海-气振子系统中未知参数的MCMC方法识别

利用MCMC方法对赤道东太平洋SST的海-气振子模型中未知参数进行识别.首先通过贝叶斯公式导出振子系统中参数的后验概率密度函数.然后采用自适应Metropolis算法构造未知参数的Markov链,截取收敛的样本序列估计参数值.数值试验结果表明：所提出的方法具有很高的估计精度,同时具有较好的抗噪声性能. 关键词： 海-气振子 参数估计 马尔科夫链蒙特卡罗方法     

The probabilistic solution of stochastic oscillators with even nonlinearity under poisson excitation

The probabilistic solutions of nonlinear stochastic oscillators with even nonlinearity driven by Poisson white noise are investigated in this paper. The stationary probability density function (PDF) of the oscillator responses governed by the reduced Fokker-Planck-Kolmogorov equation is obtained with exponentialpolynomial closure (EPC) method. Different types of nonlinear oscillators are considered. Monte Carlo simulation is conducted to examine the effectiveness and accuracy of the EPC method in this case. It is found that the PDF solutions obtained with EPC agree well with those obtained with Monte Carlo simulation, especially in the tail regions of the PDFs of oscillator responses. Numerical analysis shows that the mean of displacement is nonzero and the PDF of displacement is nonsymmetric about its mean when there is even nonlinearity in displacement in the oscillator. Numerical analysis further shows that the mean of velocity always equals zero and the PDF of velocity is symmetrically distributed about its mean.     

基于贝叶斯分层模型的MCMC方法在闪光图像重建中的应用

针对闪光图像得到的光程数据,采用贝叶斯分层模型建立了后验概率模型,运用Gibbs抽样动态构造马尔可夫链;进而获得了关于线吸收系数的统计结果及其不确定度,并与约束共轭梯度(CCG)方法进行对比分析。数值实验结果表明,马尔可夫链蒙特卡罗(MCMC)方法对理想光程图像的重建结果与真值近似完全一致;在含模糊和噪声时,重建结果与CCG方法相当;当含模糊且噪声干扰较大时,MCMC方法的重建结果要略优于CCG;更重要的是MCMC方法能够给出重建结果的不确定度。     

Joint estimation of the lognormal-Rician atmospheric turbulence model by the generalized method of moments

The generalized method of moments (GMM) is introduced in the framework of estimating the lognormal-Rician parameters. This GMM approach provides a systematic procedure for finding the moment-based parameter estimators. The GMM estimator can estimate the two shaping parameters of the lognormal-Rician PDF jointly. The asymptotic performance of the GMM estimator is compared with Monte Carlo simulation results. The results show that the GMM approach can lead to estimators with satisfactory performance over a wide range of channel conditions. The proposed method can be easily applied to both noiseless and noisy environments.     

Estimation of the power scintillation probability density function in free-space optical links by use of multicanonical Monte Carlo sampling

Free-space optics (FSO) can provide cost-effective, high-bandwidth, wireless connections. However, atmospheric turbulence may degrade the performance of FSO links by causing intensity and power scintillations at the receiver. Multicanonical Monte Carlo sampling is used in conjunction with the phase screen method to calculate the statistics, and particularly the probability density function (PDF), of the power fluctuations at an FSO receiver. This allows the efficient calculation of the PDF even for very small values with a limited number of iterations. The obtained PDF can be used to characterize the performance of the system in terms of the error probability.     

A Eulerian PDF scheme for LES of nonpremixed turbulent combustion with second-order accurate mixture fraction

Monte Carlo simulations of joint probability density function (PDF) approaches have been developed in the past largely with Reynolds averaged Navier Stokes (RANS) applications. Current interests are in the extension of PDF approaches to large eddy simulation (LES). As LES resolves accurately the large scales of turbulence in time, the Monte Carlo simulation and the flow field need to be tightly coupled. A tight coupling can be achieved if the consistency between the scalar field solution obtained via finite-volume (FV) methods and that from the stochastic solution of the PDF is ensured. For nonpremixed turbulent flames with two distinct streams, the local reactive mixture is described by the mixture fraction. A Eulerian Monte Carlo method is developed to achieve a second-order accuracy in the instantaneous filtered mixture fraction that is consistent with the corresponding FV. The performances of the proposed scheme are extensively evaluated using a one-dimensional model. Then, the scheme is applied to two cases with LES. The first one is a non-reacting mixing flow of two different fluids. The second case is the Sandia piloted turbulent flame D with a steady state flamelet model. Both results confirm the consistency of the proposed method to the level of filtered mixture fraction.     

