[Paper Review] A Hierarchical-based Greedy Algorithm for Echelon-Ferrers Construction

Xianmang He|arXiv (Cornell University)|Nov 4, 2019

Coding theory and cryptography4 citations

TL;DR

This paper proposes a hierarchical-based greedy algorithm to enhance the construction of constant-dimension subspace codes via the echelon-Ferrers technique, significantly reducing computational overhead while achieving larger code sizes. It improves lower bounds for multiple subspace code parameters, including Aq(15; 6; 6), Aq(16; 6; 6), and Aq(16; 6; 7), with over 56 new record improvements exceeding previously known bounds.

ABSTRACT

Echelon-Ferrers is one of important techniques to help researchers to improve lower bounds for subspace code. But, unfortunately, heavy computation has been paid as the cost for the construction. In this paper, we show how to attain codes of larger size for a given minimum distance d by the greedy algorithm for echelon-Ferrers construction introduced in [1]. This algorithm allows us to improve the lower bounds for several types of constant-dimension subspace codes, including Aq(15; 6; 6), Aq(16; 6; 6), and Aq(16; 6; 7) etc. More than 56 new improvements and the expression of these bounds are given. All these bounds exceeds the current best bounds.

Motivation & Objective

• To reduce the computational cost of echelon-Ferrers-based subspace code constructions while maintaining or improving code size.
• To develop a scalable and efficient algorithm for constructing larger constant-dimension subspace codes with given minimum distance d.
• To improve existing lower bounds for Aq(n; d; k) codes, particularly for parameters like Aq(15; 6; 6), Aq(16; 6; 6), and Aq(16; 6; 7).
• To provide explicit expressions and improved values for previously unknown or suboptimal bounds in subspace code theory.

Proposed method

• The paper employs a hierarchical-based greedy algorithm to systematically explore and select subspaces within the echelon-Ferrers framework.
• It leverages structural properties of echelon-Ferrers partitions to guide the greedy selection process, reducing redundant computation.
• The algorithm prioritizes subspaces that maximize code size while preserving the minimum distance d.
• It integrates pruning strategies based on hierarchical constraints to limit the search space and improve efficiency.
• The method is applied iteratively to build larger codes from smaller components, maintaining the required distance properties.
• The approach is validated by computing and comparing bounds across multiple values of n, d, and k.

Experimental results

Research questions

• RQ1 Can a hierarchical-based greedy algorithm reduce the computational cost of echelon-Ferrers construction while improving code size?
• RQ2 What is the maximum achievable size of constant-dimension subspace codes for given parameters n, d, and k using this method?
• RQ3 How do the new bounds compare to existing ones, particularly for Aq(15; 6; 6), Aq(16; 6; 6), and Aq(16; 6; 7)?
• RQ4 What specific improvements are achieved across multiple code parameters using this algorithm?
• RQ5 Can the proposed method consistently exceed the current best-known lower bounds for subspace codes?

Key findings

• The proposed algorithm achieves over 56 new lower bound improvements for constant-dimension subspace codes, all exceeding previously known bounds.
• Significant improvements are reported for Aq(15; 6; 6), Aq(16; 6; 6), and Aq(16; 6; 7), with explicit expressions provided for these enhanced bounds.
• The hierarchical greedy approach reduces computational overhead compared to standard echelon-Ferrers constructions while increasing code size.
• The method successfully constructs larger codes by efficiently navigating the subspace search space using structural constraints.
• All new bounds are strictly better than the current best-known values, demonstrating the method's effectiveness.
• The algorithm's scalability and performance make it suitable for exploring higher-dimensional subspace codes with improved efficiency.