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.     

NIFTy 3 – Numerical Information Field Theory: A Python Framework for Multicomponent Signal Inference on HPC Clusters

NIFTy , “Numerical Information Field Theory,” is a software framework designed to ease the development and implementation of field inference algorithms. Field equations are formulated independently of the underlying spatial geometry allowing the user to focus on the algorithmic design. Under the hood, NIFTy ensures that the discretization of the implemented equations is consistent. This enables the user to prototype an algorithm rapidly in 1D and then apply it to high‐dimensional real‐world problems. This paper introduces NIFTy  3, a major upgrade to the original NIFTy  framework. NIFTy  3 allows the user to run inference algorithms on massively parallel high performance computing clusters without changing the implementation of the field equations. It supports n‐dimensional Cartesian spaces, spherical spaces, power spaces, and product spaces as well as transforms to their harmonic counterparts. Furthermore, NIFTy  3 is able to handle non‐scalar fields, such as vector or tensor fields. The functionality and performance of the software package is demonstrated with example code, which implements a mock inference inspired by a real‐world algorithm from the realm of information field theory. NIFTy  3 is open‐source software available under the GNU General Public License v3 (GPL‐3) at https://gitlab.mpcdf.mpg.de/ift/NIFTy/tree/NIFTy_3 .     

An Introduction to Influence Theory: Kinematics and Dynamics

Influence theory is a foundational theory of physics that is not based on traditional empirically defined concepts, such as positions in space and time, mass, energy, or momentum. Instead, the aim is to derive these concepts, and their empirically determined relationships, from a more primitive model. It is postulated that there exist things, which are call particles, that influence one another in a discrete and directed fashion resulting in a partially ordered set of influence events. The problem of consistent quantification of the influence events is considered. Observers are modeled as particle chains (observer chains) as if an observer were able to track a particle and quantify the influence events that the particle experiences. From these quantified influence events, consistent quantification of the universe of events based on the observer chains is studied. Herein, the kinematics and dynamics of particles from the perspective of influence theory are both reviewed and further developed.     

偏缠绕模的Frobenius性质

主要给出了偏缠绕模的Frobenius性质，推广了缠绕模相应的性质。     

改进的PNC方法及其在非线性偏微分方程多解问题中的应用

2017年，李昭祥等提出了一种偏牛顿-校正法（Partial Newton-Correction Method，简记为PNC方法），并利用它成功地计算出了三类非线性偏微分方程的多重不稳定解.本文在PNC方法的基础上，提出并发展了一种改进的PNC方法.首先，利用Nehari流形$\mathcal{N}$与零平凡解的可分离性，建立并证明了$\mathcal{N}$的某特殊子流形$\mathcal{M}$上的全局分离定理及其推广（即局部分离定理）.全局分离定理只跟非线性偏微分算子或相应的非线性泛函本身有关，而与具体的计算方法无关.对一些典型的非线性偏微分方程多解问题（比如，Henon方程问题），该全局分离定理的分离条件，经验证是成立的.另一个方面，通过修改或补充原辅助变换的定义，去掉了原辅助变换的奇异性；接着建立并证明了某些非线性偏微分方程问题的新未知解与该非线性偏微分算子零核空间的密切关系；在证明中，去掉了在原奇异变换下所需的标准收敛（standard convergence）假设.最后，计算实例与数值结果验证了改进的PNC方法的可行性和有效性；同时表明子流形$\mathcal{M}$与已知解的可分离性是PNC方法和本文新方法能成功找到多解的关键.     

