🔢 mathematics

New bounds for codes over Gaussian integers based on the Mannheim distance

This paper establishes new theoretical bounds for linear codes over Gaussian integers under the Mannheim distance, including volume formulas, sphere packing limits, and MacWilliams-type identities for self-dual codes, while also presenting decoding algorithms and demonstrating the metric's advantage in correcting errors that are uncorrectable under the Hamming metric.

Original authors: Minjia Shi, Xuan Wang, Junmin An, Jon-Lark Kim

Published 2026-03-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Minjia Shi, Xuan Wang, Junmin An, Jon-Lark Kim

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are running a massive, high-speed delivery service in a city that doesn't look like a grid of streets, but rather like a giant, infinite checkerboard made of complex numbers. This is the world of Gaussian Integers.

In the old days, delivery drivers (data packets) only moved North, South, East, or West. If a package got lost, we counted how many blocks it was off using a simple "Manhattan" style count (just steps). This is called the Hamming Distance.

But modern technology (like your Wi-Fi or 4G/5G) sends data in a way that moves diagonally too. It's like a package can be thrown North-East or South-West. The old "Manhattan" counting method is terrible at measuring errors in this diagonal world. It's like trying to measure the distance between two points on a map using only a ruler that can only go up and down.

This paper is about inventing a new ruler called the Mannheim Distance that works perfectly for this diagonal, 2D world, and then using that ruler to build better, more reliable delivery systems (error-correcting codes).

Here is a breakdown of what the researchers did, using simple analogies:

1. The New Ruler: Mannheim Distance

Think of the Mannheim Distance as a "taxi fare" in a city where you can drive diagonally.

Old Way (Hamming): If you miss your destination by 1 block North and 1 block East, the old system says you are "2 blocks away" (1 step up + 1 step right).
New Way (Mannheim): The new system realizes that in this specific city, moving diagonally is efficient. It calculates the cost based on the actual "shape" of the error. Sometimes, a mistake that looks huge to the old ruler is actually a tiny, fixable error to the new ruler.

The Big Win: The paper shows that some errors that were previously "unfixable" (because the old ruler said they were too far away) are actually easy to fix with the new ruler. It's like realizing a package wasn't lost in a different city; it was just slightly off in the same neighborhood.

2. Packing Spheres (The Sphere Packing Bound)

Imagine you are trying to pack oranges (data messages) into a giant box (the communication channel).

Each orange needs a little "bubble" of empty space around it so that if the box gets shaken (noise/errors), the oranges don't smash into each other.
The Sphere Packing Bound is a mathematical rule that tells you the maximum number of oranges you can fit in the box without them touching.
The researchers calculated exactly how big these "bubbles" are in this new diagonal city. They found the absolute limit of how much data you can send reliably. If you try to send more, the bubbles overlap, and the receiver gets confused.

3. The "Perfect" Delivery System

A Perfect Code is the holy grail of delivery. It means you have packed the oranges so tightly that there is absolutely no wasted space in the box, yet they still don't touch.

The researchers asked: "Can we build a perfect system that fixes two mistakes at once?"
They did the math and found that such a system is incredibly rare. It only exists under very specific conditions (like a specific size of city).
They found that the smallest city where this "2-mistake fixer" could possibly exist is a field with 29 elements. They even gave the specific dimensions for this theoretical "perfect" system: a code of length 10 that can fix 2 errors.

4. Self-Dual Codes (The Mirror Image)

Imagine a code that is its own mirror image. If you flip the data, it looks exactly the same. These are called Self-Dual Codes. They are special because they are very symmetrical and often very strong.

The researchers used a special mathematical "mirror" (called a MacWilliams identity) to look at these codes through the new Mannheim lens.
They calculated the maximum strength (minimum distance) these mirror codes could possibly have.
They then went out and built actual examples of these codes that hit those maximums, proving their math was right. It's like saying, "Theoretically, a bridge can hold 10 tons," and then building a bridge that actually holds 10 tons.

5. The Decoder (The GPS for Errors)

Finally, they built a GPS for this new system.

