Second-order quantile methods for experts and combinatorial games

We aim to design strategies for sequential decision making that adjust to the difficulty of the learning problem. We study this question both in the setting of prediction with expert advice, and for more general combinatorial decision tasks. We are not satisfied with just guaranteeing minimax regret rates, but we want our algorithms to perform significantly better on easy data. Two popular ways to formalize such adaptivity are second-order regret bounds and quantile bounds. The underlying notions of 'easy data', which may be paraphrased as "the learning problem has small variance" and "multiple decisions are useful", are synergetic. But even though there are sophisticated algorithms that exploit one of the two, no existing algorithm is able to adapt to both. In this paper we outline a new method for obtaining such adaptive algorithms, based on a potential function that aggregates a range of learning rates (which are essential tuning parameters). By choosing the right prior we construct efficient algorithms and show that they reap both benefits by proving the first bounds that are both second-order and incorporate quantiles.

A Novel High Order Space-Time Spectral Method for the Time Fractional Fokker--Planck Equation

The fractional Fokker-Planck equation is an important physical model for simulating anomalous diffusions with external forces. Because of the non-local property of the fractional derivative an interesting problem is to explore high accuracy numerical methods for fractional differential equations. In this paper, a space-time spectral method is presented for the numerical solution of the time fractional Fokker-Planck initial-boundary value problem. The proposed method employs the Jacobi polynomials for the temporal discretization and Fourier-like basis functions for the spatial discretization. Due to the diagonalizable trait of the Fourier-like basis functions, this leads to a reduced representation of the inner product in the Galerkin analysis. We prove that the time fractional Fokker-Planck equation attains the same approximation order as the time fractional diffusion equation developed in [23] by using the present method. That indicates an exponential decay may be achieved if the exact solution is sufficiently smooth. Finally, some numerical results are given to demonstrate the high order accuracy and efficiency of the new numerical scheme. The results show that the errors of the numerical solutions obtained by the space-time spectral method decay exponentially.

Random vibration of a second order non-linear elastic system

A method is presented for obtaining, approximately, the response covariance and probability distribution of a non-linear oscillator under a Gaussian excitation. The method has similarities with the hierarchy closure and the equivalent linearization approaches, but is different. A Gaussianization technique is used to arrive at the output autocorrelation and the input-output cross-correlation. This along with an energy equivalence criterion is used to estimate the response distribution function. The method is applicable in both the transient and steady state response analysis under either stationary or non-stationary excitations. Good comparison has been observed between the predicted and the exact steady state probability distribution of a Duffing oscillator under a white noise input.

Numerical procedure for second order non-linear ordinary differential equations and application to heat transfer problem

In this paper, we have first given a numerical procedure for the solution of second order non-linear ordinary differential equations of the type y″ = f (x;y, y′) with given initial conditions. The method is based on geometrical interpretation of the equation, which suggests a simple geometrical construction of the integral curve. We then translate this geometrical method to the numerical procedure adaptable to desk calculators and digital computers. We have studied the efficacy of this method with the help of an illustrative example with known exact solution. We have also compared it with Runge-Kutta method. We have then applied this method to a physical problem, namely, the study of the temperature distribution in a semi-infinite solid homogeneous medium for temperature-dependent conductivity coefficient.

Aryl monosulphides and symmetrical disulphides based second order nonlinear optical chromophores with transparency in the visible

A series of aryl monosulphides and disulphides have been synthesized and characterized. Their molecular hyperpolarizability (beta) has been measured in solution with the hyper-Rayleigh Scattering technique and also calculated by semiempirical AMI method. The trend in the observed and calculated values of first hyperpolarizability of these compounds has been found to be in good agreement. These compounds show moderate P values and excellent transparency in the visible region.

Second order transport from anomalies

We study parity odd transport at second order in derivative expansion for a non-conformal charged fluid. We see that there are 27 parity odd transport coefficients, of which 12 are non-vanishing in equilibrium. We use the equilibrium partition function method to express 7 of these in terms of the anomaly, shear viscosity, charge diffusivity and thermodynamic functions. The remaining 5 are constrained by 3 relations which also involve the anomaly. We derive Kubo formulae for 2 of the transport coefficients and show these agree with that derived from the equilibrium partition function.

Micro-satellite angular rate estimation using second-order sliding mode observer

Development of computationally efficient and accurate attitude rate estimation algorithm using low-cost commercially available star sensor arrays and processing unit for micro-satellite mission is presented. Our design reduces the computational load of least square (LS)-based rate estimation method while maintaining the same accuracy compared to other rate estimation approaches. Furthermore, rate estimation accuracy is improved by using recently developed fast and accurate second-order sliding mode observer (SOSMO) scheme. It also gives robust estimation in the presence of modeling uncertainties, unknown disturbances, and measurement noise. Simulation study shows that rate estimation accuracy achieved by our LS-based method is comparable with other methods for a typical commercially available star sensor array. The robustness analysis of SOSMO with respect to measurement noise is also presented in this paper. Simulation test bench for a practical scenario of satellite rate estimation uses moment-of-inertia variation and environmental disturbances affecting a typical micro-satellite at 500km circular orbit. Comparison studies of SOSMO with 1-SMO and pseudo-linear Kalman filter show that satisfactory estimation accuracy is achieved by SOSMO.