钝体后湍流预混燃烧的PDF模拟

本文采用PDF方法对矩形燃烧室内钝体后的湍流预混火焰进行了数值模拟。脉动速度-频率-标量联合的PDF输运方程用Monte Carlo方法求解,质量、动量和能量的平均值由基于无结构网格的有限体积法求解,压力通过状态方程获得。PDF方程中所需的平均密度、平均速度和压力由有限体积法提供,并将用Monte Carlo方法求出的雷诺应力、化学反应源项和比热比传递给有限体积法。本文对丙烷和空气燃烧的不同简化化学反应机理进行了研究,并与实验结果进行比较,获得满意的结果。     

Probabilistic solution of some multi-degree-of-freedom nonlinear systems under external independent Poisson white noises

This paper studies the stationary probability density function (PDF) of the response of multi-degree-of-freedom nonlinear systems under external independent Poisson white noises. The PDF is governed by the high-dimensional generalized Fokker-Planck-Kolmogorov (FPK) equation. The state-space-split (3S) method is adopted to reduce the high-dimensional generalized FPK equation to a low-dimensional equation. Subsequently, the exponential-polynomial closure (EPC) method is further used to solve the reduced FPK equation for the PDF solution. Two illustrative examples are presented to examine the accuracy of the 3S-EPC solution procedure. One example involves a two-degree-of-freedom coupled nonlinear system. The other example is concerned with a ten-degree-of-freedom system with cubic terms in displacement. A Monte Carlo simulation is also performed for simulating the PDF solution of the response. The comparison with the simulated result shows that the 3S-EPC solution procedure can provide satisfactory PDF solutions. The good agreement is also observed in the tail regions of the PDF solutions.     

Probabilistic solution of nonlinear oscillators excited by combined Gaussian and Poisson white noises

The stationary probability density function (PDF) solution of the stochastic response of nonlinear oscillators is investigated in this paper. The external excitation is assumed to be a combination of Gaussian and Poisson white noises. The PDF solution is governed by the generalized Kolmogorov equation which is solved by the exponential-polynomial closure (EPC) method. In order to evaluate the effectiveness of the EPC method, different nonlinear oscillators are considered in numerical analysis. Nonlinearity exists either in displacement or in velocity for these nonlinear oscillators. The impulse arrival rate, mono-modal PDF and bi-modal PDF are also considered in this study. Compared to the PDF given by Monte Carlo simulation, the EPC method presents good agreement with the simulated result, which can also be observed in the tail region of the PDF solution.     

地声参数及传播损失不确定性估计与建模

地声参数的不确定性对水声传播具有重要的影响。通过贝叶斯理论建立水声环境不确定性推理模型,理论推导了地声参数的似然函数以及地声参数和传播损失的后验概率密度,并采用MCMC(Markov Chain Monte Carlo)进行了仿真计算,给出了地声参数的二维后验联合概率密度和一维边缘概率密度,在此基础上对传播损失的不确定性进行了估计,得到了传播损失80%的可信区间。仿真和实验结果表明,该方法适用于地声参数反演和不确定性估计,并能获取因地声参数不确定性导致的传播损失不确定性估计。     

Relaxation times of spin states of all ranks and orders of quadrupolar nuclei estimated from NMR z-spectra: Markov chain Monte Carlo analysis applied to 7Li+ and 23Na+ in stretched hydrogels

The NMR z-spectra of 7Li+ and 23Na+ in stretched hydrogels contain five minima, or critical values, with a sharp "dagger" on the central dip. The mathematical representation of such z-spectra from spin-3/2 nuclei contains nine distinct (the total is 15 but there is redundancy of the ±order-numbers) relaxation rate constants that are unique for each of the spin states, up to rank 3, order 3. We present an approach to multiple-parameter-value estimation that exploits the high level of separability of the effects of each of the relaxation rate constants on the features of the z-spectrum. The Markov chain Monte Carlo (MCMC) method is computationally demanding but it yielded statistically robust estimates (low coefficients of variation) of the parameter values. We describe the implementation of the MCMC analysis (in the present context) and posit that it can obviate the need for using multiple-quantum filtered RF-pulse sequences to estimate all relaxation rate constants/times under experimentally favorable, but readily achievable, circumstances.     

两相壁面射流PDF方程有限分析方法及数值模拟

提出求解位置-速度相空间中高维两相流PDF(probability density function)方程的有限分析方法,将位置-速度相空间颗粒PDF方程约化到速度空间,并解析求解,颗粒的位置PDF用轨道方法求解.对壁面射流两相流动进行数值模拟,并与颗粒雷诺应力轨道方法进行比较计算,结果优于颗粒雷诺应力轨道方法.     

