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.     

The energy-critical nonlinear wave equation with an inverse-square potential

We study the energy-critical nonlinear wave equation in the presence of an inverse-square potential in dimensions three and four. In the defocussing case, we prove that arbitrary initial data in the energy space lead to global solutions that scatter. In the focusing case, we prove scattering below the ground state threshold.     

双掺(Tm3+,Tb3+)LiYF4激光器1.5 μm波长激光阈值分析

由速率方程推出了双掺（Tm^3 ,Tb^3 ）离子准四能级系统的激光阈值解析式，讨论了Tm^3 和Tb^3 离子之间的相互作用。分析了1.5μm波长附近的激光阈值和Tm^3 、Tb^3 离子的掺杂原子数分数及晶体长度的关系。结果表明，对于对应Tm^3 离子^3H4→^3F4跃迁的约1.5μm波长的激光，激活离子Tm^3 的掺杂原子数分数过大时，交叉弛豫作用将使系统阈值迅速增加。Tb^3 离子的加入，一方面能抽空激光下能级，起到降低阈值的作用；另一方面亦减少了激光上能级的寿命，使阈值升高。故Tb^3 离子有最佳掺杂原子数分数。对于Tm原子数分数为y=0.01的Tm:LiYF4晶体，Tb^3 离子的最佳掺杂原子数分数为0.002左右，同时表明，激光阈值与晶体长度有关。最佳晶体长度与Tm^3 、Tb^3 离子的掺杂原子数分数以及晶体的衍射损耗和吸收损耗有关。     

Nonsensitiveness Regions for Threshold Ellipsoids

The problem is to determine nonsensitiveness regions for threshold ellipsoids within a regular mixed linear model.     

New universality classes in “Percolative” dynamics

We study a site analogue of directed percolation. Random trajectories are generated and their critical behavior is studied. The critical behavior corresponds to that of simple percolation in some of the parameter space, but elsewhere the exponents reveal new universality classes. As a byproduct, we use the model to make an improved estimate of the percolation hull exponents and to calculate the site percolation probability for the square lattice.     

光码分多址系统中光学相关接收机判决阈值的分析

介绍了光码分多址系统中常用地址码（一维扩时码、二维码和三维码）的特点，并对它们各自的互相关均值和方差进行了理论分析。基于非相干光码分多址系统中光学相关接收机的基本原理，结合不同的用户地址码，对系统误码率性能进行了分析，得到了接收机最佳判决阈值与地址码基本特性参数和系统同时用户数间的关系。最后，给出了数值仿真结果。结果表明，对于采用特定地址码的光码分多址系统，只有选择合适的接收机判决阈值，系统的误码率性能才能达到最佳。研究结果对光码分多址系统中接收机判决阈值的选取具有一定的参考作用。     

On the critical exponents of random k‐SAT

There has been much recent interest in the satisfiability of random Boolean formulas. A random k‐SAT formula is the conjunction of m random clauses, each of which is the disjunction of k literals (a variable or its negation). It is known that when the number of variables n is large, there is a sharp transition from satisfiability to unsatisfiability; in the case of 2‐SAT this happens when m/n → 1, for 3‐SAT the critical ratio is thought to be m/n ≈ 4.2. The sharpness of this transition is characterized by a critical exponent, sometimes called ν = ν_k (the smaller the value of ν the sharper the transition). Experiments have suggested that ν₃ = 1.5 ± 0.1. ν₄ = 1.25 ± 0.05, ν₅ = 1.1 ± 0.05, ν₆ = 1.05 ± 0.05, and heuristics have suggested that ν_k → 1 as k → ∞. We give here a simple proof that each of these exponents is at least 2 (provided the exponent is well defined). This result holds for each of the three standard ensembles of random k‐SAT formulas: m clauses selected uniformly at random without replacement, m clauses selected uniformly at random with replacement, and each clause selected with probability p independent of the other clauses. We also obtain similar results for q‐colorability and the appearance of a q‐core in a random graph. © 2002 Wiley Periodicals, Inc. Random Struct. Alg., 21: 182–195, 2002     

颜色视觉匹配中明度阈值的评价

为了对颜色视觉匹配中的明度阈值特性进行评价,采用在阴极射线管显示器上获得的颜色刺激,在5个不同明度的中性色背景下,标准刺激的明度L*从5到95以5个CIELAB单位为变化步长,共对19个不同明度的中性标准颜色进行视觉匹配实验。通过对实验结果的处理和分析,验证了关于颜色刺激视觉辨别明度阈值与背景明度之间相关性的韦伯法则和crispening效应,同时利用颜色视觉匹配数据对基于CIELAB的相关色差公式进行了评价和比较,CIEDE2000具有最好的明度差预测性能,CMC次之,CIELAB和CIE94最差。     

Existence and Uniqueness of Endemic States for the Age-structured MSEIR Epidemic Model

The existence and uniqueness of positive steady states for the age-structured MSEIR epidemic model with age-dependent transmission coefficient is considered. Threshold results for the existence of endemic states are established; under certain conditions, uniqueness is also shown.     

修饰态布居的选择性激发对无反转激光的作用

以三能级V型系统为例研究修饰态布居的选择性激发对无反转激光增益的作用. 当非 相干驱动场的频谱宽度远小于驱动场产生的修饰态能级的间距时，非相干驱动场只将一个修 饰态的布居抽运至激发态. 借助原子的衰减通道，系统中形成单向布居转移通道，从而建立 修饰态布居的选择性激发. 利用修饰态布居的选择性激发，可以摆脱裸态共振无反转激光的 三个限制: (1) 不再要求辅助的低频驱动跃迁比高频激光跃迁具有更高的衰减速率；(2) 显 著降低非相干激发速率的阈值；(3) 无反转激光的线性增益不再反比于相干驱动场的强 关键词： 修饰态布居的选择性激发 无反转激光增益 原子衰减速率 非相干激发阈值速率