Fact-checked by Grok 2 weeks ago

Lattice path

A lattice path is a path on the \mathbb{Z}^d, typically d=2, consisting of a of consecutive points where each step connects adjacent lattice points via unit vectors in a predefined step set, such as east (1,0), north (0,1), or more general small steps in \{-1,0,1\}^2 \setminus \{(0,0)\}. In , lattice paths serve as a core model for counting problems, with the unrestricted paths from the origin (0,0) to (a,b) using only east and north steps enumerated by the \binom{a+b}{a}, which counts the ways to arrange a east steps and b north steps in . This basic extends to higher dimensions via multinomial coefficients and forms the foundation for more complex variants, such as paths confined to cones or strips. Notable restricted lattice paths include Dyck paths, which are paths from (0,0) to (2n,0) with up (1,1) and down (1,-1) steps that never go below the x-axis, counted by the Catalan number C_n = \frac{1}{n+1} \binom{2n}{n}, and they bijectively correspond to balanced parentheses, binary trees, and non-crossing partitions. The ballot theorem, which quantifies paths staying above the diagonal using the reflection principle, further refines these counts, with applications to election outcomes and queueing theory. Lattice paths have broad applications beyond pure , modeling random walks in probability and statistics, vicious walkers in statistical physics, configurations in chemistry, and solutions in . Advanced techniques, including generating functions, the , and the Lindström-Gessel-Viennot for non-intersecting paths, enable and connections to orthogonal polynomials, symmetric functions, and plane partitions.

Basic Concepts

Definition and Terminology

A lattice path is a sequence of points in the integer lattice \mathbb{Z}^d, typically \mathbb{Z}^2, where each consecutive pair of points differs by a fixed set of allowed steps, connecting lattice points with integer coordinates. These paths often start at the origin (0,0) and end at a specified point (m,n), forming a walk along the grid lines of the lattice. Standard steps in lattice paths include the east step (1,0) and the north step (0,1), though other directions such as south (0,-1) or west (-1,0) may be permitted depending on the context. Particular emphasis is placed on lattice paths, which use only non-decreasing steps like east and north, ensuring the path never moves left or down. Key terminology includes lattice points, the integer-coordinate vertices visited by the path; edges or steps, the line segments connecting consecutive points; and the length of the path, defined as the number of steps taken. Paths are commonly notated as sequences of step abbreviations, such as R for right (east) and U for up (north), e.g., the sequence RURRU represents a specific path from (0,0) to (3,2). The starting and ending points are specified to define the path's scope, with the endpoint determining the total displacement in each coordinate. The study of lattice paths originated in 18th-century combinatorics, with foundational work by Leonhard Euler on related enumeration problems in grids, predating explicit developments in the 19th century.

Grid and Step Types

Lattice paths are primarily constructed on the \mathbb{Z}^2, where points have integer coordinates, allowing paths to traverse the entire plane unless otherwise constrained. This grid structure provides a discrete framework for modeling sequences of moves between adjacent points. In many combinatorial applications, paths are restricted to the first quadrant \mathbb{N}_0^2 (non-negative integers), ensuring that coordinates remain non-negative and preventing excursions into negative regions. Extensions to the full \mathbb{Z}^2 plane accommodate paths that may cross axes, while higher-dimensional lattices like \mathbb{Z}^d generalize the concept for multidimensional paths. Toroidal grids, formed by identifying opposite boundaries of a finite rectangular lattice to create periodic conditions, appear in certain graph-theoretic variants of lattice paths. The step sets defining lattice paths consist of vectors corresponding to moves between adjacent lattice points, typically unit length for simplicity. Common steps include horizontal moves (1,0) (east) or (-1,0) (west), and vertical moves (0,1) (north) or (0,-1) (south), forming a sequence of connected line segments. For monotone paths, the step set is restricted to non-decreasing directions, such as only east (1,0) and north (0,1), ensuring no backtracking or descent in coordinates. In contrast, paths allowing down or left steps, such as including (0,-1) or (-1,0), enable closed or returning configurations, though these may violate monotonicity. More general directed steps, like diagonals (1,1) or (1,-1), expand the set while maintaining a positive horizontal drift. Path constraints emphasize monotonicity in many cases, where each step increases at least one coordinate without reversal, defining the path length as the total number of steps taken. For instance, a from (0,0) to (2,2) requires exactly two east steps and two north steps in some order, forming a of four moves. Variations distinguish directed paths, which enforce steps in predefined orientations (e.g., only positive directions), from undirected paths that permit symmetric moves without directional bias, though the latter are less common in . Finite grids impose limits on path extent, while infinite grids like \mathbb{Z}^2 allow unbounded exploration.

