Local linear convergence analysis of Primal–Dual splitting methods

In this paper, we study the local linear convergence properties of a versatile class of Primal–Dual splitting methods for minimizing composite non-smooth convex optimization problems. Under the assumption that the non-smooth components of the problem are partly smooth relative to smooth manifolds, we present a unified local convergence analysis framework for these methods. More precisely, in our framework, we first show that (i) the sequences generated by Primal–Dual splitting methods identify a pair of primal and dual smooth manifolds in a finite number of iterations, and then (ii) enter a local linear convergence regime, which is characterized based on the structure of the underlying active smooth manifolds. We also show how our results for Primal–Dual splitting can be specialized to cover existing ones on Forward–Backward splitting and Douglas–Rachford splitting/ADMM (alternating direction methods of multipliers). Moreover, based on these obtained local convergence analysis result, several practical acceleration techniques are discussed. To exemplify the usefulness of the obtained result, we consider several concrete numerical experiments arising from fields including signal/image processing, inverse problems and machine learning. The demonstration not only verifies the local linear convergence behaviour of Primal–Dual splitting methods, but also the insights on how to accelerate them in practice.     

Quasi-linear PDEs and forward–backward stochastic differential equations: Weak solutions

In this paper, we study the existence, uniqueness and the probabilistic representation of the weak solutions of quasi-linear parabolic and elliptic partial differential equations (PDEs) in the Sobolev space $H_{\rho }^1({\mathbb{R}}^d)$. For this, we study first the solutions of forward–backward stochastic differential equations (FBSDEs) with smooth coefficients, regularity of solutions and their connection with classical solutions of quasi-linear parabolic PDEs. Then using the approximation procedure, we establish their convergence in the Sobolev space to the solutions of the FBSDES in the space $L_{\rho }^2({\mathbb{R}}^d;{\mathbb{R}}^d)?L_{\rho }^2({\mathbb{R}}^d;{\mathbb{R}}^k)?L_{\rho }^2({\mathbb{R}}^d;{\mathbb{R}}^{k\times d})$. This gives a connection with the weak solutions of quasi-linear parabolic PDEs. Finally, we study the unique weak solutions of quasi-linear elliptic PDEs using the solutions of the FBSDEs on infinite horizon.     

Explicit Solutions of Quadratic FBSDEs Arising From Quadratic Term Structure Models

We provide explicit solutions of certain forward-backward stochastic differential equations (FBSDEs) with quadratic growth. These particular FBSDEs are associated with quadratic term structure models of interest rates and characterize the zero-coupon bond price. The results of this paper are naturally related to similar results on affine term structure models of Hyndman (Math. Financ. Econ. 2(2):107–128, 2009) due to the relationship between quadratic functionals of Gaussian processes and linear functionals of affine processes. Similar to the affine case a sufficient condition for the explicit solutions to hold is the solvability in a fixed interval of Riccati-type ordinary differential equations. However, in contrast to the affine case, these Riccati equations are easily associated with those occurring in linear-quadratic control problems. We also consider quadratic models for a risky asset price and characterize the futures price and forward price of the asset in terms of similar FBSDEs. An example is considered, using an approach based on stochastic flows that is related to the FBSDE approach, to further emphasize the parallels between the affine and quadratic models. An appendix discusses solvability and explicit solutions of the Riccati equations.     

Main Components of Xinjiang Lavender Essential Oil Determined by Partial Least Squares and Near Infrared SpectroscopySCIEI北大核心CSCD

This work was undertaken to establish a quantitative analysis model which can rapid determinate the content of linalool, linalyl acetate of Xinjiang lavender essential oil. Totally 165 lavender essential oil samples were measured by using near infrared absorption spectrum(NIR), after analyzing the near infrared spectral absorption peaks of all samples, lavender essential oil have abundant chemical information and the interference of random noise may be relatively low on the spectral intervals of 7100-4 500 cm(-1). Thus, the PLS models was constructed by using this interval for further analysis. 8 abnormal samples were eliminated. Through the clustering method, 157 lavender essential oil samples were divided into 105 calibration set samples and 52 validation set samples. Gas chromatography mass spectrometry (GC-MS) was used as a tool to determine the content of linalool and linalyl acetate in lavender essential oil. Then the matrix was established with the GC-MS raw data of two compounds in combination with the original NIR data. In order to optimize the model, different pretreatment methods were used to preprocess the raw NIR spectral to contrast the spectral filtering effect, after analysizing the quantitative model results of linalool and linalyl acetate, the root mean square error prediction(RMSEP) of orthogonal signal transformation (OSC) was 0.226, 0.558, spectrally, it was the optimum pretreatment method. In addition, forward interval partial least squares (FiPLS) method was used to exclude the wavelength points which has nothing to do with determination composition or present nonlinear correlation, finally 8 spectral intervals totally 160 wavelength points were obtained as the dataset. Combining the data sets which have optimized by OSC-FiPLS with partial least squares(PLS) to establish a rapid quantitative analysis model for determining the content of linalool and linalyl acetate in Xinjiang lavender essential oil, numbers of hidden variables of two components were 8 in the model. The performance of the model was evaluated according to root mean square error of cross-validation (RMSECV) 9 root mean square error of prediction (RMSEP). In the model, RESECV of linalool and linalyl acetate were 0.170 and 0.416, respectively; RMSEP were 0.188 and 0.364. The results indicated that raw data was pretreated by OSC and FiPLS, the NIR-PLS quantitative analysis model with good robustness, high measurement precision; it could quickly determine the content of linalool and linalyl acetate in lavender essential oil. In addition, the model has a favorable prediction ability. The study also provide a new effective method which could rapid quantitative analysis the major components of Xinjiang lavender essential oil.     