When a message arrives with errors, the decoder looks at the "syndrome" (a clue left behind by the error).
In the old system, some clues were too confusing to solve.
In the new Mannheim system, the researchers showed that the decoder can solve these clues.
Example: They showed a scenario where a message had an error that the old system couldn't fix (it thought the error was too big). But the new system looked at the error, realized it was actually a small diagonal shift, and fixed it perfectly.

Summary: Why Does This Matter?

We live in a world of QAM (Quadrature Amplitude Modulation), which is the technology behind your Wi-Fi, 4G, and digital TV. This technology sends data in a 2D grid (diagonally).

Old Math: Treated this 2D grid like a 1D street. It wasted space and couldn't fix certain errors.
This Paper: Provides the correct math for the 2D grid. It tells engineers exactly how much data they can pack into a signal and how to fix errors that were previously impossible to correct.

In short, the authors took a complex, abstract mathematical problem and gave us a better map and a better ruler for the digital highways of the future.

1. Problem Statement

Classical coding theory primarily focuses on codes over finite fields using the Hamming distance. However, the Hamming distance (and the related Lee distance) is ill-suited for Quadrature Amplitude Modulation (QAM) signals, which are ubiquitous in modern communication systems (e.g., LTE, Wi-Fi, HSDPA). QAM constellations exist in a two-dimensional signal space where errors manifest as shifts in both real and imaginary components.

To address this, Huber introduced codes over Gaussian integers ( $\mathbb{Z}[i]$ ) equipped with the Mannheim distance. While significant work has been done on constructing such codes (e.g., OMEC codes, quantum codes, and self-dual codes), there is a lack of comprehensive theoretical bounds on the minimum Mannheim distance and a systematic study of optimal codes meeting these bounds. This paper aims to fill that gap by establishing Mannheim-metric analogues of classical coding bounds and investigating the existence of optimal codes.

2. Methodology

The authors employ a combination of algebraic number theory, combinatorial enumeration, and linear algebra to derive bounds and construct codes.

Algebraic Framework: The study focuses on linear codes over the quotient ring $G_\pi = \mathbb{Z}[i]/(\pi)$ $G_{π} = Z [i] / (π)$ , where $\pi$ $π$ is a Gaussian prime.
- For $p \equiv 3 \pmod 4$ , the ring is isomorphic to $\mathbb{F}_{p^2}$ .
- For $p \equiv 1 \pmod 4$ , the ring is isomorphic to $\mathbb{F}_p$ .
Metric Analysis: The Mannheim weight of an element $\alpha \in G_\pi$ is defined as $wt_\pi(\alpha) = |a| + |b|$ , where $a+bi$ is the representative minimizing this sum. The distance is $d_\pi(x, y) = wt_\pi(x-y)$ .
Coset Decomposition: The authors utilize the decomposition of the multiplicative group $\mathbb{F}_p^*$ into cosets of a subgroup $H = \{\pm 1, \pm \gamma\}$ (where $\gamma$ is a primitive 4th root of unity). This allows for the classification of elements by their Mannheim weights.
Sphere Packing & Enumeration: They derive an explicit formula for the volume of Mannheim balls ( $V_\pi(s, n)$ ) to establish sphere packing bounds.
MacWilliams Identity: A Mannheim version of the MacWilliams identity is used to analyze self-dual codes, relating the weight enumerator of a code to its dual.
Constraint Satisfaction: To find upper bounds for self-dual codes, the authors formulate a system of linear equations with non-negative integer constraints on the number of codewords with specific compositions (coset distributions).

3. Key Contributions

A. Volume of Mannheim Balls and Sphere Packing Bound

The paper derives an explicit formula for the number of vectors of a given Mannheim weight.

Result: A Mannheim-metric version of the Sphere Packing Bound is established: $|C| \cdot V_\pi(e, n) \le p^n$ , where $e = \lfloor (d_\pi - 1)/2 \rfloor$ .
Implication: This leads to necessary conditions for the existence of perfect codes (codes meeting the bound with equality).
2-Error Correcting Perfect Codes: The authors prove that if a 2-error-correcting perfect code exists over $G_\pi$ $G_{π}$ , the field size $p$ $p$ must satisfy $p^r = 8n^2 + 4n + 1$ $p^{r} = 8 n^{2} + 4 n + 1$ . They show that positive integer solutions only exist for $r=1$ $r = 1$ and $r=2$ $r = 2$ .
- The smallest candidate parameters are identified as $[n, k, d_\pi] = [10, 8, 5]$ over $\mathbb{F}_{29}$ .