Monotone Paths

North-East Paths

North-East lattice paths represent the simplest form of monotone lattice paths, consisting exclusively of horizontal and vertical movements in positive directions. A North-East path from the origin (0,0) to a point (m,n) in the is defined as a sequence of steps, where each step is either an East move of (1,0) or a North move of (0,1), resulting in exactly m East steps and n North steps to reach the endpoint. These paths inherently remain within the non-negative , as no step decreases the x- or y-coordinate, ensuring all intermediate points have non-negative coordinates. The total length of such a path is m + n steps, with the order of East and North steps determining the specific route taken. For example, a path to (2,2) might follow the sequence East-East-North-North (EENN), which proceeds along the x-axis to (2,0) before ascending to (2,2); alternatively, East-North-East-North (ENEN) alternates directions, passing through points (1,0), (1,1), (2,1), and (2,2); or North-North-East-East (NNEE), which first climbs to (0,2) along the y-axis and then moves right to (2,2). These examples illustrate how different interleavings of the steps create varied trajectories while adhering to the monotone constraint. In combinatorial terms, North-East paths correspond directly to sequences (or words) in the {E, N} of length m + n, containing precisely m E's and n N's, where the position of each step dictates the path's shape on the grid. This sequential representation underscores their role as foundational objects in , facilitating analysis of their structural properties without imposing additional boundaries.

Enumeration via Binomial Coefficients

The number of unrestricted north-east lattice paths from (0,0) to (m,n), consisting of m east steps (1,0) and n north steps (0,1), is given by the \binom{m+n}{m}, which equals \frac{(m+n)!}{m! \, n!}. This counts the distinct sequences of steps required to reach the endpoint, where the total number of steps is m + n. To derive this formula, consider that any such path is a sequence of exactly m east steps and n north steps in some order. The total number of distinct sequences is the number of ways to choose m positions out of the m + n total steps for the east steps (with the remaining n positions assigned to north steps), which directly yields \binom{m+n}{m}. This combinatorial argument provides a straightforward mechanism without reference to generating functions or recurrences. A proof of this enumeration can be established via a bijection to the set of subsets of size m from a set of m + n elements. Specifically, map each path to the subset of positions where east steps occur; this correspondence is one-to-one and onto, confirming the count \binom{m+n}{m}. An inductive proof also holds: the base case for m = 0 or n = 0 is trivial (1 path each), and for larger values, paths to (m,n) either end with an east step (preceded by paths to (m-1,n)) or a north step (preceded by paths to (m,n-1)), giving the recurrence \binom{m+n}{m} = \binom{m+n-1}{m-1} + \binom{m+n-1}{m}, which aligns with the Pascal identity for binomial coefficients. Combinatorially, these paths interpret as sequences of length m + n with exactly m zeros (east steps) and n ones (north steps), providing a direct equivalence to the definition of coefficients. In a context, they model sequences where m voters favor one candidate and n favor another, with the path tracing the cumulative vote tally without restrictions on lead changes. This enumeration generalizes to paths with multiple step types via multinomial coefficients. For instance, the number of paths from (0,0,0) to (m,n,k) using east (1,0,0), north (0,1,0), and up (0,0,1) steps is \binom{m+n+k}{m,n,k} = \frac{(m+n+k)!}{m! \, n! \, k!}, obtained by choosing positions for each step type in the total sequence of m + n + k steps.

