Scholarlink

The Analog Ensemble (AnEn) method is used to reconstruct incomplete time series, particularly in meteorological contexts. However, the large volume of data and the need for repeated searches for nearest neighbours makes the method computationally intensive. This study explores the k-d tree data structure as an efficient alternative to accelerate the analog search process within the AnEn framework. By organising high-dimensional data hierarchically, the k-d tree significantly reduces search time. CPU-based implementations in the C language of the classical AnEn and k-d tree-based AnEn method were developed and evaluated. The results show that even in single-threaded execution, the k-d tree approach delivers substantial efficiency gains without compromising reconstruction accuracy or introducing significant memory overhead.

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

Using k-d Trees to Leverage the Performance of the Analog Ensemble Method

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Allan Clementino,
  • Claudio Schepke,
  • Carlos Balsa,
  • José Rufino

摘要

The Analog Ensemble (AnEn) method is used to reconstruct incomplete time series, particularly in meteorological contexts. However, the large volume of data and the need for repeated searches for nearest neighbours makes the method computationally intensive. This study explores the k-d tree data structure as an efficient alternative to accelerate the analog search process within the AnEn framework. By organising high-dimensional data hierarchically, the k-d tree significantly reduces search time. CPU-based implementations in the C language of the classical AnEn and k-d tree-based AnEn method were developed and evaluated. The results show that even in single-threaded execution, the k-d tree approach delivers substantial efficiency gains without compromising reconstruction accuracy or introducing significant memory overhead.