Test Problem Generator by Neural Network for Algorithms that Try Solving Nonlinear Programming Problems Globally

A test problem generator, by means of neural networks nonlinear function approximation capability, is given in this paper which provides test problems, with many predetermined local minima and a global minimum, to evaluate nonlinear programming algorithms that are designed to solve the problem globally.     

神经网络在非线性规划中的应用

本文对蔡少棠[1 ] ,[2 ] 等提出的非线性规划神经网络作了改进 ,使得该神经网络适合于通用VLSI技术实现 .     

TE011腔使用扁平样品管提高水溶液样品ESR波谱灵敏度的研究

实验表明,在TE₀₁₁模式园柱形腔中,不但可以使用扁平样品管测量水溶液样品的ESR波谱,而且获得好的腔品质因子Q和高品质的ESR信号。所观察到的信号高度,是使用最佳尺寸毛细管样品时最大信号高度的3.2倍;又是在具有波导尺寸的TE₁₀₂腔中使用扁平样品管时所获得信号高度的3倍。本文对此作了定性的理论分析。     

The statistics of random directed graphs

The analysis of random graphs developed by the author, principally as a model for polymerization processes, is extended to the case of directed random graphs, with models of neural nets in mind. The principal novelty of the directed case is the representation of the partition function by a complex rather than a real integral, and the replacement of simple maxima in asymptotic evaluations by an interesting form of saddle point.     

Finding a feasible flow in a strongly connected network

We give a linear-time algorithm to find a feasible flow in a strongly connected network with fixed supplies and demands, each summing to a common value that is at most the minimum arc capacity. This algorithm speeds up the Goldberg-Rao maximum flow method by a constant factor.     

Behavior of a quantum particle in contact with a classical heat bath

We investigate the behavior of a two-level quantum system in contact with a classical heat bath, e.g., a solute particle with internal degrees of freedom immersed in a solvent of massive particles. Using a combination of analytical and numerical methods, we obtain precise information about localization, time-displaced correlation functions, and the frequency-dependent susceptibility of such solute particles. We find that these quantities can have a strong dependence on the density of the solvent fluid, with the maximum changes from the behavior of the corresponding isolated quantum system occurring in many cases at very low densities. We compare the exact results with those obtained by path integral Monte Carlo. There is good agreement with the imaginary time correlations, but analytic continuation to real time proves elusive: even with the best numerical data on the former, we can only get very gross features of the latter.     

On Ito-Kurzweil-Henstock Integral and Integration-by-Part Formula

In this paper we derive the Integration-by-Parts Formula using the generalized Riemann approach to stochastic integrals, which is called the Ito-Kurzweil-Henstock integral.     

On the problem of crack system with an elliptic hole

The problem of cracks with an elliptic hole in an infinite plane is investigated. By introducing the fictitious loads on the hole edge and using the Muskhelishvili method, the problem is reduced to solving a system of mixed-type integral equations in which some are Fredholm equations but others Cauchy-type singular ones. A numerical method is suggested and can be used for the treatment of other similar equations. The numerical results for some typical examples are given, showing that the method is really effective.     

Integral models of representations of current groups

We suggest a new construction of nonlocal representations of the current group. Instead of the Fock space, which is usually used in this situation, we consider the direct integral of countable tensor products of representations over the trajectories of some stochastic process. The construction substantially uses the invariance of the so-called infinite-dimensional Lebesgue measure.