Bounded and Non-Intersecting Paths

Dyck Paths

A Dyck path of semilength n is a lattice path from (0,0) to (n,n) on the integer grid, consisting of n east steps (1,0) and n north steps (0,1), that never rises above the diagonal line y = x. This condition is equivalent to requiring that, in the sequence of steps, the number of east steps is at least the number of north steps in every prefix. As a result, the path stays weakly below the diagonal, touching it at the start and end points, and possibly at intermediate lattice points along the way. This two-dimensional formulation of Dyck paths on the is closely related to a one-dimensional , where the path is viewed as a walk from (0,0) to (2n,0) using up steps (1,1) and down steps (1,-1), never descending below the x-axis. In this equivalent model, the up steps correspond to east steps in the version, and down steps to north steps, preserving the non-negativity condition that mirrors the prefix balance in the original lattice . This reformulation grounds the combinatorial structure in a simpler on the line while maintaining the bounded nature inherent to the two-dimensional paths. For illustration, consider n=1: the unique Dyck path is the sequence east then north, tracing (0,0) \to (1,0) \to (1,1), which satisfies y \leq x at every point. For n=2, the two Dyck paths are:
  • East, east, north, north: (0,0) \to (1,0) \to (2,0) \to (2,1) \to (2,2)
  • East, north, east, north: (0,0) \to (1,0) \to (1,1) \to (2,1) \to (2,2)
Both remain below or on the diagonal y=x. Dyck paths admit natural bijections to other combinatorial objects, such as sequences of n balanced parentheses, where an east step maps to an opening parenthesis and a north step to a closing one, ensuring the prefix condition matches the balance requirement. They also biject to full trees with n+1 leaves, constructed recursively by associating subpaths to left and right subtrees. These correspondences highlight the role of Dyck paths as a foundational structure in , with further applications detailed elsewhere. The number of Dyck paths of semilength n from (0,0) to (2n,0) is given by the nth C_n = \frac{1}{n+1} \binom{2n}{n}. This explicit formula arises from the applied to lattice paths. The total number of unrestricted paths with n up steps (1,1) and n down steps (1,-1) is \binom{2n}{n}. To count only those staying non-negative, subtract the number of paths that touch or cross below the x-axis, which by reflecting the initial segment until the first touch of y=-1 equals \binom{2n}{n-1}. Thus, C_n = \binom{2n}{n} - \binom{2n}{n-1} = \frac{1}{n+1} \binom{2n}{n}. Catalan numbers also satisfy a recursive relation derived from the decomposition of Dyck paths. Specifically, C_0 = 1 and for n \geq 1, C_n = \sum_{i=0}^{n-1} C_i C_{n-1-i}, corresponding to the first return to the x-axis after $2(i+1) steps, followed by a Dyck path of semilength i and one of semilength n-1-i. The ordinary for the is C(x) = \sum_{n=0}^{\infty} C_n x^n = \frac{1 - \sqrt{1-4x}}{2x}, which satisfies the C(x) = 1 + x C(x)^2 from the recursive structure. Fuss-Catalan numbers provide a generalization for lattice paths with higher up steps. The m-Fuss-Catalan number is C_n^{(m)} = \frac{1}{m n + 1} \binom{(m+1) n}{n}, counting m-Dyck paths from (0,0) to ((m+1)n, 0) using n up steps of (1, m) and m n down steps of (1, -1), staying non-negative.

Non-Intersecting Paths

Non-intersecting lattice paths refer to a collection of two or more lattice paths from specified starting points to specified ending points that do not cross or touch each other except possibly at endpoints. These configurations are crucial in modeling non-crossing structures and are enumerated using the Lindström–Gessel–Viennot lemma, which states that the number of such k-tuples of non-intersecting paths is given by the determinant of a k×k matrix whose entries count the number of unrestricted paths between corresponding starts and ends. This lemma has applications in counting tilings, permutations, and plane partitions, with further details in advanced enumeration techniques.

