obtained from image correlation decreased with the percent elongation, a correlation coefficient greater than 0.8 was achieved for the tissue experiments and exceeded that obtained in the evaluation of the elastomer. Based on results from this study, DIC may serve as a valuable method for the determination of mechanical properties of arteries and other soft tissues.

Acoustic band gaps created by rotating square rods in a two-dimensional lattice

Abstract: Acoustic band gaps can be opened and tuned by rotating square rods in two-dimensional liquid sonic crystals. For the systems of mercury rods with square cross section in a water host, the width of lowest gaps increases as the rotation angle of the square rods increases. But opposite results are found for the inverse systems of water rods in mercury, where the lowest gaps narrow with an increase in the rotation angle. This gap-tuning effect becomes more evident with the filling fraction increase. Such an effect should open up a new way for designing acoustic band gaps in two-dimensional phononic crystals.

Solving the symbolic regression problem with tree-adjunct grammar guided genetic programming: the comparative results

Abstract: In this paper, we show some experimental results of tree-adjunct grammar-guided genetic programming (TAG3P) on the symbolic regression problem, a benchmark problem in genetic programming. We compare the results with genetic programming (GP) and grammar-guided genetic programming (GGGP). The results show that TAG3P significantly outperforms GP and GGGP on the target functions attempted in terms of the probability of success. Moreover, TAG3P still performed well when the structural complexity of the target function was scaled up.

Spin filtering with curve windows for interferometric fringe patterns.

Abstract: Filtering off noise from fringe patterns is important for the processing and analysis of interferometric fringe patterns. Spin filters proposed by the author have been proven to be effective in filtering off noise without blurring and distortion to fringe patterns with small fringe curvature. But spin filters still have problems processing fringe patterns with large fringe curvature. We develop a new spin filtering with curve windows that is suitable for all kinds of fringe pattern regardless of fringe curvature. The experimental results show that the new spin filtering provides optimal results in filtering off noise without distortion of the fringe features, regardless of fringe curvature.

Self-consistent average-atom scheme for electronic structure of hot and dense plasmas of mixture.

Abstract: Department of Applied Physics, National University of Defense Technology, Changsha 410073, People’s Republic of China~Received 14 April 2002; published 9 October 2002!An average-atom model is proposed to treat the electronic structures of hot and dense plasmas of mixture Itis assumed that the electron density consists of two parts The first one is a uniform distribution with a constantvalue, which is equal to the electron density at the boundaries between the atoms The second one is the totalelectron density minus the first constant distribution The volume of each kind of atom is proportional to thesum of the charges of the second electron part and of the nucleus within each atomic sphere By this way, onecan make sure that electrical neutrality is satisfied within each atomic sphere Because the integration of theelectron charge within each atom needs the size of that atom in advance, the calculation is carried out in a usualself-consistent way The occupation numbers of electron on the orbitals of each kind of atom are determined bythe Fermi-Dirac distribution with the same chemical potential for all kinds of atoms The wave functions andthe orbital energies are calculated with the Dirac-Slater equations As examples, the electronic structures of themixture of Au and Cd, water (H

Fatigue and fracture of bovine dentin

Abstract: In this paper, the fatigue and fracture properties of bovine dentin are evaluated usingin vitro experimental analyses. Double cantilever beam (DCB) specimens were prepared from bovine maxillary molars and subjected to zeroto-tension cyclic loads. The fatigue crack growth rate was evaluated as a function of the dentin tubule orientation using the Paris law. Wedge-loaded DCB specimens were also prepared and subjected to monotonic opening loads. Moire interferometry was used to acquire the in-plane displacement field during stable crack growth, and the instantaneous wedge load and crack length were acquired to evaluate the crack growth resistance and crack tip opening displacement (CTOD) with crack extension. The rate of fatigue crack growth was generally larger for crack propagation occurring perpendicular to the dentin tubules. The Moire fringe fields documented during monotonic crack growth exhibited non-linear deformation occurring within a confined region adjacent to the crack tip. Both the wedge load and CTOD response provided evidence that a fracture process zone contributes to energy dissipation during crack extension and that dentin exhibits a risingR-curve behavior. Results from this preliminary investigation are being used as a guide for an evaluation of the fatigue and fracture properties of human dentin.

