求解大规模矛盾方程组的最小二乘支持向量机算法

房价预测、共享单车出租数量预测、空气污染情况预测等常涉及矛盾方程组求解,对其数值求解方法研究具有重要的理论意义与应用价值。当矛盾方程组规模过大时,用传统的最小二乘法求解,不仅计算量大,而且由于误差积累使最终结果的准确性不高。鉴于此,采用机器学习中的最小二乘支持向量机（least squares support vector machine,LS-SVM）算法求解大规模矛盾方程组,并分别针对线性、非线性、单变量、多变量矛盾方程组进行了数值求解。数值结果表明,数据类型和数据量的变化对结果的影响不大,因此只要选取适当的参数就可建立合适的模型,得到高精度的预测结果。     

敏感性问题调查方法的探讨

本文探讨了敏感性问题的四种调查方法，并结合实例进行比较，最后，对各种方法的优劣和应用范围提出作者的见解     

Criterion for a unique non-relativistic quark model

We have calculated the mean square charge radii of the neutron and proton, and compared them with the experimental values, to construct a unique non-relativistic quark potential model. It is shown for the first time that, contrary to general belief, it is possible to reproduce simultaneously the baryon mass spectrum and the electromagnetic sizes of neutron and proton using a single potential model. It was found necessary to add admixtures of excited states of the nucleon in the unperturbed ground state wavefunctions.     

ON THE CONCAVE φ-INEQUALITIES FOR NONNEGATIVE SUBMARTINGALES

Let φ1, φ2 be nonnegative nondecreasing functions, and φl be concave. The authors prove the equivalence of the following two conditions:(i) Eφ1 (Mf) ≤ cEφ2(Zo+A∞) for every nonnegative submartingale f = (fn)n≥0 with it's Doob's Decomposition: f = Z + A, where Z is a martingale in L1 and A is a nonnegative incrasing and predictable process.(ii) There exists positive constants c, to such that ft∞φ1(s)/s2 ds ≤ c φ2(t)/t, t ＞ to.When φ1 = φ2 the condition (ii) above is equivalent to the classical condition -Pφ＜1. As a consequence, for a concave function φ, -Pφ＜ 1 if and only if Eφ1 (Mf) ≤ cEφ2 (Z0 + A∞)for every nonnegative submartingale f.     

Admissible Square Quadruplets and Semisimple Symmetric Spaces

We show that there is a one to one correspondence between the set of semisimple locally symmetric spaces and the set of admissible square quadruplets. This correspondence leads to a new approach in the classification and a simple parametrization of irreducible locally symmetric spaces.     

Finiteness of the strong global dimension of radical square zero algebras

The strong global dimension of a finite dimensional algebra A is the maximum of the width of indecomposable bounded differential complexes of finite dimensional projective A-modules. We prove that the strong global dimension of a finite dimensional radical square zero algebra A over an algebraically closed field is finite if and only if A is piecewise hereditary. Moreover, we discuss results concerning the finiteness of the strong global dimension of algebras and the related problem on the density of the push-down functors associated to the canonical Galois coverings of the trivial extensions of algebras by their repetitive algebras.     

用最小二乘法测弹簧的有效质量

在气垫层轨上组装一简谐振动系统，利用最小二乘法测出弹簧的有效质量，并与理论值进行比较。     

The research of near field acoustical holographic imaging method

This paper reviews the theoretical analysis of the imaging method andcharacteristic of the Nearfield Acoustical Holography(NAH).We present anew filtering method,least squares method for filtering with restraint,whichimproves the stability of the reconstruction.The computer simulation experi-ment for the vibration and radiation of a round plate by means of the numericalmethod of NAH has been done and the effect of the new filtering method hasbeen verified.The NAH imaging of radiation of a small loudspeaker using ascanning linear receiving array of 32 elements has been made and the prelimina-ry results are given in the paper.     

Tracer dynamics in Dyson's model of interacting Brownian particles

We prove that the mean square displacement of a tracer particle grows as logt for larget. We point out a connection to the low-temperature floating phase of the ANNNI model.     

A necessary and sufficient convergence condition of orthomin(k) methods for least squares problem with weight

A class of Orthomin-type methods for linear systems based on conjugate residuals is extended to a form suitable for solving a least squares problem with weight. In these algorithms a mapping matrix as preconditioner is brought into use. We also give a necessary and sufficient condition for the convergence of the algorithm. Furthermore, we also study the construction of the mapping matrix for which the necessary and sufficient condition holds.