Hyperparameter on-line learning of stochastic resonance based threshold networks

Aiming at training the feed-forward threshold neural network consisting of nondifferentiable activation functions, the approach of noise injection forms a stochastic resonance based threshold network that can be optimized by various gradient-based optimizers. The introduction of injected noise extends the noise level into the parameter space of the designed threshold network, but leads to a highly non-convex optimization landscape of the loss function. Thus, the hyperparameter on-line learning procedure with respective to network weights and noise levels becomes of challenge. It is shown that the Adam optimizer, as an adaptive variant of stochastic gradient descent, manifests its superior learning ability in training the stochastic resonance based threshold network effectively. Experimental results demonstrate the significant improvement of performance of the designed threshold network trained by the Adam optimizer for function approximation and image classification.     

Full counting statistics of transport electrons through a two-level quantum dot with spin–orbit coupling

We study the full counting statistics of transport electrons through a semiconductor two-level quantum dot with Rashba spin–orbit (SO) coupling, which acts as a nonabelian gauge field and thus induces the electron transition between two levels along with the spin flip. By means of the quantum master equation approach, shot noise and skewness are obtained at finite temperature with two-body Coulomb interaction. We particularly demonstrate the crucial effect of SO coupling on the super-Poissonian fluctuation of transport electrons, in terms of which the SO coupling can be probed by the zero-frequency cumulants. While the charge currents are not sensitive to the SO coupling.     

Integral Kernel Operators on Fock Space – Generalizations and Applications to Quantum Dynamics

A general theory of operators on Boson Fock space is discussed in terms of the white noise distribution theory on Gaussian space (white noise calculus). An integral kernel operator is generalized from two aspects: (i) The use of an operator-valued distribution as an integral kernel leads us to the Fubini type theorem which allows an iterated integration in an integral kernel operator. As an application a white noise approach to quantum stochastic integrals is discussed and a quantum Hitsuda–Skorokhod integral is introduced. (ii) The use of pointwise derivatives of annihilation and creation operators assures the partial integration in an integral kernel operator. In particular, the particle flux density becomes a distribution with values in continuous operators on white noise functions and yields a representation of a Lie algebra of vector fields by means of such operators.     

On a stochastic nonlinear equation in one-dimensional viscoelasticity

In this paper we discuss an initial-boundary value problem for a stochastic nonlinear equation arising in one-dimensional viscoelasticity. We propose to use a new direct method to obtain a solution. This method is expected to be applicable to a broad class of nonlinear stochastic partial differential equations.

     

简谐速度噪声及其与势场之间的频率共振

为了描述复杂的噪声环境，考虑了一种具有频率结构的噪声——简谐速度噪声，包括它的产生、关联函数、功率谱以及作为热噪声时的频率特性所导致的一些行为.结果表明：在频谱空间中简谐速度噪声是一种带通噪声，存在一个峰值频率，且噪声带宽由参量Γ控制.当简谐势中的一个布朗粒子受热简谐速度噪声驱动时，粒子能量极大值出现在两种频率相等的情况下.这表明噪声和势场的频率之间存在动力学共振，决定着粒子能量的大小. 关键词： 简谐噪声 简谐速度噪声 功率谱 频率共振     

天文斑点成像技术中的噪声影响分析

为进一步说明天文像复原新方法-迭代位移叠加法的性质，对探测器的附加噪声在斑点干涉术、斑点掩模法和迭代位移叠加法中的影响分别进行了分析。结果表明：在斑点掩模法中，复杂的噪声高阶矩统计被引入了重谱的噪声偏差，因此对噪声极为敏感，使数据处理更加复杂化；与斑点掩模法相比，迭代位移叠加法仅涉及对噪声的一阶矩统计平均，所以此新方法的特点是对噪声不敏感，具有图像复原处理过程简单快捷的优越性。     

利用振幅调制器进行光电负反馈抑制激光强度噪声

采用振幅调制器作为抑制激光强度噪声的元件，首先对光电负反馈回路进行了简要的理论分析，然后利用该光电负反馈进行抑制激光二极管抽运全固化单频环形Nd:YVO4红外激光器强度噪声的实验。结果表明，在0－1MHz的低频段，强度噪声大幅度降低，最大降低约15dB。     

GAUSSIAN WHITE NOISE CALCULUS OF GENERALIZED EXPANSION

A new framework of Gaussian white noise calculus is established, in line with generalized expansion in [3, 4, 7]. A suitable frame of Fock expansion is presented on Gaussian generalized expansion functionals being introduced here, which provides the integral kernel operator decomposition of the second quantization of Koopman operators for chaotic dynamical systems, in terms of annihilation operators dt and its dual, creation operators t*.     

The measurement of subcritical electroconvection

Under idealized conditions, electroconvection in a nematic liquid crystal sets in at a well-defined threshold of the driving ac voltage. Fluctuating convection rolls of a small amplitude below that threshold have been observed recently. The measurement of the amplitude of these fluctuations is described in detail in this paper. It is based on averaging the light intensity signals using the structure function and a quantitative analysis of the light deflection.