Synthesis of polyaluminocarbosilane and reaction mechanism study

Abstract: Polyaluminocarbosilane (PACS) as the precursor of high-temperature resistant SiC fibers was synthesized by reacting polycarbosilane (PCS) with aluminum(III)acetylacetonate [Al(AcAc)3] at 310°C in N2 under ambient pressure. The reaction mechanism and the structure of PACS were investigated in detail by FTIR, GPC, GC/MS, ESCA, and elemental analysis. The reaction was proven complex involving the formation of SiOAl and SiAlSi bonds, which were accompanied by the evolution of 3-methoxy-2,2-dimethyl-oxirane, 2,3-dihydro-[1,4]dioxine, pent-3-en-2-one, and 3-ethyl-but-3-en-2-one, and acetylacetone. © 2002 Wiley Periodicals, Inc. J Appl Polym Sci 85: 2787–2792, 2002

FDTD Analysis of Electromagnetic Reflection by Conductive Plane Covered with Magnetized Inhomogeneous Plasmas

Abstract: A novel finite-difference time-domain (FDTD) methodology which incorporates both anisotropy and frequency dispersion at the same time is developed for electromagnetic wave propagation in anisotropic magnetoactive plasmas in this paper. The numerical verification of the method are confirmed by computing the reflection and transmission of right-handed/left-handed circularly polarized (RCP/LCP) wave through a magnetized plasma layer, with the direction of propagation parallel to the direction of the biasing field. And, the right-handed / left-handed polarized wave reflection coefficients for electromagnetic signals normally incident upon a conductive plane covered with a layer of magnetized plasma are computed using the new FDTD method. The parabolic electron-number density profile varies only in the direction perpendicular to the plane. The function dependence of reflection coefficients on the number density, collision frequency and external magnetic field is studied.

Detailed-term-accounting approximation calculations of the radiative opacity of aluminum plasmas: a systematic study.

Abstract: The spectrally resolved radiative opacity and the Rosseland and Planck mean opacities are calculated by using the detailed-term-accounting approximation for aluminum plasmas with varieties of density and temperature. The results are presented along a 40 eV isothermal sequence, a $0.01 {\mathrm{g}/\mathrm{c}\mathrm{m}}^{3}$ isodense sequence, and a sequence with average ionization degree $Z*\ensuremath{\sim}7.13.$ Particular attention is given to the influence of the detailed treatment of spectral lines on the Rosseland mean opacity under different thermodynamical conditions. The results show that at densities of $0.004 {\mathrm{g}/\mathrm{c}\mathrm{m}}^{3}$ and higher, the opacities are not very sensitive to the spectral linewidth within a reasonable range. As examples, the Rosseland mean opacity, which is most sensitive to the detailed linewidth, at 40 eV and $0.004 {\mathrm{g}/\mathrm{c}\mathrm{m}}^{3}$ changes no more than 15%, when we change the electron impact spectral linewidth artificially by reducing it by 50% or increasing it twice, and at 40 eV and $0.1 {\mathrm{g}/\mathrm{c}\mathrm{m}}^{3}$ it changes less than 5%. For comparison, we also carried out calculations by using an average atom model. For the Rosseland mean opacities, the two models show quite large differences, in particular at low densities, while for the Planck mean opacities the results of the two models are much closer.

A novel fractal microstrip PBG structure

Abstract: A novel fractal microstrip PBG structure is proposed. It is realized by replacing the etched rectangular holes in a ground plane with a Sierpinski carpet. A dual-stopband characteristic not related to harmonics is observed. Different numbers of iterations of Sierpinski carpet are studied experimentally. © 2002 John Wiley & Sons, Inc. Microwave Opt Technol Lett 32: 136–138, 2002.

