Finding Generating Functions from Recurrence Relations | Discrete Math Tutorial

The symbolic method for labeled combinatorial classes provides a systematic way to derive exponential generating functions (EGFs) by translating combinatorial constructions into algebraic operations: disjoint union corresponds to addition of EGFs, labeled product corresponds to multiplication of EGFs, sequences correspond to 1/(1-A(z)), sets correspond to e^(A(z)), and cycles correspond to ln(1/(1-A(z))). This framework allows us to analyze complex combinatorial structures like permutations, derangements, and involutions by expressing them as combinations of basic operations and extracting coefficients from the resulting generating function equations.

Exponential generating functions (EGFs) are used for counting labeled combinatorial objects. An EGF has the form ∑ a_i x^i / i!. When combining labeled objects into structures, the EGF of the combined structure is the product of the EGFs of the components. For example, to count pairs of labeled objects, multiply their EGFs. This differs from ordinary generating functions, which are used for unlabeled objects.

An exponential generating function (EGF) encodes a sequence a₀, a₁, a₂, ... as a power series where the coefficient of xⁿ is aₙ/n!; unlike ordinary generating functions, EGFs often provide nicer convergence properties and are particularly useful for combinatorial problems involving labeled structures. For example, the EGF for the constant sequence 1 is eˣ, and the EGF for permutations of n elements is 1/(1-x). To count derangements (permutations with no fixed points), we use the combinatorial identity that any permutation can be decomposed into a fixed set and a deranged set, leading to the EGF equation eˣ × D̃(x) = 1/(1-x), which solves to D̃(x) = eˣ/(1-x), providing an elegant method to derive the number of derangements of a finite set.

Exponential generating functions (EGFs) are used for counting labeled objects. An EGF has the form ∑(aₙxⁿ/n!). When combining labeled objects, the product of EGFs corresponds to forming ordered pairs of objects. For example, if T(x) describes objects of type T and U(x) describes objects of type U, then T(x)U(x) describes ordered pairs (T, U) where the first object has size k and the second has size n-k, with the factor n! accounting for the labeling.

This comprehensive section establishes the theoretical framework for analyzing labeled combinatorial structures using exponential generating functions. It begins by defining exponential generating functions as sum_{n≥0} a_n * z^n / n! for labeled structures on n elements. The section introduces three fundamental combinatorial operations: the product operation (A × B) which combines structures by partitioning elements and placing structures on subsets, corresponding to multiplying generating functions; the sequence operation which creates ordered collections with generating function 1/(1-B(z)); and the set operation which creates unordered collections with generating function e^{B(z)}. Throughout, concrete examples illustrate these concepts: permutations (a_n = n! with generating function 1/(1-z)), cycles (a_n = (n-1)! with generating function -ln(1-z)), sets (a_n = 1 with generating function e^z), and the atomic class (a_1 = 1 with generating function z). The section demonstrates how these operations allow complex structures to be built from simpler ones, with the algebraic operations providing direct computational tools for enumeration.

The Catalan recurrence is C(n+1) = Σ_{i=0}^n C(i)C(n-i) for n ≥ 0, with C(0) = 1. This nonlinear recurrence counts parenthesizations of n+1 symbols. Applying generating functions: multiply both sides by x^{n+1} and sum over n ≥ 0, yielding F(x) = x(F(x))² + 1. Solving the quadratic equation gives F(x) = [1 - √(1-4x)]/(2x). The coefficient of x^n in this expansion is the nth Catalan number, demonstrating how generating functions solve nonlinear recurrences.

For non-linear recurrences like Catalan numbers, generating functions yield quadratic equations. Solving F(x)^2 - F(x) + 1 = 0 gives F(x) = (1 - √(1-4x))/2. The square root extraction requires choosing the negative root to satisfy initial conditions. This demonstrates how generating functions transform combinatorial recurrences into algebraic problems solvable through standard techniques.

Catalan numbers Cₙ satisfy the non-linear recurrence Cₙ = Σₖ₌₀ⁿ⁻¹ CₖCₙ₋₁₋ₖ for n ≥ 1, with C₀ = 1. This recurrence appears in triangulations of polygons, correct bracket sequences, and binary trees. The generating function C(x) = Σₙ₌₀^∞ Cₙxⁿ satisfies the quadratic equation x·C(x)² - C(x) + 1 = 0. Solving gives C(x) = [1 - √(1 - 4x)]/(2x), with the minus sign chosen for finiteness at x = 0. The square root of 1 + x is √(1 + x) = Σₙ₌₀^∞ (1/2 choose n)xⁿ, where (1/2 choose n) uses generalized binomial coefficients with non-integer upper indices.

This section introduces generating functions to solve the Catalan recurrence. Let f(x) = sum C_n x^n. The recurrence translates to f(x) = 1 + x f(x)^2, which is a quadratic equation in f(x). Solving with the quadratic formula gives f(x) = [1 ± √(1 - 4x)] / (2x). The correct sign is determined by requiring f(0) = C_0 = 1, which forces the minus sign.

The generating function C(x) = Σₙ₌₀^∞ Cₙxⁿ satisfies the functional equation C(x) = 1 + x[C(x)]². This equation is derived by starting with the generating function definition, applying the recurrence relation, recognizing the Cauchy product structure, and substituting the generating function itself into the equation.

Partition identities can be proven using generating functions. The generating function for partitions into distinct parts congruent to 1, 2, or 4 mod 7 is the infinite product ∏(1 + q^k) where k ≡ 1,2,4 mod 7. The generating function for unrestricted partitions into parts congruent to 1, 9, or 11 mod 14 is ∏(1/(1 - q^k)) where k ≡ 1,9,11 mod 14. By manipulating these products and showing they are equal, we prove the identity that these two sets of partitions have the same count for each n.

This section presents a rigorous mathematical proof using mathematical induction. The base cases m=1 and m=2 are verified directly. The inductive hypothesis assumes πₖ(q) = ∏ⱼ=1^k 1/(1 - q^j). For the inductive step, partitions with parts ≤ k+1 split into those using 0, 1, 2, ... copies of k+1, with remaining parts ≤ k. This yields πₖ₊₁(q) = πₖ(q) × 1/(1 - q^(k+1)), completing the induction and proving the general formula.

Combinatorics extends to advanced applications including graph theory (counting vertex-edge relationships in networks), partition theory (dividing sets into non-overlapping subsets for probability analysis), and generating functions (encoding recurrence relations for algorithm analysis). Graph theory applies combinatorial counting to connectivity problems. Partition theory enables the base theorem in probability by dividing sample spaces into disjoint events. Generating functions transform recurrence relations into solvable forms, bridging combinatorics with algorithm complexity analysis.

Generating functions can prove combinatorial identities by showing that two seemingly different counting problems yield the same generating series; specifically, the number of partitions of a positive integer into odd parts equals the number of partitions into distinct parts, as demonstrated by manipulating the generating function for odd-partitions to reveal it equals the generating function for distinct-partitions.

Euler established that the generating function for partitions is the infinite product ∏_{k=1}^∞ 1/(1-x^k). He also proved that the number of partitions into odd parts equals the number of partitions into distinct parts—a striking combinatorial identity. The pentagonal number theorem provides a recurrence: p(n) = ∑(-1)^{k+1}p(n-g_k) where g_k = k(3k-1)/2 are generalized pentagonal numbers. These results demonstrate deep connections between combinatorial structures and analytic functions.