Abstract
Finding all the mutually unbiased bases in various dimensions is a problem of fundamental interest in quantum information theory and pure mathematics. The general problem formulated in finite-dimensional Hilbert spaces is open. In the categorical approach to quantum mechanics, one can find examples of categories which behave "like" the category of finite-dimensional Hilbert spaces in various ways but are subtly different. One such category is the category of sets and relations, Rel. One can formulate the concept of mutually unbiased bases here as well. In this note, we classify all the mutually unbiased bases in this category by relating it to a standard question in combinatorics.
Introduction
In quantum mechanics, two bases are "mutually unbiased" if knowing a state in one basis provides no information about its state in the other. These Mutually Unbiased Bases (MUBs) are foundational for quantum cryptography and state tomography. In standard finite-dimensional Hilbert spaces (FDHilb), a $d$-dimensional space allows for at most $d+1$ MUBs. This limit is achievable when $d$ is a prime power. However, for other dimensions—most notably $d=6$—the existence of a full set of MUBs remains one of the most famous open problems in quantum theory.
This paper addresses the problem by shifting the investigation to a "toy universe": the category Rel (sets and binary relations). While Rel is not a physical model, it shares significant structural properties with quantum mechanics. By classifying MUBs in Rel, we can identify patterns and constraints that may offer insights into the mystery of MUBs in standard Hilbert spaces.
Classical Structures as Partitions
In the framework of Categorical Quantum Mechanics, "classical structures" correspond to bases and are defined by their ability to copy and delete information. In Rel, every classical structure is defined by a partition of a set and an abelian group structure on each part of that partition.
A key insight of this work is that the property of being "unbiased" depends solely on the underlying partition of the set, rather than the specific group operation chosen. Furthermore, for a basis in Rel to possess a complementary (unbiased) partner, its partition must be "uniform," meaning every group in the partition must be of equal size.
The Latin Square Equivalence
The core contribution of this paper is the establishment of a rigorous equivalence between MUBs in Rel and a classical combinatorial problem. The authors prove that finding $k$ mutually unbiased structures on a set of $d^2$ elements in Rel is mathematically equivalent to finding $k-2$ Mutually Orthogonal Latin Squares (MOLS) of size $d \times d$.
A Latin Square is a $d \times d$ array where every symbol appears exactly once in every row and column. Two such squares are orthogonal if, when superimposed, every possible pair of symbols appears exactly once. This mapping allows the authors to apply deep results from combinatorics to the quantum problem.
The Dimension 6 Mirror
This combinatorial connection provides a clear explanation for the difficulty of the $d=6$ case. In combinatorics, Tarry's theorem famously states that there are no pairs of orthogonal Latin squares of order 6. Consequently, in Rel, there are exactly 3 MUBs for a set of 36 elements ($6^2$). This provides a "toy" version of the $d=6$ problem in Hilbert space, where the maximum number of MUBs is also suspected (though not yet proven) to be 3.
Key Results and Conclusion
The paper provides a complete classification of MUBs in Rel based on the existence of MOLS. The primary findings include:
- Upper Bound: There are at most $d+1$ mutually complementary classical structures on $d^2$ elements.
- Prime Power Success: If $d$ is a prime power, the maximum of $d+1$ structures is always achievable.
- The d=6 Case: For a set with 36 elements, there are exactly 3 mutually unbiased structures.
While Rel is a simplified model, these results suggest that the constraints on MUBs in physical Hilbert spaces may be rooted in similar combinatorial structures. Future work may explore whether more complex combinatorial objects, such as Young tableaux, provide the necessary mapping for standard quantum mechanics.
Summary updated