[Paper Review] Some Open Problems in Combinatorial Physics

Gérard Duchamp, Hayat Cheballah|ArXiv.org|Jan 17, 2009

Advanced Combinatorial Mathematics9 references3 citations

TL;DR

This paper formulates four open problems in combinatorial physics centered on multiplicative structures in exponential generating functions, Riordan-Sheffer matrix groups, combinatorial vector fields, and probabilistic approximations of substitution matrices. It introduces combinatorial formulas for multiplicities in Feynman-Bender diagrams, seeks combinatorial proofs for matrix power operations, and analyzes the asymptotic rarity of substitution matrices in random unipotent matrices using probabilistic bounds.

ABSTRACT

We point out four problems which have arisen during the recent research in the domain of Combinatorial Physics.

Motivation & Objective

• To derive a closed-form formula for the multiplicity of Feynman-Bender diagrams in terms of unordered set partitions and their intersection matrices.
• To provide a combinatorial proof—without algebraic geometry—for the fact that the t-th power of a substitution matrix with prefunction remains a substitution matrix for rational t.
• To interpret the coefficients of the infinitesimal generator (vector field) arising from the logarithm of a substitution matrix in terms of graph enumeration and combinatorial structures.
• To study the frequency of substitution matrices within random unipotent matrices and derive probabilistic bounds on their appearance as matrix size and variable range increase.

Proposed method

• Uses Hadamard exponential product of exponential generating functions to relate combinatorial structures to labeled partitions and their intersection matrices.
• Applies Bell polynomials and unordered partition types to express the Hadamard product as a sum over bicolored Feynman-Bender diagrams with multiplicities.
• Employs matrix logarithms and limits to define the infinitesimal generator of a one-parameter group of substitutions, yielding a differential operator.
• Employs probabilistic sampling of unipotent matrices with bounded integer entries to estimate the frequency of substitution matrices.
• Derives an upper bound on the probability of observing a substitution matrix as a function of matrix size n and variable range r.
• Uses Zariski-like arguments and combinatorial enumeration to analyze the closure properties of substitution matrices under matrix powers and logarithms.

Experimental results

Research questions

• RQ1 What is a closed-form combinatorial expression for the multiplicity of a given Feynman-Bender diagram in the Hadamard product of two exponential generating functions?
• RQ2 Can the closure of the Riordan-Sheffer group under rational powers be proven combinatorially, without relying on pro-algebraic structures?
• RQ3 What is the combinatorial interpretation of the vector field obtained as the logarithmic derivative of a substitution matrix in the context of labeled graphs?
• RQ4 How does the probability of a randomly generated unipotent matrix being approximately a substitution matrix behave as matrix size increases?
• RQ5 Does the dependence of this probability on the variable range vanish asymptotically as matrix size tends to infinity?

Key findings

• The multiplicity of a diagram d in the Hadamard product is given by the number of unordered pairings of set partitions whose intersection matrix lies in the class of d.
• An upper bound on the probability of observing a substitution matrix in random unipotent matrices is derived as $ p_n \leq \frac{r^{2n-3}}{r^{\frac{n(n-1)}{2}}} $, which tends to zero as $ n \to \infty $.
• For $ 3 \times 3 $ unipotent matrices, all such matrices are exact substitution matrices due to the constrained form of their third column generating series.
• As the range of integer entries increases while keeping matrix size and number of draws fixed, the observed probability of substitution matrices rapidly decreases toward zero.
• The logarithmic derivative of a substitution matrix yields a differential operator of the form $ q(z)\frac{d}{dz} + v(z) $, which corresponds to a combinatorial vector field with rational coefficients.
• The probability of finding a substitution matrix in random unipotent matrices is higher for smaller matrices and lower for larger ones, with the bound indicating exponential decay in size.