前馈神经网络隐性单调与泛化能力

应用前馈人工神经网络对广域单调的两组样本进行了模拟反演，引入单调前馈网络的概念对其权值和阈值定解问题和泛化能力进行了较诉研究。表明前馈人工神经网络是一个表达形式简单的复杂系统，其单调特征是隐性的，而且训练网络的成熟性对样本数量和样本内在规律性有一定依赖。强调了前馈人工神经网络的应用效果，指出单调与复合问题还需进一步深入研究。     

重力学反问题中零外部位密度的研究

对重力学反问题研究中零外部位密度的数学性质及物理意义进行了系统研究和分析．首先从基本定义及条件的描述入手，推演出其数学表示和限制，由此揭示重力学反问题的非适定性形态，进而论及其与非零外部位密度之间的数学关系．由此证明对应于给定外部位的场源函数可以分解成零外部位密度与调和密度的直和．通过对其物理意义的讨论，为地球物理反演非零外部位场源及物理大地测量学构制等效场源提供理论依据．     

反演二维弹性波动方程组多参数的渐近展开法

采用摄动法对二维弹性波动方程组的系数反问题进行渐近近似,给出了反演系数的具体计算方法。用摄动法解微分方程组的系数反问题还是首次。此外,还采用了地球物理分层模型,这种分层模型便于更精细地反映地球不同圈层的物理特征,更接近于实际情况。同时对进一步提高反演分辨率、增加反演参数、对病态方程组如何克服不适定性等问题,作了具体阐述,具有重要的理论意义与应用价值。     

Impulsive time-free transfers between halo orbits

A methodology is developed to design optimal time-free impulsive transfers between three-dimensional halo orbits in the vicinity of the interior L ₁ libration point of the Sun-Earth/Moon barycenter system. The transfer trajectories are optimal in the sense that the total characteristic velocity required to implement the transfer exhibits a local minimum. Criteria are established whereby the implementation of a coast in the initial orbit, a coast in the final orbit, or dual coasts accomplishes a reduction in fuel expenditure. The optimality of a reference two-impulse transfer can be determined by examining the slope at the endpoints of a plot of the magnitude of the primer vector on the reference trajectory. If the initial and final slopes of the primer magnitude are zero, the transfer trajectory is optimal; otherwise, the execution of coasts is warranted. The optimal time of flight on the time-free transfer, and consequently, the departure and arrival locations on the halo orbits are determined by the unconstrained minimization of a function of two variables using a multivariable search technique. Results indicate that the cost can be substantially diminished by the allowance for coasts in the initial and final libration-point orbits.An earlier version was presented as Paper AIAA 92-4580 at the AIAA/AAS Astrodynamics Conference, Hilton Head Island, SC, U.S.A., August 10–12, 1992.     

Adiabatic invariants for the nonconservative Kepler's problem

This paper considers adiabatic invariants for the classical Kepler problem with resisting forces. The analysis is based on the theory of integrating factors and theory of adiabatic invariants in the Krylov-Bogoliubov-Mitropolski variables. The adiabatic invariants are series with respect to a small parameter. Also, for every particular case of nonconservative forces, it is shown that, with a complete set of adiabatic invariants, an approximate solution of the problem can be obtained. Four problems are analyzed in detail where approximate solutions are compared with numerical.     

On the global solution of the N-body problem

In connection with the publication (Wang Qiu-Dong, 1991) the Poincaré type methods of obtaining the maximal solution of differential equations are discussed. In particular, it is shown that the Wang Qiu-Dong'sglobal solution of the N-body problem has been obtained in Babadzanjanz (1979). First the more general results on differential equations have been published in Babadzanjanz (1978).     

Conditioning mean steady state flow on hydraulic head and conductivity through geostatistical inversion

Nonlocal moment equations allow one to render deterministically optimum predictions of flow in randomly heterogeneous media and to assess predictive uncertainty conditional on measured values of medium properties. We present a geostatistical inverse algorithm for steady-state flow that makes it possible to further condition such predictions and assessments on measured values of hydraulic head (and/or flux). Our algorithm is based on recursive finite-element approximations of exact first and second conditional moment equations. Hydraulic conductivity is parameterized via universal kriging based on unknown values at pilot points and (optionally) measured values at other discrete locations. Optimum unbiased inverse estimates of natural log hydraulic conductivity, head and flux are obtained by minimizing a residual criterion using the Levenberg-Marquardt algorithm. We illustrate the method for superimposed mean uniform and convergent flows in a bounded two-dimensional domain. Our examples illustrate how conductivity and head data act separately or jointly to reduce parameter estimation errors and model predictive uncertainty.This work is supported in part by NSF/ITR Grant EAR-0110289. The first author was additionally supported by scholarships from CONACYT and Instituto de Investigaciones Electricas of Mexico. Additional support was provided by the European Commission under Contract EVK1-CT-1999-00041 (W-SAHaRA-Stochastic Analysis of Well Head Protection and Risk Assessment).