基于Labview的一级环形倒立摆逆系统轨迹法自动摆起控制

利用求解非线性方程两点边值问题,得到逆系统参考轨迹;通过设计逆系统前馈控制及LQR反馈控制器,对参考轨迹实时跟踪。基于QNET 2.0旋转倒立摆实验平台,采用Labview编程软件,实现了一级环形倒立摆的自动摆起与稳定的实物控制。实验结果证明控制算法的有效性,可以使一级环形倒立摆在一个摆周期内自动摆起,并保持稳定状态。与能量法自动起摆实验比较,逆系统算法具有较好的快速性和稳定性。     

Backward–forward algorithms for structured monotone inclusions in Hilbert spaces

In this paper, we study the backward–forward algorithm as a splitting method to solve structured monotone inclusions, and convex minimization problems in Hilbert spaces. It has a natural link with the forward–backward algorithm and has the same computational complexity, since it involves the same basic blocks, but organized differently. Surprisingly enough, this kind of iteration arises when studying the time discretization of the regularized Newton method for maximally monotone operators. First, we show that these two methods enjoy remarkable involutive relations, which go far beyond the evident inversion of the order in which the forward and backward steps are applied. Next, we establish several convergence properties for both methods, some of which were unknown even for the forward–backward algorithm. This brings further insight into this well-known scheme. Finally, we specialize our results to structured convex minimization problems, the gradient-projection algorithms, and give a numerical illustration of theoretical interest.     

近景大视场角分区域标定方法研究

针对飞行试验测量视场大相机标定精度低的问题,提出一种高精度CCD相机分区域标定方法。该方法首先通过将标靶均匀布置在摄像机视场内,使得标靶尽可能均匀错落地充满整个视场范围,再结合人眼判读的方式求解靶标的像面位置,最终与全站仪三维坐标形成精确的空间标定点集。接着,将像平面按横向方向等间距分割成Ｎ个区域,并结合后方交会的方法分别对每个子区域进行相机参数的计算。实验结果表明：经过分区域标定,相机采集点的总误差比单区域标定法降低了4％(Ｎ＝3)。算法可实现指定区域的相机参数计算,基本满足中高等精度的工业测量要求。所本文研究可应用于位置相对固定不变的工业视觉测量,特别是大工件测量领域。     

用于校准能见度仪的标准散射体定标系统的校准方法

为了实现对用于校准能见度仪的标准散射体的快速准确定标,通过测量标准散射体不同散射角的散射系数,实现对标准散射体不同散射角所模拟的大气能见度进行准确定标并对定标系统进行校准.根据定标系统的组成与工作原理,确定定标系统的校准方法并建立相应的校准链,并设计校准链的各组成部分的校准方法.通过对已知散射系数的标准散射体进行定标,验证校准后定标系统定标结果的准确性,进而验证了定标系统校准方法的正确性.实验结果证明:定标系统对标准散射体的散射系数定标结果的误差为7.93%;经由定标系统定标的标准散射体所模拟的大气能见度的最大误差为8.61%,满足用于校准能见度仪的标准散射体的使用要求.     

飞秒激光与固体靶相互作用的前向快电子束发散研究

为了研究前向快电子束的发散问题，采用飞秒激光辐照平面铜靶，以电子角分布仪和氟化锂(LiF)热释光探测器探测快电子的前向发散角。实验显示前向快电子束发散角均在95°(半高宽FWHM)以上。理论计算指出预等离子体中电子的初速度以及加速后的电子在靶内碰撞效应均不是产生大发散角的重要原因。根据激光-等离子体相互作用理论，等离子体中的电磁场才是导致前向快电子束产生较大发散角的主要原因。这个研究为获取较小发散角的前向快电子束提供了实验设置的依据。     

Probabilistic interpretation of a system of quasilinear parabolic PDEs

Using a forward–backward stochastic differential equations (FBSDE) associated to a transmutation process driven by a finite sequence of Poisson processes, we obtain a probabilistic interpretation for a non-degenerate system of quasilinear parabolic partial differential equations (PDEs). The novetly is that the linear second order differential operator is different on each line of the system.