Introduction
In classical algebra and cryptographic constructions, binary operations such as XOR play a critical role due to their algebraic simplicity and reversibility. XOR is a prototypical example of a commutative involution—a binary function \(f(a, b)\) satisfying:
\[f(a, b) = f(b, a), \quad \text{and} \quad f(f(a, b), b) = a\]
This paper aims to generalize this notion by studying commutative, self-reversing binary functions over finite \(n\)-state systems. Our focus is on functions that are not expressible as field additions over \(\text{GF}(2^k)\), yet still satisfy involutory and commutative properties.
Preliminaries and Definitions
Let \(S = \{0, 1, \dots, n-1\}\) where \(n = 2^k\). Define a function \(f: S \times S \rightarrow S\) to be a commutative involution if it satisfies:
- Commutativity: \(f(a, b) = f(b, a)\)
- Self-reversibility: \(f(f(a, b), b) = a\)
These properties imply that knowledge of any two among \(a, b, c = f(a, b)\) uniquely determines the third.
Inverter Permutations
Define a reversible \(n\)-state inverter inv as a permutation of \(S\):
inv: S \rightarrow S \quad \text{bijective}\]
A permutation inv is self-reversing if:
inv(inv(x))=x
Example: the permutation inv4=[2 3 0 1] is self-reversing over \(S = \{0, 1, 2, 3\}\).
The Involution Conjecture
- Part 1: Involutions only exist for \(n = 2^k\).
- Part 2: For \(n > 2\), there exist \(p > n\) distinct self-reversing inverters.
- Part 3: Some of these operations are not isomorphic to field addition in \(\text{GF}(2^k)\).
We conjecture a vast class of such operations beyond XOR, constructed from rows/columns of self-reversing inverters and satisfying commutativity and reversibility.
Constructive Examples
The following table represents the classic XOR operation over \(\text{GF}(4)\):
\[\begin{bmatrix} 0 & 1 & 2 & 3 \\ 1 & 0 & 3 & 2 \\ 2 & 3 & 0 & 1 \\ 3 & 2 & 1 & 0 \\ \end{bmatrix}\]
We now construct a novel involution that is not equivalent to XOR:
\[\begin{bmatrix} 0 & 1 & 3 & 2 \\ 1 & 0 & 2 & 3 \\ 3 & 2 & 1 & 0 \\ 2 & 3 & 0 & 1 \\ \end{bmatrix}\]
All rows and columns in this table are self-reversing inverters. The operation is commutative and involutory but cannot be derived from addition in \(\text{GF}(4)\).
Enumeration and Computational Results
Using MATLAB, we enumerated 10 valid self-reversing inverters for \(n=4\). For \(n=8\), the count is significantly larger. By assembling operation tables and enforcing symmetry, we identified hundreds of involutions not reducible to field arithmetic.
Extensions and Cryptographic Relevance
While the primary thrust of this paper is mathematical, these structures naturally invite cryptographic interest:
Involutions can replace XOR in cipher structures such as AES’s AddRoundKey, offering potential increases in algebraic complexity.
A transformation method has been developed that lifts an \(n\)-state involution to a \(2n\)-state one, expanding the design space of secure permutations.
The Finite Lab-Transform (FLT) [1] introduced in previous work can yield \((n-3)!\) variations of a given involution, enhancing algorithmic entropy [2].
These cryptographic applications are secondary to the core mathematical interest of characterizing the involution space.
Usage and Patent Notice
The Finite Lab-Transform (FLT) [1] and other concepts[2] discussed in this article are protected under issued USA patents. While this material may be freely used for educational, review, research, and testing purposes, any operational use—including commercial applications—requires compliance with the relevant patent protections. No license for operational use is provided by this publication. For further inquiries regarding licensing and implementation, please contact the author.
References
-
[1] Peter Lablans. Website: Finite Lab-Transform (FLT). https://labtransform.com, 2022.
[2] Peter Lablans. Website: Insanely Better Encryption by CFT, https://www.labcyfer.com/articleCFT/articleCFT.htm, 2025.