平行式向列液晶实现相移剪切电子散斑干涉术的研究

研究了向列型液晶的相位调制特性，采用平行式向列液昌作为相移器建立了相移剪切电子散斑干涉计量系统。该方法将相移技术引入到剪切电子散斑干涉术中，提高了检测精度，采用了相关条纹法求解相位，简化了计算；相移方法简单，可靠。用该系统进行应变场的测量实验，取得了满意的效果。     

Line profile study by diode laser spectroscopy in the CH4ν2 + ν4 band

We present a complete study on four methane lines for two atmospheric micro-windows (in the ν₂ + ν₄ absorption band) used for the determination of atmospheric methane concentrations with ground-based Fourier transform spectrometers. Thanks to our tunable diode laser (TDL) spectrometer with active wavenumber control and step-by-step recording mode we have improved the accuracy on intensity, broadening, narrowing, and pressure shift parameters. To make our results directly useable in atmospheric models which usually assume a Voigt line shape, we have parameterised an effective-broadening parameter γ_Voigt (P) for each line and each gas mixture (CH₄-N₂ and CH₄-O₂). When this parameterisation is used to fit a “true” line profile, the same concentration as with more sophisticated models is retrieved using a consistent set of spectroscopic parameters in both approaches.     

相移法获取三维光弹特征相位差的方法

对于三维光弹模型,次主应力及其方向沿入射光连续变化,根据光学等效的原理,可以将三维光弹模型等效为一个包含线性相位延迟器和旋光器的光学模型。在集成光弹法中,对于每一个测量点,研究者必需通过实验获取三个特征参数。提出了一种用于确定集成光弹中全场特征相位差的五步相移法,给出了偏振光场中元器件的设置方法,并运用Jones算子推导出相移法得出的光强公式。最后使用斜射法对径向受压圆盘进行实验,验证了方法的有效性。     

基于白光干涉测试技术的改进Carré相移算法

为了减小白光相移干涉法测量物体微观形貌时产生的误差,对极值法的搜索路径进行优化后与Carré相移算法相结合,提出一种基于白光干涉测试技术的改进Carré相移算法.该算法减小了光源扰动及电荷耦合元件散粒噪声带来的影响,且对相移器的线性误差不敏感,免去了相位解包裹过程,提高了运算效率.采用单刻线样块对扫描步距进行校准实现了光源中心波长的在线修正,减小了由于光源特性及环境扰动误差带来的影响.采用不同算法对标准粗糙度样块进行三维形貌恢复以及表面粗糙度的重复性测量实验,结果表明:该算法对表面粗糙度的测量结果较传统白光干涉算法准确度提高,测量重复性优于1%.     

弱化限制等距性的稀疏信号恢复

众所周知,传统的信号压缩和重建遵循香农一耐奎斯特采样定律,即采样率必须至少为信号最高频率的两倍,才能保证在重建时不产生失真,这无疑将给信号采样,传输和存储过程带来越来越大的压力.随着科技的飞速发展,特别是近年来传感器技术获取数据能力提高,物联网等促使人类社会的数据规模遽增,大数据时代正式到来.大数据的规模效应给数据存储,传输,管理以及数据分析带来了极大的挑战.压缩采样应运而生.限制等距性(Restricted Isometry Property,RIP)在压缩传感中起着关键的作用.只有满足限制等距条件的压缩矩阵才能平稳恢复原始信号.RIP作为衡量矩阵是否能作为测量矩阵得到了认可,但是此理论的缺陷在于对任一矩阵,很难有通用,快速的算法来验证其是否满足RIP条件.很多学者尝试弱化RIP条件以找到测量矩阵构造的突破口.首先构造了新的限制等距条件δ_(1.5k)+θ_(k,1.5k)≤1,然后证明在这个条件下无噪声稀疏信号能被精确的恢复,并且噪声稀疏信号能被平稳的估计.最后,通过比较表明δ_(1.5k)+θ_(k,1.5k)≤1优于现存的条件.     

Blue-shifting hydrogen bond in the benzene-benzene and benzene-naphthalene complexes

Ab initio complete optimizations at MP2/6-31++G** level have been performed in the T-shaped geometry of the benzene-benzene and benzene-naphthalene complexes. To check the effect of the basis set superposition error (BSSE), optimizations have been done in the BSSE corrected and BSSE uncorrected potential energy surfaces. The BSSE effect in the calculation of the Hessian has also been evaluated to check its influence in the frequency values. Quantum theory atoms in molecules (QTAIM) calculations have also been performed on both dimers. Intermolecular energies differ around a 25% when the optimization is performed with or without counterpoise corrected gradients. The influence of BSSE is also noticeable in the distances. Frequency shifts show big changes because of the BSSE. Thus, uncorrected values are up 350% larger than corrected ones. The hypotheses given in the literature to explain the origin of the blue-shifting hydrogen bond do not seem to give a suitable explanation for all characteristics of the behavior found in the studied systems.     