Mathematical representation of a chinese character and its applications

Abstract: In this paper, a novel method to express Chinese characters mathematically is presented based on the knowledge of the structure of Chinese characters. Each Chinese character can be denoted by a mathematical expression in which the operands are components of Chinese characters and the operators are the location relations between the components. Five hundred five components are selected and 6 operators are defined to express all the Chinese characters successfully. These mathematical expressions of Chinese characters are simple, natural, and can be operated like the common mathematical expression of numbers. It makes Chinese information processing much simpler than before. This theory has been applied successfully in fonts automation, Chinese information transmission among different platforms and different operating systems on Internet, and knowledge discovery of the structure of Chinese characters. It can also be applied extensively to many areas such as typesetting, advertising, packing design, virtual library, network transmission, pattern recognition and Chinese mobile communication.

Structural evolution and associated properties on conversion from Si-C-O-Al ceramic fibers to Si-C-Al fibers by sintering

Abstract: Ceramic fibers with the Si-C-O-Al composition were prepared by melt-spinning of polyaluminocarbosilane (PACS), initial curing in air and then thermal curing, before finally being pyrolyzed at 1300 °C. The ceramic fibers so-obtained were sintered at 1800 °C in argon to produce Si-C-Al ceramic fibers as a super high temperature-resistant reinforcement. When the Si-C-O-Al fibers were sintered to the Si-C-Al ceramic fibers, the structural evolution and the associated properties were studied in comparison to PCS-derived Si-C-O ceramic fibers with respect to tensile strength, creep resistance, electrical resistivity, morphology, and crystalline grain size using 29Si MAS NMR, 13C MAS NMR, AES, XRD and SEM. The Si-C-O-Al ceramic fibers had low creep resistance because they contained high levels of silicon oxycarbide, poorly organized SiC crystalline grains at the nanometer level and high levels of free carbon. The electrical resistivity of the Si-C-O-Al ceramic fibers was high due to the high level of silicon oxycarbide. However, the Si-C-Al ceramic fibers were strongly resistant to creep due to a silicon oxycarbide free structure, a well-organized and crystallized SiC content and a lower content of free carbon. The absence of silicon oxycarbide in the Si-C-Al ceramic fibers is responsible for their low electrical resistivity and high tensile strength retention at high temperature.

Preparation of hybrid polymer as a near-stoichiometric SiC precursor by blending of polycarbosilane and polymethysilane

Abstract: A hybrid polymer as a near-stoichiometric SiC precursor was prepared by simply blending polycarbosilane (PCS) as a carbon-rich SiC precursor and polyborazine-modified polymethylsilane (M-PMS) as a silicon-rich SiC precursor at an optimized ratio of 0.25∶1.00 (PCS∶M-PMS) by weight. The hybrid precursor showed a very high ceramic yield of 82 wt.% and higher molecular weight than that obtained using a single precursor. The excellent processibility of the hybrid precursor was proven by dry-spinning it into green fibers. Based on XRD, 13C MAS-NMR and 29Si MAS-NMR measurements, the SiC ceramic derived from the hybrid precursor at 1400 °C showed no excess Si and little residual carbon, and a chemical composition of 98.08 wt.% SiC and 1.92 wt.% SiO2 by SEM-EDX analysis.

Joint estimation of Doppler and time-difference-of-arrival exploiting cyclostationary property

Abstract: For the problem of estimating the time-difference-of-arrival (TDOA) of radio waves impinging on one or a pair of antennas for the purpose of actively or passively locating the source of a communications or telemetry signal in the presence of interfering signals and noise, W A Gardner has developed some effective methods, such as CCCC and SPECCOA etc However, it is shown that CCCC, SPECCOA and related methods do not apply to moving targets because the Doppler shift changes the cycle frequencies of the signal of interest Two new methods based on the cyclic cross-ambiguity (CCA) function and the spectrally cyclic cross-ambiguity (SCCA) function are introduced in the paper for the joint TDOA and Doppler estimation, and their consistency in estimating the TDOA and Doppler is also discussed Practical experiments, utilising FM radio signals emitting from a remote FM transmitter to detect the planes and estimate the values of TDOA and Doppler, have been carried out Experimental results demonstrate the efficiency of the CCA and SCCA methods in the estimation of TDOA and Doppler