Mode coupling between first- and second-order whispering-gallery modes in coupled microdisks

The mode characteristics for two coupled microdisks are investigated by the finite-difference time-domain technique. In the two coupled micodisks, mode coupling between the same order whispering-gallery modes (WGMs) results in coupled WGMs with split mode wavelengths. The numerical results show that the split mode wavelengths of the coupled first- and second-order WGMs can have a crossing point in some cases, which can induce anticrossing mode coupling between them and greatly reduce the mode Q factor of the coupled first-order WGMs. The time variation of mode field pattern shows the transformation between the coupled first- and second-order WGMs. (C) 2007 Optical Society of America

Effect of the electrostatic interactions of chromophores on the macroscopic second-order nonlinear optical properties of the polymer films

The polyetherketone (PEK-c) guest-host system thin films doped with 3-(1,1-dicyanothenyl)-1-phenyl-4,5-dihydro-1H-pryazole (DCNP) were prepared. Their second-order nonlinear optical (NLO) coefficients chi(33)((2)) were measured by using Maker fringe method for the polymer films doped with different weight percents of DCNP. Experimental results indicate that the second-order NLO properties of the poled polymer films could decrease with the chromophore loading increasing when the chromophore loading reaches a fairly high level. In this paper, the relationship between the macroscopic second-order NLO coefficient and the chromophore number density was modified under considering the role of the electrostatic interactions of chromophores in the polymer film. According to the modified relationship, the macroscopic second-order NLO coefficient is no longer in direct proportion with the chromophore number density in the polymer film. The effect of the electrostatic interactions of chromophores on second-order NLO properties was discussed. The attenuation of the macroscopic second-order NLO activity can be demonstrated by the role of the chromophore electrostatic interactions at high loading of chromophore in the polymer systems.

An automatic step adjustment method for average power analysis technique used in fiber amplifiers

An automatic step adjustment (ASA) method for average power analysis (APA) technique used in fiber amplifiers is proposed in this paper for the first time. In comparison with the traditional APA technique, the proposed method has suggested two unique merits such as a higher order accuracy and an ASA mechanism, so that it can significantly shorten the computing time and improve the solution accuracy. A test example demonstrates that, by comparing to the APA technique, the proposed method increases the computing speed by more than a hundredfold under the same errors. By computing the model equations of erbium-doped fiber amplifiers, the numerical results show that our method can improve the solution accuracy by over two orders of magnitude at the same amplifying section number. The proposed method has the capacity to rapidly and effectively compute the model equations of fiber Raman amplifiers and semiconductor lasers. (c) 2006 Optical Society of America

Second-order Stokes solutions for internal waves in three-layer density-stratified fluid

In this paper, internal waves in three-layer stratified fluid are investigated by using a perturbation method, and the second-order asymptotic solutions of the velocity potentials and the second-order Stokes solutions of the associated elevations of the interfacial waves are presented based on the small amplitude wave theory. As expected, the first-order solutions are consistent with ordinary linear theoretical results, and the second-order solutions describe the second-order modification on the linear theory and the interactions between the two interfacial waves. Both the first-order and second-order solutions derived depend on the depths and densities of the three-layer fluid. It is also noted that the solutions obtained from the present work include the theoretical results derived by Umeyama as special cases.

Representation of microscale ocean wavenumber spectrum correct to the second order

In this paper, the analytical representations of four wave source functions in high-frequency spectrum range are given on the basis of ocean wave theory and dimensional analysis, and the perturbation method is used to solve the governing equations of ocean wave high-frequency spectrum on the basis of the temporally stationary and locally homogeneous scale relations of microscale wave. The microscale ocean wavenumber spectrum correct to the second order has an explicit structure, its first order part represents the equilibrium between different source functions, and its second order part represents the contribution of microscale wave propagation.

Statistical distribution of depth-integrated local horizontal momentum for second-order random ocean waves in finite water depth

Based on the second-order random wave solutions of water wave equations in finite water depth, statistical distributions of the depth- integrated local horizontal momentum components are derived by use of the characteristic function expansion method. The parameters involved in the distributions can be all determined by the water depth and the wave-number spectrum of ocean waves. As an illustrative example, a fully developed wind-generated sea is considered and the parameters are calculated for typical wind speeds and water depths by means of the Donelan and Pierson spectrum. The effects of nonlinearity and water depth on the distributions are also investigated.

Statistical distribution of wave-surface elevation for second-order random directional ocean waves in finite water depth

Based on the second-order random wave solutions of water wave equations in finite water depth, a statistical distribution of the wave-surface elevation is derived by using the characteristic function expansion method. It is found that the distribution, after normalization of the wave-surface elevation, depends only on two parameters. One parameter describes the small mean bias of the surface produced by the second-order wave-wave interactions. Another one is approximately proportional to the skewness of the distribution. Both of these two parameters can be determined by the water depth and the wave-number spectrum of ocean waves. As an illustrative example, we consider a fully developed wind-generated sea and the parameters are calculated for various wind speeds and water depths by using Donelan and Pierson spectrum. It is also found that, for deep water, the dimensionless distribution reduces to the third-order Gram-Charlier series obtained by Longuet-Higgins [J. Fluid Mech. 17 (1963) 459]. The newly proposed distribution is compared with the data of Bitner [Appl. Ocean Res. 2 (1980) 63], Gaussian distribution and the fourth-order Gram-Charlier series, and found our distribution gives a more reasonable fit to the data. (C) 2002 Elsevier Science B.V. All rights reserved.