B. General Bounds on Code Parameters

The authors establish several new bounds for linear codes over $G_\pi$ (specifically when $p \equiv 1 \pmod 4$ ):

Relation to Hamming Distance: They prove that $d_\pi(n, k) \le \frac{4S(a,b)}{p-1} d_H(n, k)$ , where $S(a,b)$ is a constant derived from the sum of Mannheim weights of coset leaders.
Tightness: They exhibit families of codes (including Simplex codes) that meet these bounds, proving they are tight for specific parameters.
Specific Values: Explicit formulas are derived for $d_\pi(n, 1)$ and $d_\pi(n, n-1)$ for specific primes (e.g., $\pi = 2+3i$ and $\pi = 1+4i$ ).

C. Self-Dual Codes

A significant portion of the paper is dedicated to self-dual codes over $\mathbb{F}_{13}$ and $\mathbb{F}_{17}$ (viewed as Gaussian integer rings).

Upper Bounds: Using the MacWilliams identity and linear programming techniques (solving for non-negative integer solutions of weight enumerator equations), they derive upper bounds on the minimum Mannheim distance $d_\pi^{SD}(n)$ .
Optimal Codes: They construct and classify optimal self-dual codes for lengths $n \le 14$ $n \leq 14$ .
- For $n=8$ over $\mathbb{F}_{13}$ , the maximum $d_\pi$ is 6.
- For $n=8$ over $\mathbb{F}_{17}$ , the maximum $d_\pi$ is 7.
MDS Connection: They prove that if certain high-distance self-dual codes exist (e.g., $[10, 5, 9]$ over $\mathbb{F}_{13}$ ), they must be Maximum Distance Separable (MDS) codes.

D. Decoding Algorithms

The paper presents a Syndrome Decoding Algorithm for codes over Gaussian integer residue rings.

Mechanism: It constructs a syndrome look-up table where coset leaders are chosen based on the minimum Mannheim weight rather than Hamming weight.
Key Insight: The authors provide examples demonstrating that certain error patterns are uncorrectable under the Hamming metric (due to high Hamming weight) but correctable under the Mannheim metric (due to low Mannheim weight). This highlights the practical advantage of the Mannheim metric for QAM channels.

4. Key Results

Perfect Code Existence: A 2-error-correcting perfect Mannheim code can only exist if $p^r = 8n^2 + 4n + 1$ . The smallest such field is $\mathbb{F}_{29}$ with parameters $[10, 8, 5]$ .
Bounds Tightness: The derived upper bounds for $d_\pi(n, k)$ are shown to be tight for specific parameters, particularly for Simplex codes and certain self-dual codes.
Self-Dual Classification: Complete classifications of self-dual codes up to length 8 over $\mathbb{F}_{13}$ and $\mathbb{F}_{17}$ are provided, identifying the optimal Mannheim distances.
Decoding Efficacy: Numerical examples confirm that the Mannheim metric allows for the correction of error vectors that would be considered uncorrectable (weight $> \lfloor (d_H-1)/2 \rfloor$ ) under the Hamming metric.

5. Significance

This paper significantly advances the theoretical foundation of coding over Gaussian integers.

Theoretical Rigor: It moves beyond ad-hoc constructions to provide rigorous bounds (Sphere Packing, Singleton-like, and MacWilliams-based) for the Mannheim metric.
Practical Relevance: By demonstrating that Mannheim-based decoding can correct errors that Hamming-based decoding cannot, the paper validates the use of Gaussian integer codes for modern 2D modulation schemes like QAM.
Optimization: The identification of optimal self-dual codes and the conditions for perfect codes provide a roadmap for designing efficient error-correcting codes for high-speed communication systems.
Algorithmic Implementation: The inclusion of decoding algorithms and explicit generator matrices for optimal codes bridges the gap between theoretical bounds and practical implementation.

In summary, the paper establishes that the Mannheim distance is not just a theoretical curiosity but a robust metric for 2D signal spaces, offering tighter bounds and superior error correction capabilities compared to traditional metrics in specific contexts.