Scholarlink

Locating a good initial point for a local iterative optimization method is an important task since they are strongly dependent on it. In this paper, a pre-processing step that exploits time series forecasting is proposed to deal with this problem. The suggested step could work alongside any local method and has the advantage of increasing the percentage of initial points that lead to the global minimizer, as shown by the numerical results. The proposed technique is compared to the corresponding utilized local method for a variety of well-known, as well as randomly generated, test objective functions with promising performance.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Time Series Forecasting: Initializing Line Search Methods for Unconstrained Optimization

  • 章节

  • 发表日期：

    2026年

查看全文
  • Athanasia N. Papanikolaou,
  • Theodoula N. Grapsa,
  • Christina D. Nikolakakou,
  • George S. Androulakis

摘要

Locating a good initial point for a local iterative optimization method is an important task since they are strongly dependent on it. In this paper, a pre-processing step that exploits time series forecasting is proposed to deal with this problem. The suggested step could work alongside any local method and has the advantage of increasing the percentage of initial points that lead to the global minimizer, as shown by the numerical results. The proposed technique is compared to the corresponding utilized local method for a variety of well-known, as well as randomly generated, test objective functions with promising performance.