Improved analysis for the baryon masses to order Λ QCD / m Q from QCD sum rules

Abstract: We use the QCD sum rule approach to calculate the masses of the Lambda_Q and Sigma_Q baryons to the Lambda_{QCD}/m_Q order within the framework of heavy quark effective theory. We compare the direct approach and the covariant approach to this problem. Two forms of currents have been adopted in our calculation and their effects on the results are discussed. Numerical results obtained in both direct and covariant approaches are presented. The splitting between spin 1/2 and 3/2 doublets derived from our calculation is {\Sigma_Q^*}^2-\Sigma_Q^2\simeq 0.35\pm 0.03 GeV^2 which is in good agreement with the experiment.

Residual-gradient-based neural reinforcement learning for the optimal control of an acrobot

Abstract: Based on the idea of dynamic programming, reinforcement learning (RL) has become an important model-free method to solve difficult optimal control problems. In this paper, a novel neural RL method is proposed to solve the time-optimal control problem of a class of under-actuated robots, which is called the acrobot. The RL method uses a modified residual gradient reinforcement learning algorithm called RGNP (residual gradient with nonstationary policy). The RGNP algorithm not only has guaranteed convergence under certain conditions but also can ensure the performance of the approximated optimal policy, which is superior to the previous residual gradient algorithms. Simulation results of the learning control of the acrobot illustrate the effectiveness of the proposed method.

A novel PBG structure for microstrip lines

Abstract: A new photonic bandgap (PBG) structure for microstrip lines is proposed. Different from conventional PBG microstrip lines, the periodic patterns are etched in the conductor line of the microstrip in this structure. Simulations and measurements show that the bandgap can be achieved using this method. Compared with traditional PBG microstrip structures, this new transmission line can be installed in a metal shield box without performance loss.

Acceleration element for femtosecond electron pulse compression.

Abstract: An acceleration element is proposed for compressing the electron pulse duration in a femtosecond photoelectron gun. The element is a compact metal cavity with curved-shaped walls. An external voltage is applied to the cavity where a special electric field forms in such a way that the slow electrons in the electron pulse front are accelerated more than the fast electrons, and consequently the electron pulse duration will be compressed. The distribution of the electric field inside the acceleration cavity is analyzed for the geometry of the cavity. The electron dynamics in this acceleration cavity is also investigated numerically. Numerical results show that the electron pulse front and pulse duration can be improved by compensating for the effects of space charge and the initial energy spread of photoelectrons with a Lambertian angular distribution. Depending on the design parameters and the shape of the electron pulse, for a femtosecond electron gun with an electron energy of 30 keV, ${10}^{3}$ electrons per pulse, and an electron drift length of 40 cm, the electron pulse duration can be reduced from 550 to 200 fs when using a compensating cavity with an average radius of 1.7 and 5.6 cm in length. Electron pulses shorter than 200 fs can be achieved if the length of the drift region is reduced.

KF-Diff+: Highly Efficient Change Detection Algorithm for XML Documents

Abstract: Most previous work in change detection on XML documents used the ordered tree, with the best complexity of O(nlogn), where n is the size of the document. The best algorithm we had ever known for unordered model achieves polynomial time in complexity. In this paper, we propose a highly efficient algorithm named KF-Diff+. The key property of our algorithm is that the algorithm transforms the traditional tree-to-tree correction into the comparing of the key trees which are substantially label trees without duplicate paths with the complexity of O(n), where n is the number of nodes in the trees. In addition, KF-Diff+ is tailored to both ordered trees and unordered trees. Experiment shows that KF-Diff+ can handle XML documents at extreme speed.

Multi-model SAR image despeckling

Abstract: A multi-model despeckling approach for SAR image is presented. The chi-squared test is used to segment the image into homogeneous and heterogeneous regions. Then, the heterogeneous regions are separated into subregions, each of which consists of the points with same edge orientations. Homogeneous regions and the separated subregions are despeckled according to their characteristics. Experimental results are reported.