Advanced Enumeration Techniques

Ballot Theorem

The ballot theorem is a fundamental result in combinatorics that enumerates the number of lattice paths satisfying a strict inequality condition relative to a boundary line. In its classical probabilistic formulation, suppose candidate A receives a votes and candidate B receives b votes, where a > b \geq 0 are integers, and all (a + b)!\ / (a! \, b!) possible vote sequences are equally likely. The probability that A maintains a strict lead over B throughout the entire counting process—meaning the number of votes for A exceeds those for B in every prefix of the sequence—is \frac{a - b}{a + b}. Combinatorially, this corresponds to lattice paths on the integer grid from (0,0) to (b, a) consisting of b east steps (1,0) (representing votes for B) and a north steps (0,1) (votes for A), such that the path remains strictly above the diagonal line y = x at every point after the , i.e., the y-coordinate exceeds the x-coordinate along the entire path. The number of such paths is \frac{a - b}{a + b} \binom{a + b}{a}. Equivalently, in terms of paths from (0, b) to (a, 0) with a east steps (1, 0) and b south steps (0, -1), the theorem counts those that do not touch the line y = 0 except at the starting point, yielding the same formula \frac{a - b}{a + b} \binom{a + b}{a}. The proof relies on a refined application of the , which partitions the bad paths (those violating the condition) into two classes B_0 and B_1. Each class is placed in with the unrestricted paths from a reflected via reflection or rotation of path segments, with |B_0| = |B_1| = \binom{a + b - 1}{a}. Thus, the total number of bad paths is $2 \binom{a + b - 1}{a}. The number of good paths is then \binom{a + b}{a} - 2 \binom{a + b - 1}{a} = \frac{a - b}{a + b} \binom{a + b}{a}, where the algebraic simplification follows from the identity \binom{a + b - 1}{a} = \frac{b}{a + b} \binom{a + b}{a}. The theorem was first stated by in 1887, who provided an inductive outline but sought a direct proof; the proof was given shortly thereafter by Désiré André in the same year. Its roots trace to earlier work on recurrent sequences and probabilistic applications by in 1777. Generalizations of the ballot theorem extend the boundary condition and the number of candidates. For the weak inequality version, where A stays ahead or tied (y ≥ x throughout), the number of paths from (0,0) to (b, a) with a ≥ b is \frac{a - b + 1}{a + 1} \binom{a + b}{b}. A broader form, due to Takács, handles cases where A maintains a lead by at least k times B's votes (u > k v), giving \frac{u - k v}{u + v} \binom{u + v}{u} paths, which includes multiple boundary constraints for scenarios like several candidates. As a special case, setting a = n + 1 and b = n yields the nth , connecting to the enumeration of Dyck paths.

Cycle Lemma and Reflections

The cycle lemma, introduced by Dvoretzky and Motzkin, asserts that given a sequence of a up steps (+1) and b down steps (-1) with a = kb + r for integers k \geq 1 and $0 \leq r < k, there are exactly r cyclic rotations of the sequence in which the number of up steps is at least k times the number of down steps in every prefix. The proof proceeds by analyzing the circular sequence and identifying starting positions where the dominance condition holds throughout, using the total excess r to count the valid origins via a bijective correspondence with certain subpaths. In the context of Dyck paths consisting of n up steps and n down steps, the lemma implies that exactly the nth Catalan number C_n such paths, when viewed cyclically, have a rotation starting at their minimal height that returns to it without descending below. This follows from shifting each balanced sequence to align with its unique minimal partial sum position, yielding a non-negative path, with the multiplicity accounted for by the returns in composite paths. Applying the lemma to unbalanced sequences with n+1 up steps and n down steps (where r=1) directly yields C_n = \frac{1}{n+1} \binom{2n}{n} good rotations, each corresponding to a Dyck path after adjustment. The Dvoretzky-Motzkin refinement generalizes the lemma to paths with arbitrary step sets, such as combinations of up, down, and level steps (as in ), where the dominance condition is adapted to weighted increments. For instance, in paths with steps of +1, -1, and 0, the number of cyclic rotations staying non-negative is determined by the net displacement parameter, enabling enumeration of generalized Catalan structures like or . For primitive Dyck paths of semilength n—those touching height 0 only at the endpoints—the cycle lemma yields the count \frac{1}{n} \binom{2n-2}{n-1}, equivalent to the (n-1)th C_{n-1}. This arises by applying the lemma to sequences of n up steps and n-1 down steps, where the unique good rotation (with r=1) stays strictly positive after the initial up step until the end, and appending a final down step produces the primitive excursion. Advanced reflection methods extend the basic reflection principle, commonly linked to of the ballot problem despite his use of a cycle-based bijection rather than geometric reflection. In for the , bad paths crossing the diagonal are paired with paths starting from the reflected origin by inverting the initial segment until the first crossing, yielding the formula for paths from (0,b) to (n,a) with a > b \geq 0 staying above the line y = x. For non-intersecting lattice paths, multiple reflections are employed via the generalized reflection principle of Gessel and Zeilberger, which sums signed path counts over the reflections in the associated (e.g., the for type A). Specifically, the number of k-tuples of non-intersecting paths from starts A_1, \dots, A_k to ends E_1, \dots, E_k staying in a Weyl chamber is \sum_{\sigma \in S_k} \operatorname{sgn}(\sigma) \prod_{i=1}^k N(A_i \to E_{\sigma(i)}), where N denotes unrestricted path counts; this aligns with the Lindström-Gessel-Viennot determinant for non-intersecting configurations.

