Weight Hierarchies and Distributions in Linear Codes

This direction is split into other directions: weight distribution analysis. Its last version stays readable and citable here.

Version 1 of 165 papers108 conceptsfrozen 2026-10-04

In coding theory, the strength of an error-correcting code is traditionally measured by its minimum distance. However, a single number rarely captures the full story. To understand how a code performs under different noise conditions, or how it can be used for cryptography and data storage, researchers look at the weight distribution—the complete spectrum of how many non-zero elements exist across all possible codewords.

Recent research has shifted from simply counting these weights to characterizing the weight hierarchy, a more granular view that describes the structural "support" of entire subcodes.

The Geometry of Codeword Weights

The Hamming weight of a codeword is the number of its non-zero coordinates. The collection of these weights for all codewords is the weight distribution. Knowing this distribution allows researchers to calculate the exact probability of an undetected error.

A common construction method involves "defining sets" over finite fields. By choosing a specific set of elements and applying a trace function, researchers like Dabin Zheng and Cunsheng Ding have produced codes with only a few distinct weights (e.g., two-weight or three-weight codes). As shown in Two classes of p-ary linear codes and their duals, these "few-weight" codes are often optimal, meaning they achieve the highest possible minimum distance for a given length and dimension.

From Single Weights to Subcode Hierarchies

While the weight of a single codeword is useful, modern applications like secret sharing and distributed storage require understanding the Generalized Hamming Weight (GHW). Instead of looking at one codeword, GHW looks at an $r$-dimensional subcode and asks: what is the minimum number of coordinate positions where at least one codeword in that subcode is non-zero?

This sequence of weights—the weight hierarchy—acts as a structural fingerprint. For example, Jie Hao and Bin Chen demonstrated in Generalized Hamming Weights of (r, δ)-Locally Repairable Codes that these hierarchies determine how efficiently a distributed system can recover lost data. The hierarchy reveals the "locality" of the code—how many other nodes must be contacted to repair a failure.

The Dual Code as a Mirror

One of the most powerful tools in this field is the MacWilliams Identity. It provides a mathematical bridge between a code and its dual (the set of all vectors orthogonal to the code). Often, it is computationally impossible to count the weights of a large code, but its dual might be much smaller or simpler.

Ching–Yi Lai and colleagues extended this concept in On the MacWilliams Identity for Classical and Quantum Convolutional Codes, showing that the relationship holds even for complex quantum and convolutional structures. By treating the weight enumerator as a linear Fourier transform, researchers can "mirror" the known properties of a dual code to find the weight distribution of the primary code. This is often paired with Pless power moments, a set of identities that relate the moments of these distributions to verify a code's optimality.

Counting with Exponential Sums

To move from theoretical bounds to exact formulas, researchers use the "engine" of finite field arithmetic: exponential sums. The weight of a codeword is essentially the length of the code minus the number of zeros in a specific polynomial. Counting these zeros is equivalent to evaluating sums like Weil sums, Gauss sums, and Kloosterman sums.

For instance, Gaopeng Jian and Zhouchen Lin used Weil sums in Two-weight and three-weight linear codes based on Weil sums to transform abstract algebraic equations into concrete weight counts. Similarly, Minglong Qi and Shengwu Xiong utilized "twisted" Kloosterman sums to solve previously intractable counting problems, allowing for the design of codes with a predictable, small number of weights.

Beyond Hamming: New Metrics for New Media

As data storage moves toward ultra-high-density formats (like DNA storage or symbol-pair channels), the standard Hamming metric is being supplemented by the $b$-symbol weight and pair weight. These metrics account for the fact that errors often occur in adjacent clusters rather than isolated bits.

Hongwei Zhu and Minjia Shi have established bounds on the "weight spectrum" for these metrics, as seen in Some bounds on the cardinality of the b-symbol weight spectrum of codes. This research ensures that as the physical medium of storage changes, the mathematical framework for protecting that data evolves to match the new error patterns.

Applications in Combinatorial Design

A surprising byproduct of this research is the construction of Steiner systems and $t$-designs. These are highly symmetric arrangements of sets where every subset of a certain size appears exactly the same number of times.

By proving that the supports of codewords of a specific weight form a design, researchers like Can Xiang and Chunming Tang have used codes to solve long-standing problems in combinatorial geometry. As detailed in An infinite family of antiprimitive cyclic codes supporting Steiner systems, the weight distribution isn't just a performance metric—it is a map to deep structural symmetries in mathematics.

Papers in this version

Showing 60 of 65 papers.

History & citationVersion 1 · 2026-10-04

Split into other directions · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_weight_hierarchies_and_distributions_in__62d31457,
  title = {Weight Hierarchies and Distributions in Linear Codes},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/weight-hierarchies-and-distributions-in-linear-codes/rdr_62d3145722bc4534b7e4a996bcbfd05b},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_weight_hierarchies_and_distributions_in__62d31457_v1,
  title = {Weight Hierarchies and Distributions in Linear Codes},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/weight-hierarchies-and-distributions-in-linear-codes/rdr_62d3145722bc4534b7e4a996bcbfd05b/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}