摘要：目的比较三种传统优化方法和遗传算法基于中心复合设计的单目标优化时的特点。方法利用提高木聚糖酶产量的结果,分别用直接法、等高线法、最速上升法、遗传算法寻找最优结果,比较后得出结论。结果直接法、最速上升法、等高线法、遗传算法寻优的木聚糖酶含量分别为27.77UA/ml、23.93UA/ml、47.07UA/ml、51.04UA/ml。结论直接法可在其设计水平上较快找出结果,但不能保证局部或是全局达最优;在星点设计基础上的单目标寻优时等高线法的结果和遗传算法相当,等高线法可观察到部分因素对响应的影响大小,但存在局部最优和主观性较强的不足;最速上升法易陷入局部最优而忽略其他可能最优解。

摘要：A three-year research project based in British Columbia, Canada, is attempting to develop a framework and tools to assist healthcare system decision-makers achieve “high performance” in resource allocation. In pursuit of this objective, a literature search was conducted and two phases of primary data collection are being undertaken: an online survey of senior healthcare decision-makers, and in-depth case studies of potential “high performing” organizations. This paper addresses the survey phase;our aim is to provide a practical example of the mechanics of survey design, of benefit to those who want to better understand our forthcoming results, but also as an aid to other researchers grappling with the hard choices and trade-offs involved in the survey development process. Survey content is described in light of the existing literature, with discussion of the choices made by the research team to decide what questions and items would be included and excluded. The target population for the survey was senior managers in Canadian regional health authorities (or the closest equivalent organizations) in each of the 10 provinces and 3 territories. The paper dis- cusses how this sample was obtained, and describes the survey implementation process.