Novel Convolutional Neural Network with Variational Information Bottleneck for P300 Detection

In the area of brain-computer interfaces (BCI), the detection of P300 is a very important technique and has a lot of applications. Although this problem has been studied for decades, it is still a tough problem in electroencephalography (EEG) signal processing owing to its high dimension features and low signal-to-noise ratio (SNR). Recently, neural networks, like conventional neural networks (CNN), has shown excellent performance on many applications. However, standard convolutional neural networks suffer from performance degradation on dealing with noisy data or data with too many redundant information. In this paper, we proposed a novel convolutional neural network with variational information bottleneck for P300 detection. Wiht the CNN architecture and information bottleneck, the proposed network termed P300-VIB-Net could remove the redundant information in data effectively. The experimental results on BCI competition data sets show that P300-VIB-Net achieves cutting-edge character recognition performance. Furthermore, the proposed model is capable of restricting the flow of irrelevant information adaptively in the network from perspective of information theory. The experimental results show that P300-VIB-Net is a promising tool for P300 detection.     

Stimulus‐Responsive Nanoparticles and Associated (Reversible) Polymorphism via Polymerization Induced Self‐assembly (PISA)

Polymerization‐induced self‐assembly (PISA) is an extremely versatile method for the in situ preparation of soft‐matter nanoparticles of defined size and morphologies at high concentrations, suitable for large‐scale production. Recently, certain PISA‐prepared nanoparticles have been shown to exhibit reversible polymorphism (“shape‐shifting”), typically between micellar, worm‐like, and vesicular phases (order–order transitions), in response to external stimuli including temperature, pH, electrolytes, and chemical modification. This review summarises the literature to date and describes molecular requirements for the design of stimulus‐responsive nano‐objects. Reversible pH‐responsive behavior is rationalised in terms of increased solvation of reversibly ionized groups. Temperature‐triggered order–order transitions, conversely, do not rely on inherently thermo‐responsive polymers, but are explained based on interfacial LCST or UCST behavior that affects the volume fractions of the core and stabilizer blocks. Irreversible morphology transitions, on the other hand, can result from chemical post‐modification of reactive PISA‐made particles. Emerging applications and future research directions of this “smart” nanoparticle behavior are reviewed.

     

The effect of moving bottlenecks on a two-lane traffic flow

In this paper, we study the effect of moving bottlenecks on traffic flow. The full velocity difference （FVD） model is extended to the traffic flow on a two-lane highway, and new lane changing rule is proposed to reproduce the vehicular lane changing behavior. Using this model, we derive the fundamental current-density diagrams for the traffic flow with the effect of moving bottleneck. Moreover, typical time-space diagram for a two-lane highway shows the formation and dissipation of a moving bottleneck. Results demonstrate that the effect of moving bottleneck enlarges with the increase of traffic density, but the effect can be reduced by increasing the maximum velocity of heavy truck. The effects of multiple moving bottlenecks under different conditions are investigated. The effect becomes more remarkable when the coupling effect of multiple moving bottlenecks occurs.     

Information Flows of Diverse Autoencoders

Deep learning methods have had outstanding performances in various fields. A fundamental query is why they are so effective. Information theory provides a potential answer by interpreting the learning process as the information transmission and compression of data. The information flows can be visualized on the information plane of the mutual information among the input, hidden, and output layers. In this study, we examine how the information flows are shaped by the network parameters, such as depth, sparsity, weight constraints, and hidden representations. Here, we adopt autoencoders as models of deep learning, because (i) they have clear guidelines for their information flows, and (ii) they have various species, such as vanilla, sparse, tied, variational, and label autoencoders. We measured their information flows using Rényi’s matrix-based α-order entropy functional. As learning progresses, they show a typical fitting phase where the amounts of input-to-hidden and hidden-to-output mutual information both increase. In the last stage of learning, however, some autoencoders show a simplifying phase, previously called the “compression phase”, where input-to-hidden mutual information diminishes. In particular, the sparsity regularization of hidden activities amplifies the simplifying phase. However, tied, variational, and label autoencoders do not have a simplifying phase. Nevertheless, all autoencoders have similar reconstruction errors for training and test data. Thus, the simplifying phase does not seem to be necessary for the generalization of learning.