Spin filtering with curve windows for interferometric fringe patterns

Abstract: Filtering off noise form fringe patterns is important for processing and analysis of interferometric fringe patterns. Spin filtering proposed by the author has been proven to be effective in filtering off noise without blurring and distortion to the fringe patterns with small fringe curvature. But spin filters still have problems for the fringe patterns with large fringe curvature. In this paper we develop a new spin filtering with curve windows to be suitable for all kinds of fringe patterns regardless of fringe curvature. The curve windows in the filtering coincide with the local fringe contours of fringe gray levels. The experimental result show that the new spin filtering provides the optimal results in both filtering off noise and remaining fringe feature.© (2002) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

New family of lapped biorthogonal transform via lifting steps

Abstract: By scaling all discrete cosine transform (DCT) intermediate output coefficients of the lapped transform and employing the type-II and type-IV DCT based on lifting steps, a new family of lapped biorthogonal transform is introduced, called the IntLBT. When all the elements with a floating point of each lifting matrix in the IntLBT are approximated by binary fractions, the IntLBT is implemented by a series of dyadic lifting steps and provides very fast, efficient in-place computation of the transform coefficients, and all internal nodes have finite precision. When each lifting step in the IntLBT is implemented using the same nonlinear operations as those used in the well known integer-to-integer wavelet transform, the IntLBT maps integers to integers, so it can express lossless image information. As an application of the novel IntLBT to lossy image compression, simulation results demonstrate that the IntLBT has significantly less blocking artefacts, higher peak signal-to-noise ratio, and better visual quality than the DCT. More importantly, the IntLBT's coding performance is approximately the same as that of the much more complex Cohen-Daubechies-Feauveau (CDF) 9/7-tap biorthogonal wavelet with floating-point coefficients, and in some cases even surpasses that of the CDF 9/7-tap biorthogonal wavelet.

Photoionization of Be-like neon (Ne VII) from the low-lying states: Energies, widths, branching ratios, and oscillator strengths of the 1s-2p resonances

Abstract: Seventeen-state close-coupling calculations are performed on the photoionization near the 1 s-2 p resonance region from the terms belonging to the configurations 1s 2 2s 2 ,1 s 2 2s2p, and 1s 2 2p 2 , of Be-like neon ions, Ne VII. The calculations include the total and partial photoionization cross sections and the contributions of the main ionization channels to the partial cross sections. The resonance energies, widths, and branching ratios of the 1s-2 p core-excited states are determined from these cross sections. The resonance oscillator strengths of the corresponding 1s-2 p transitions from Ne VII low-lying states are also obtained by integrating the photoionization cross sections. Our theoretical resonance energies, widths, and branching ratios are compared with existing experiments on the Auger spectra and other theoretical results. The calculated resonance energies are in rather good agreement with the latest experiment on the Auger spectra. For the autoionization width, good agreement is also found with recent theoretical results wherever available obtained using a saddle-point complex-rotation method.

General analytical solution of transverse vibration for orthotropic rectangular thin plates

Abstract: A general solution of differential equation for transverse displacement function of orthotropic rectangular thin plates in free vibration is established in this paper. It can be used to solve the vibration problem of plate with arbitrary boundaries. As an example, the frequencies of a composite laminated plate with four free edges have been solved. The result as compared with the experiment is satisfactory.

Necessary and sufficient condition for distillability with unit fidelity from finite copies of a mixed state: The most efficient purification protocol

Abstract: It is well known that any entangled mixed state in $2\ensuremath{\bigotimes}2$ systems can be purified via infinite copies of the mixed state. But, can one distill a pure maximally entangled state from finite copies of a mixed state in any bipartite system by local operation and classical communication? This is more meaningful in practical application. We give a necessary and sufficient condition for this distillability. This condition requires that there exist distillable subspaces. According to this condition, one can judge easily whether a mixed state is distillable or not. We also analyze some properties of distillable subspaces, and discuss the most efficient purification protocol. Finally, we discuss the distillable enanglement of a two-qubit system for the case of finite copies.