SparseArray/SparseVector.compact should be split into 2-3 functions · scalanlp/breeze#422

(0 评论) (0 反应) (0 负责人)Scala (3,453 star) (690 fork)batch import

good first issuehelp wanted

描述

Currently, SparseArray/SparseVector.compact does two separate things which are usually not intended to be done together. It should be split into three functions:

1: Add a def filter(predicate: T => Boolean): Unit function that allows specifying a predicate which is then used to strip certain values from the array/vector without reallocation.

2: Add a function called def tailor(): Unit that cuts down the memory usage to whatever is required to hold the values.

3: For backwards compatibility, replace the function compact with a function that calls filter(x => x == zero.zero) and then tailor().

贡献者指南

技术栈: scala
领域: dataperformance
议题类型: feature
难度: 3
预计时间: 1-2 days
活动状态: stale
清晰度: clear
前置要求: Scala knowledgeFamiliarity with sparse data structures
新手友好度: 65
研究方向: Examine the current implementation of SparseArray.compact and SparseVector.compact in the breeze source code (likely under breeze/src/main/scala/breeze/linalg/). Understand how it combines filtering zeros and memory tailoring. Implement a new filter(predicate) method, a tailor() method, and redefine compact to call both. Ensure backward compatibility. Look at existing tests for guidance.