随机阈值冲击模型金属化膜脉冲电容器可靠性分析

基于惯性约束聚变强激光装置中金属化膜脉冲电容器"自愈"的失效机理,提出了更为适合的冲击模型金属化膜脉冲电容器可靠性评估方法。现有的加速退化试验中,失效阈值往往是事先确定的,但考虑可靠性产品个体差异及环境影响的不同,这是不合理的,针对此问题提出了基于随机阈值的冲击模型金属化膜脉冲电容器退化建模的方法。由于模型过于复杂,采用MCMC方法进行参数估计。最后通过对电容器的仿真实验,将随机阈值下的评估结果与固定阈值的情况相对比,说明了该模型和方法的合理性,并进一步分析了不同的阈值分布均值和方差对产品可靠性的影响。     

长程差分吸收光谱技术气体浓度反演误差的定量估计

长程差分吸收光谱法(LP-DOAS)是基于最小二乘原理来反演大气痕量气体浓度的。LP-DOAS能对痕量气体进行高灵敏的测量,但是还没有统计的方法定量确定LP-DOAS反演误差。痕量气体的吸收通常很弱,外来影响因素决定了检测限和测量精度,其被误当做真正的吸收,增加了没有统计特性的噪声到残差中,导致最小二乘拟合误差(err(LSQ))有一个明显的误估计。研究采用蒙特卡罗方法,通过残差的循环移位定量确定差分吸收光谱法反演气体浓度的误差。实验结果表明,蒙特卡罗方法可以定量估计差分吸收光谱法反演误差,误估计因子为1.13,而err(LSQ)为3.12。     

基于水平集的两步MCMC方法对河道形态的识别

首先运用符号距离函数刻画河道的复杂几何形态，然后通过求解水平集演化方程和两步马尔科夫链蒙特卡罗（MCMC）算法拟合生产历史数据，逐步更新河道的边界.在两步MCMC方法的第一步，应用流线模拟计算的敏感性矩阵获取近似的似然函数，修改MCMC的推荐概率分布；第二步，为确保MCMC算法的严密性，对通过第一步的油藏模型进行完整的数值模拟以获取精准的似然函数，并用更改的接受概率作为模型接受的判断准则.最后通过二维计算实例验证该方法的有效性.     

Particle filtering with path sampling and an application to a bimodal ocean current model

This paper introduces a recursive particle filtering algorithm designed to filter high dimensional systems with complicated non-linear and non-Gaussian effects. The method incorporates a parallel marginalization (PMMC) step in conjunction with the hybrid Monte Carlo (HMC) scheme to improve samples generated by standard particle filters. Parallel marginalization is an efficient Markov chain Monte Carlo (MCMC) strategy that uses lower dimensional approximate marginal distributions of the target distribution to accelerate equilibration. As a validation the algorithm is tested on a 2516 dimensional, bimodal, stochastic model motivated by the Kuroshio current that runs along the Japanese coast. The results of this test indicate that the method is an attractive alternative for problems that require the generality of a particle filter but have been inaccessible due to the limitations of standard particle filtering strategies.     

路径积分Monte Carlo方法与非线性随机振动

给出了一种求解非线性随机振动问题的新方法-基于路径积分表述的Monte Carlo方法。受白噪声激励的非线性系统的各响应统计值被表成路径积分形式,并采用Monte Carlo方法进行计算。讨论了应用方面的问题,并计算了两个实例。     

Fast Compression of MCMC Output

We propose cube thinning, a novel method for compressing the output of an MCMC (Markov chain Monte Carlo) algorithm when control variates are available. It allows resampling of the initial MCMC sample (according to weights derived from control variates), while imposing equality constraints on the averages of these control variates, using the cube method (an approach that originates from survey sampling). The main advantage of cube thinning is that its complexity does not depend on the size of the compressed sample. This compares favourably to previous methods, such as Stein thinning, the complexity of which is quadratic in that quantity.     

Some Spin Glass Ideas Applied to the Clique Problem

In this paper we introduce a new algorithm to study some NP-complete problems. This algorithm is a Markov Chain Monte Carlo (MCMC) inspired by the cavity method developed in the study of spin glass. We will focus on the maximum clique problem and we will compare this new algorithm with several standard algorithms on some DIMACS benchmark graphs and on random graphs. The performances of the new algorithm are quite surprising. Our effort in this paper is to be clear as well to those readers who are not in the field.     

Relative intensity noise induced by four-wave mixing in the case of two-wave transmission in an optical fiber

Four-wave mixing (FWM) is a significant nonlinear effect in wavelength division multiplexing (WDM) fiber-optic systems. For two-wave transmission, it is easily found that the FWM noise power decreases with frequency spacing and increases with signal power. However, the variation of relative intensity noise (RIN) with frequency spacing and signal power is only 2 dB at most. The intensity fluctuations induced by the energy exchange between the FWM generated new waves and the original ones are trivial and the influence of FWM on RIN can be neglected. It is also found that the increase of RIN with signal power is mainly attributed to stimulated Brillouin scattering (SBS) rather than FWM.