A Modified Quasi-Newton Method for Structured Optimization with Partial Information on the Hessian

This paper develops a modified quasi-Newton method for structured unconstrained optimization with partial information on the Hessian, based on a better approximation to the Hessian in current search direction. The new approximation is decided by both function values and gradients at the last two iterations unlike the original one which only uses the gradients at the last two iterations. The modified method owns local and superlinear convergence. Numerical experiments show that the proposed method is encouraging comparing with the methods proposed in [4] for structured unconstrained optimization Presented at the 6th International Conference on Optimization: Techniques and Applications, Ballarat, Australia, December 9–11, 2004     

基因芯片荧光图象采集与分析

描述了基因芯片荧光图象的光学共聚焦成象,大范围高速扫描与控制,数据采集的原理和方法.并讨论了基因芯片荧光图象的分析算法,即模板定位与阈值分割相结合的半自动算法,并将其计算结果与完全采用人工分割算法的结果进行了比较.     

Algorithms for flattened Gaussian beams passing through apertured and unapertured paraxial ABCD optical systems

The propagation property of flattened Gaussian beams (FGBs) is studied based on the Collins formula, and analytical propagation equations for FGBs with and without an aperture are derived. It is shown that the closed-form expressions for the FGB of order N passing through paraxial ABCD optical systems without an aperture can be written as the combinations of Confluent Hyper-geometric Functions, and that the analytical equations for the FGB of order N passing through apertured paraxial ABCD optical systems may involve Error Functions. The proposed algorithms for the considered beams can save computing time, provide intuitive physical insights, and help to design optical systems. Our results are confirmed and illustrated with numerical examples.     

T 1 theorem for Besov spaces on nonhomogeneous spaces

Suppose μ is a Radon measure on ℝ ^d , which may be non doubling. The only condition assumed on μ is a growth condition, namely, there is a constant C₀>0 such that for all x∈supp(μ) and r>0, μ(B(x, r))⪯C₀rⁿ, where 0<n⪯d. We prove T1 theorem for non doubling measures with weak kernel conditions. Our approach yields new results for kernels satisfying weakened regularity conditions, while recovering previously known Tolsa’s results. We also prove T1 theorem for Besov spaces on nonhomogeneous spaces with weak kernel conditions given in [7].     

Substructuring preconditioners for saddle-point problems arising from Maxwell's equations in three dimensions

This paper is concerned with the saddle-point problems arising from edge element discretizations of Maxwell's equations in a general three dimensional nonconvex polyhedral domain. A new augmented technique is first introduced to transform the problems into equivalent augmented saddle-point systems so that they can be solved by some existing preconditioned iterative methods. Then some substructuring preconditioners are proposed, with very simple coarse solvers, for the augmented saddle-point systems. With the preconditioners, the condition numbers of the preconditioned systems are nearly optimal; namely, they grow only as the logarithm of the ratio between the subdomain diameter and the finite element mesh size.

     

Deposition of Hydrogen-Free Silicon Nitride Thin Films by Microwave ECR plasma Enhanced Magnetron Sputtering at Room Temperature

Hydrogen-free silicon nitride （SiNx） films were deposited at room temperature by microwave electron cyclotron resonance （MW-ECR） plasma enhanced unbalance magnetron sputtering system. Both Fourier-transform infrared spectroscopy and x-ray photoelectron spectroscopy are used to study the bonding type and the change of bonding structures of the silicon nitride films. The results indicate that the chemical structure and composition of SiNx films deposited by this technique depend strongly on the N2 flow rates, the stoichiometric SiNx film, which has the highest hardness of 22.9 GPa, could be obtained at lower N2 flow rate of 4 sccm.     

Sobolev-Hardy不等式与临界双重调和问题

该文讨论一类带有奇异系数的双重调和方程{△^2u-μu/|x|^s=f(x,u),x∈Ω,u=δu/δv=0,x∈δΩ，这里Ω包含R^N是包含0的有界光滑区域，u∈H0^2(Ω），μ∈R是参数，0≤s≤2,△^2=△△表示双重拉普拉斯算子，当f(x,u)=u^p,p=2N/N-4时，上述问题就是一个临界双重调和问题，该文运用Sobolev-Hardy不等式和变分方法，得到它的解的存在性的一些结果。     

3,7-二硝基亚氨基-2,4-二氮杂双环[3,3,0]辛烷的放热分解反应动力学

在程序升温条件下 ,用DSC研究了标题化合物的放热分解反应动力学 .用线性最小二乘法、迭代法以及二分法与最小二乘法相结合的方法 ,以积分方程、微分方程和放热速率方程拟合DSC数据 .在逻辑选择建立了微分和积分机理函数的最可几一般表达式后 ,用放热速率方程得到相应的表观活化能 (Ea)、指前因子 (A)和反应级数 (n)的值 .结果表明 :该反应的微分形式的经验动力学模式函数、Ea 和A值分别为 (1-α) 0 .44、2 30 .4kJ/mol和 10 18.16s-1.借助加热速率和所得动力学参数值 ,提出了标题化合物放热分解反应的动力学方程 .该化合物的热爆炸临界温度为 30 2 .6℃ .上述动力学参数对分析、评价标题化合物的稳定性和热变化规律十分有用 .     

Crystallographic report: catena‐Aqua(2,2′‐bipyrimidine)lithium(I) perchlorate

A lithium(I) coordination polymer has been formed from LiClO₄ and the 2,2′‐bipyrimidine (bpym) ligand in which each square pyramidal lithium(I) atom is coordinated in the basal plane by four nitrogen donor atoms derived from two bpym ligands and one water molecule at the apical position. These are connected into a layer structure via hydrogen‐bonding interactions involving the perchlorate anions. Copyright © 2004 John Wiley & Sons, Ltd.     

Generation of UV laser light by stimulated Raman scattering in D2, D2/Ar and D2/He using a pulsed Nd:YAG laser at 355nm

A pulsed Nd:YAG laser at 355nm is used to pump Raman cell filled with D_2, D_2/Ar and D_2/He. With adequately adjusted parameters, the maximum photon conversion efficiency of the first-order Stokes light (S_1, 396.796nm) reaches 33.33% in D_2/Ar and the stability of S_1 in pure D_2 is fairly high, the energy drift being less than 10% when the pump energy drifts in the range of 5%. The conversion efficiency and stability, which are functions of the composition and pressure of the Raman medium and the energy of pump laser, are investigated. The result has been used to optimize the laser transmitter system for a differential absorption lidar system to measure NO_2 concentration profiles.