Supervised Gaussian Process Latent Variable Model for Dimensionality Reduction

The Gaussian process latent variable model (GP-LVM) has been identified to be an effective probabilistic approach for dimensionality reduction because it can obtain a low-dimensional manifold of a data set in an unsupervised fashion. Consequently, the GP-LVM is insufficient for supervised learning tasks (e. g., classification and regression) because it ignores the class label information for dimensionality reduction. In this paper, a supervised GP-LVM is developed for supervised learning tasks, and the maximum a posteriori algorithm is introduced to estimate positions of all samples in the latent variable space. We present experimental evidences suggesting that the supervised GP-LVM is able to use the class label information effectively, and thus, it outperforms the GP-LVM and the discriminative extension of the GP-LVM consistently. The comparison with some supervised classification methods, such as Gaussian process classification and support vector machines, is also given to illustrate the advantage of the proposed method.

Effect of electron-electron scattering on spin dephasing in a high-mobility low-density two-dimensional electron gas

By utilizing time-resolved Kerr rotation techniques, we have investigated the spin dynamics of a high-mobility low density two-dimensional electron gas in a GaAs/Al0.35Ga0.65As heterostructure in the dependence on temperature from 1.5 to 30 K. It is found that the spin relaxation/dephasing time under a magnetic field of 0.5 T exhibits a maximum of 3.12 ns around 14 K, which is superimposed on an increasing background with rising temperature. The appearance of the maximum is ascribed to that at the temperature where the crossover from the degenerate to the nondegenerate regime takes place, electron-electron Coulomb scattering becomes strongest, and thus inhomogeneous precession broadening due to the D'yakonov-Perel' mechanism becomes weakest. These results agree with the recent theoretical predictions [J. Zhou et al., Phys. Rev. B 15, 045305 (2007)], which verify the importance of electron-electron Coulomb scattering to electron spin relaxation/dephasing.

Spin dynamics of electrons in the first excited subband of a high-mobility low-density two-dimensional electron system

We report on time-resolved Kerr rotation measurements of spin coherence of electrons in the first excited subband of a high-mobility low-density two-dimensional electron system in a GaAs/Al0.35Ga0.65As heterostructure. While the transverse spin lifetime (T-2(*)) of electrons decreases monotonically with increasing magnetic field, it has a nonmonotonic dependence on the temperature and reaches a peak value of 596 ps at 36 K, indicating the effect of intersubband electron-electron scattering on the electron-spin relaxation.

Dimensional crossover of heat conduction in low dimensions

The dimensional crossover phenomena of heat conduction is studied by a two-dimensional (2D) Fermi-Pasta-Ulam lattice. The 2D divergence law of the thermal conductivity is confirmed by the simulations results. The divergence law of the thermal conductivity will change from the 2D class to 1D class as delta=N-y/N-x decreases, here N-y is the size in transverse direction and N-x in longitude direction. The simulation's results suggest that the dimensional crossover happens in delta(*)-> 0 as N-x ->infinity.

The cellular structure of a two-dimensional H2/O2/Ar detonation wave

In this paper, the cellular structure of a two-dimensional detonation wave in a low pressure H2/O2/Ar mixture calculated with a detailed chemical reaction model, high order scheme and high resolution grids is investigated. The regular cellular structure is produced about 1 ms after introducing perturbations in the reaction zone of a steady one-dimensional detonation wave. It is found from the present resolution study that the discrepancies concerning the structure type arising from the coarser grid employed can be resolved using a sufficiently fine grid size of 0.05 mm and below and shows a double-Mach-like strong-type configuration. During the structure evolution process, the structure configuration does not change much in the periods before and after the triple point collision. Through the triple point collision, three regular collision processes are observed and are followed by a quick change to the double-Mach-like configuration. The simulated structure tracks show that there are three different tracks associated with different triple points or the kink on the transverse wave. Comparisons with previous work and experiments indicate the presence of a strong structure for an ordinary detonation.

Some characteristics of low-speed streaks under sheared air-water interfaces

The characteristics of low-speed fluid streaks occurring under sheared air-water interfaces were examined by means of hydrogen bubble visualization technique. A critical shear condition under which the streaky structure first appears was determined to be u(tau) approximate to 0.19 cm/s. The mean spanwise streak spacing increases with distance from the water surface owing to merging and bursting processes, and a linear relationship describing variation of non-dimensional spacing <(+)over bar> versus y(+) was found essentially independent of shear stress on the interface. Values of <(+)over bar>, however, are remarkably smaller than their counterparts in the near-wall region of turbulent boundary layers. Though low-speed streaks occur randomly in time and space, the streak spacing exhibits a lognormal probability distribution behavior. A tentative explanation concerning the formation of streaky structure is suggested, and the fact that <(+)over bar> takes rather smaller values than that in wall turbulence is briefly discussed.

The Structure and Evolution of a Two-Dimensional H2/O2/Ar Cellular Detonation

This paper reports on two-dimensional numerical simulation of cellular detonation wave in a / / mixture with low initial pressure using a detailed chemical reaction model and high order WENO scheme. Before the final equilibrium structure is produced, a fairly regular but still non-equilibrium mode is observed during the early stage of structure formation process. The numerically tracked detonation cells show that the cell size always adapts to the channel height such that the cell ratio is fairly independent of the grid sizes and initial and boundary conditions. During the structural evolution in a detonation cell, even as the simulated detonation wave characteristics suggest the presence of an ordinary detonation, the evolving instantaneous detonation state indicates a mainly underdriven state. As a considerable region of the gas mixture in a cell is observed to be ignited by the incident wave and transverse wave, it is further suggested that these two said waves play an essential role in the detonation propagation.