基于视觉辐条图案识别方法的蓝宝石引晶机制研究

工业生产蓝宝石晶体过程中,引晶步骤有着至关重要的地位。引晶必须在温度梯度较小,温度分布趋于稳定的条件下进行。目前,工业生产蓝宝石主要依靠人工经验操控籽晶杆实现引晶操作,但是人工引晶操作的准确性不高会导致成品品质不佳、资源浪费。为此,本文提出一种基于蓝宝石视觉辐条图案识别方法来检测蓝宝石熔体状态自由液面状态,从而实现一种高效率引晶的机制。此方法利用经典骨架化算法细化辐条图案,Harris算子实现特征信息的提取,提取的特征信息放入运动轨迹模型中判断熔体稳定性,分析液面温度分布稳定性从而实现引晶。结果表明,此算法具有有效性,蓝宝石晶体引晶效率大大提高,生产出的成品良率也有提升,可有效指导蓝宝石的工业生产。     

近红外光谱结合共识模型快速检测果酒的总酚含量

果酒发酵中的多酚是引起果酒口感、颜色变化的重要因素。为保证果酒品质,有必要开发一种快速监测发酵过程中多酚含量变化的技术。收集不同批次成熟期的蓝莓、桑葚为原料,分别碾压成汁,同时按比例混合二者,于小型发酵罐进行发酵。通过离线收集不同发酵时段的发酵液于离心管,高速离心后取上清液置于棕色瓶保存,共计得到48个果酒发酵样本。将上清液置于三个平行样比色皿,以傅里叶快速变换近红外光谱仪(FT-NIR)采集其透射光谱,取平均值作为该样本的光谱信号。然后将棕色瓶内的发酵液以国标法(即以标准液的吸光度值制定标准曲线)测定各样品的总酚含量,以duplex法计算样本光谱之间的距离且按2∶1的比例划分为训练集和预测集。采用间隔偏最小二乘法(iPLS)将训练集样本的透射光谱与总酚含量之间构建定量模型,间隔数从2依次变化到60个。该研究创新之处是使用共识方法融合多个已构建好的iPLS成员模型,按一定的共识规则分配权系数。通过各成员模型交互验证的残差及其残差之间的相关性来优化各成员模型的线性组合,以拉格朗日乘数法求解各成员模型的权系数,使间隔偏最小二乘-共识模型(consensual iPLS,CiPLS)的交互验证均方根误差最小。相比于全局PLS模型、划分不同间隔数量时的iPLS模型,CiPLS均具有较小的预测误差。当划分39个间隔时由三个iPLS成员模型(即14th,16th,18th)组成的共识模型误差最小为124.2,交互验证相关系数为0.944,对预测集样本的预测均方根误差为163.4,预测相关系数为0.931,预测性能均优于PLS和iPLS模型。另外,作为对比选用连续投影算法与无信息变量剔除法来优化光谱模型,其预测性能均不及本文提出的共识模型。分析各iPLS模型预测残差之间的相关性,发现共识模型主要是融合那些具有较高预测性能且模型间较低相关性的成员模型。结果表明,光谱分析结合共识方法可提高回归模型的预测精度、减少建模所需变量数,能够用于果酒总酚含量的离线快速检测。     

On-chip multiphoton Greenberger–Horne–Zeilinger state based on integrated frequency combs

One of the most important multipartite entangled states, Greenberger–Horne–Zeilinger state (GHZ), serves as a fundamental resource for quantum foundation test, quantum communication and quantum computation. To increase the number of entangled particles, significant experimental efforts should been invested due to the complexity of optical setup and the difficulty in maintaining the coherence condition for high-fidelity GHZ state. Here, we propose an ultra-integrated scalable on-chip GHZ state generation scheme based on frequency combs. By designing several microrings pumped by different lasers, multiple partially overlapped quantum frequency combs are generated to supply as the basis for on-chip polarization-encoded GHZ state with each qubit occupying a certain spectral mode. Both even and odd numbers of GHZ states can be engineered with constant small number of integrated components and easily scaled up on the same chip by only adjusting one of the pump wavelengths. In addition, we give the on-chip design of projection measurement for characterizing GHZ states and show the reconfigurability of the state. Our proposal is rather simple and feasible within the existing fabrication technologies and we believe it will boost the development of multiphoton technologies.