动态联合库存的决策模型及算法

考虑生产商、销售商联合库存的动态优化问题,建立的随机需求下生产-销售运作系统的排队模型,得到了系统的稳态概率分布和队长分布.以成本最小化为目标,模型算法找到了最优的运作策略和机器使用数量.数值模拟的结果表明,依赖于指定机器数量的动态调整策略明显优于静态系统.     

An existence theorem of Walrasian equilibrium

The purpose of this note is to establish an existence theorem for Walrasian equilibrium under weaker conditions.     

深圳市供水量的最优组合预测

城市供水系统是一个复杂的大系统,供水量受多种因素的共同影响。本文以深圳市最近20多年的供水量历史数据为基础,建立了深圳市供水量的最优组合预测模型。该模型具有较高的预测精度,组合预测的预测效果优于任意一种单一预测的预测效果,供水量预测结果对深圳市未来供水的短期或长期规划能起到重要的宏观指导作用。     

A sensitive, rapid ferricyanide-mediated toxicity bioassay developed using Escherichia coli

A need for rapid toxicity techniques has seen recent research into developing new microbiological assays and characterising their toxicity responses using a range of substances. A microbiological bioassay that determines changes in ferricyanide-mediated respiration for toxicity measurement (FM-TOX) shows particular promise. The development and optimisation of an improved FM-TOX method, incorporating novel features, is described using Escherichia coli as the biocatalyst. Omission of an exogenous carbon source, used in previously described FM-TOX assays, substantially improves the assay sensitivity. In addition, the development of a two-step procedure (toxicant exposure followed by determination of microbial respiratory activity) was found to enhance the inhibition of E. coli by 3,5-dichlorophenol and four other toxicants, compared to single-step procedures. Other assay parameters, such as the ferricyanide concentration, exposure times and optical density of the biocatalyst were also optimised, sometimes based on practical aspects. Toxicity tests were carried out using the adopted technique on both organic and inorganic toxicants, with classic sigmoid-shaped dose-response curves observed, as well as some non-standard responses. IC₅₀ data is presented for a number of common toxicants. The optimised assay provides a good foundation for further toxicity testing using E. coli, as well as the potential for expanding the technique to utilise other bacteria with complementary toxicity responses, thereby allowing use of the assay in a range of applications.     

MANAGEMENT OF A RENEWABLE RESOURCE WITH A FINITE DEMAND ELASTICITY

Optimal management policies are derived for a renewable resource when the demand for output has finite elasticity. The analysis modifies traditional views of this problem by allowing for the possibility that static optimizing policies, which maximize myopic net revenues, may be superior to “Golden Rule” policies which account for longer-run stock effects.     

基于GIS的杭州市地面水有机污染可视化研究

通过利用地理信息系统（GIS）的空间分析功能，对杭州市地面水中的有机污染现状（用高锰酸盐指数CODMn表示）进行区域分析，使环境信息数据定性、定位、可视，并能便捷地找出环境信息在时间上和空间上所呈现的规律性，为GIS深入应用于环境科学领域作了有益的尝试。     

带常数项的最优加权组合预测

We propose a model based on the optimal weighted combinational forecasting with constant terms,give formulae of the weights and the average errors as well as a rela- tion of the model and the corresponding model without constant terms,and compare these models.Finally an example was given,which showed that the fitting precision has been enhanced.     

产销平衡运输问题的表上作业法解法的一个注记

本文给出了用表上作业法求解产销平衡运输问题当出现退化时在相应空格填“O”的更为明确的规则，利用该规则可以避免可能存在的多余计算。本文还给出了用改进后的表上作业法求解指派问题的方法和步骤，该方法与求解指派问题的常用方法“匈牙利法”相比，具有手工计算更为简便的优点。     

基于分数阶差分的ARFIMA模型及预测效果研究

采用MRS分析法对香港恒生指数周数据序列的长期记忆性进行研究,并建立ARFIMA模型,推导了分数阶差分的计算过程。对分数阶差分的ARFIMA模型与一阶差分的ARFIMA模型进行了比较,发现应进行分数阶差分的序列,简化成一阶差分后,就有可能丢失许多有价值的信息,导致建模误差增大。进一步使用ARFIMA模型预测公式进行预测,结果显示ARFIMA模型预测效果不理想。在对香港恒生指数周数据进行预测时,ARFIMA模型几乎是失效的,并从两个不同的角度论证了这一结果出现的必然性。     

A two-echelon inventory model with lost sales

This paper considers a single-item, two-echelon, continuous-review inventory model. A number of retailers have their stock replenished from a central warehouse. The warehouse in turn replenishes stock from an external supplier. The demand processes on the retailers are independent Poisson. Demand not met at a retailer is lost. The order quantity from each retailer on the warehouse and from the warehouse on the supplier takes the same fixed value Q, an exogenous variable determined by packaging and handling constraints. Retailer i follows a (Q, R_i) control policy. The warehouse operates an (SQ, (S − 1)Q) policy, with non-negative integer S. If the warehouse is in stock then the lead time for retailer i is the fixed transportation time L_i from the warehouse to that retailer. Otherwise retailer orders are met, after a delay, on a first-come first-served basis. The lead time on a warehouse order is fixed. Two further assumptions are made: that each retailer may only have one order outstanding at any time and that the transportation time from the warehouse to a retailer is not less than the warehouse lead time. The performance measures of interest are the average total stock in the system and the fraction of demand met in the retailers. Procedures for determining these performance measures and optimising the behaviour of the system are developed.