Three-Dimensional Network Model And Its Parameter Calibration

A material model, whose framework is parallel spring-bundles oriented in 3-D space, is proposed. Based on a discussion of the discrete schemes and optimum discretization of the solid angles, a 3-D network cell consisted of one-dimensional components is developed with its geometrical and physical parameters calibrated. It is proved that the 3-D network model is able to exactly simulate materials with arbitrary Poisson ratio from 0 to 1/2, breaking through the limit that the previous models in the literature are only suitable for materials with Poisson ratio from 0 to 1/3. A simplified model is also proposed to realize high computation accuracy within low computation cost. Examples demonstrate that the 3-D network model has particular superiority in the simulation of short-fiber reinforced composites.

Numerical exploration of new features on three-dimensional transition of a cylinder wake

The three-dimensional transition of the wake flow behind a circular cylinder is studied in detail by direct numerical simulations using 3D incompressible N-S equations for Reynolds number ranging from 200 to 300. New features and vortex dynamics of the 3D transition of the wake are found and investigated. At Re = 200, the flow pattern is characterized by mode A instability. However, the spanwise characteristic length of the cylinder determines the transition features. Particularly for the specific spanwise characteristic length linear stable mode may dominate the wake in place of mode A and determine the spanwise phase difference of the primary vortices shedding. At Re = 250 and 300 it is found that the streamwise vortices evolve into a new type of mode - "dual vortex pair mode" downstream. The streamwise vortex structures switch among mode A, mode B and dual vortex pair mode from near wake to downstream wake. At Re = 250, an independent low frequency f(m) in addition to the vortex shedding frequency f(s) is identified. Frequency coupling between f(m) and f(s) occurs. These result in the irregularity of the temporal signals and become a key feature in the transition of the wake. Based on the formation analysis of the streamwise vorticity in the vicinity of cylinder, it is suggested that mode A is caused by the emergence of the spanwise velocity due to three dimensionality of the incoming flow past the cylinder. Energy distribution on various wave numbers and the frequency variation in the wake are also described.

Proper orthogonal decomposition of oscillatory Marangoni flow in half-zone liquid bridges of low-Pr fluids

Proper orthogonal decomposition (POD) using method of snapshots was performed on three different types of oscillatory Marangoni flows in half-zone liquid bridges of low-Pr fluid (Pr = 0.01). For each oscillation type, a series of characteristic modes (eigenfunctions) have been extracted from the velocity and temperature disturbances, and the POD provided spatial structures of the eigenfunctions, their oscillation frequencies, amplitudes, and phase shifts between them. The present analyses revealed the common features of the characteristic modes for different oscillation modes: four major velocity eigenfunctions captured more than 99% of the velocity fluctuation energy form two pairs, one of which is the most energetic. Different from the velocity disturbance, one of the major temperature eigenfunctions makes the dominant contribution to the temperature fluctuation energy. On the other hand, within the most energetic velocity eigenfuction pair, the two eigenfunctions have similar spatial structures and were tightly coupled to oscillate with the same frequency, and it was determined that the spatial structures and phase shifts of the eigenfunctions produced the different oscillatory disturbances. The interaction of other major modes only enriches the secondary spatio-temporal structures of the oscillatory disturbances. Moreover, the present analyses imply that the oscillatory disturbance, which is hydrodynamic in nature, primarily originates from the interior of the liquid bridge. (C) 2007 Elsevier B.V. All rights reserved.

Compact finite difference - Fourier spectral method for three-dimensional incompressible Navier-Stokes equations

A new compact finite difference-Fourier spectral hybrid method for solving the three dimensional incompressible Navier-Stokes equations is developed in the present paper. The fifth-order upwind compact finite difference schemes for the nonlinear convection terms in the physical space, and the sixth-order center compact schemes for the derivatives in spectral space are described, respectively. The fourth-order compact schemes in a single nine-point cell for solving the Helmholtz equations satisfied by the velocities and pressure in spectral space is derived and its preconditioned conjugate gradient iteration method is studied. The treatment of pressure boundary conditions and the three dimensional non-reflecting outflow boundary conditions are presented. Application to the vortex dislocation evolution in a three dimensional wake is also reported.

Two-dimensional simulation of inductive plasma sources with self-consistent power deposition

A time averaged two-dimensional fluid model including an electromagnetic module with self-consistent power deposition was developed to simulate the transport of a low pressure radio frequency inductively coupled plasma source. Comparsions with experiment and previous simulation results show, that the fluid model is feasible in a certain range of gas pressure. In addition, the effects of gas pressure and power input have been discussed.

Transient Temperature Distribution in a Sheet Metal by Low-Energy Laser Scanning

Low-energy laser-heating techniques are widely used in engineering applications such as, thinfilm deposition, surface treatment, metal forming and micro-structural pattern formation. In this paper,under the conditions of ignoring the thermo-mechanical coupling, a numerical simulation on the spatialand temporal temperature distribution in a sheet metal produced by the laser beam scanning in virtue of thefinite element method is presented. Both the three-dimensional transient temperature field and thetemperature evolution as a function of heat penetrating depth in the metal sheet are calculated. Thetemperature dependence of material properties was taken into account. It was shown that, after taking thetemperature dependence of the material absorbance effect into consideration, the temperature change ratealong the scanning direction and the temperature maximum were both increased.