Generalizations and Variants

Higher-Dimensional Paths

Lattice paths in higher dimensions extend the two-dimensional case to the \mathbb{Z}^d for d \geq 3. A higher-dimensional lattice path is defined as a sequence of points starting from the origin (0, \dots, 0) and ending at a point (m_1, \dots, m_d) with non-negative integer coordinates, where each step is a e_i along one of the d coordinate axes, increasing the i-th coordinate by 1. This generalizes the north-east paths in 2D, where steps are restricted to positive directions along the axes. The enumeration of unrestricted higher-dimensional lattice paths relies on multinomial coefficients. The number of such paths from (0, \dots, 0) to (m_1, \dots, m_d) is given by the multinomial coefficient \binom{m_1 + \dots + m_d}{m_1, \dots, m_d} = \frac{(m_1 + \dots + m_d)!}{m_1! \cdots m_d!}, which counts the ways to order the sequence of m_1 + \dots + m_d steps, with m_i steps in the i-th direction for each i. For example, the number of paths from (0,0,0) to (1,1,1) in is \binom{3}{1,1,1} = 6, corresponding to the permutations of one step in each of the three directions. Properties of higher-dimensional paths include generalizations of structures like Delannoy paths, which in allow steps along axes or diagonals (1,1). In n dimensions, generalized Delannoy paths are minimal paths from one point to another in \mathbb{Z}^n permitting diagonal steps that increase multiple coordinates simultaneously, forming a poset isomorphic to ordered partitions of a . Enumerating these involves Thue systems and congruence classes, extending the 2D Delannoy numbers. However, challenges arise in restricted settings, such as non-intersecting paths in 3D, where exact counts require advanced techniques like analytic in several variables due to increased geometric complexity.

Weighted and Decorated Paths

In lattice path enumeration, weighted paths extend the basic counting by assigning a weight w(s) to each step s in the path, where the total weight of a path P = (s_1, s_2, \dots, s_k) is the product \prod_{i=1}^k w(s_i). The generating function for the set of all such paths from a starting point A to an ending point E is then the sum over all paths P from A to E of the total weight of P. This framework allows for algebraic refinements that capture additional statistics, such as area or height, by choosing appropriate weight functions; for instance, assigning weight q to steps based on the area below the path yields q-analogues of classical counts. Decorated paths introduce labels or colors on steps or points, transforming the into multivariate generating functions or q-analogues that track these decorations. For example, coloring steps with multiple types leads to weighted counts where each color contributes a distinct , enabling the study of refined structures like those with specified ascent or turn patterns. Such decorations often arise in connections to orthogonal polynomials or symmetric functions, where the encodes the distribution of labels along the . A prominent example is the weighted enumeration of Dyck paths, where weights on peaks or valleys produce rational generating functions that generalize the Catalan generating function C(x) = \sum_{n=0}^\infty C_n x^n = \frac{1 - \sqrt{1-4x}}{2x}. Specifically, for Dyck paths of semilength n with weight q raised to the area between the path and the diagonal, the q-Catalan number is given by C_n(q) = \sum_{\pi} q^{\mathrm{area}(\pi)}, where the sum is over all Dyck paths \pi from (0,0) to (n,n) staying weakly above the diagonal, and \mathrm{area}(\pi) counts the number of unit squares below \pi but above the diagonal. This q-analogue satisfies a rational derived from the path decomposition into irreducible components. For non-intersecting paths, the Lindström-Gessel-Viennot lemma extends to the weighted setting, providing a formula for the signed sum of weights of tuples of non-intersecting paths from starting points A_1, \dots, A_k to ending points B_1, \dots, B_k. The lemma states that this equals \det(M), where M_{i,j} is the for weighted paths from A_i to B_j, assuming the paths are in a . This weighted version, building on earlier unweighted formulations, is crucial for enumerating systems like plane partitions or tilings via path interpretations.

Combinatorial Applications

Connections to Permutations and Trees

Lattice paths, particularly Dyck paths, exhibit deep bijective correspondences with through the first return decomposition. A Dyck path of semilength n from (0,0) to (2n,0) can be uniquely decomposed by identifying the first return to the x-axis at some point (2k,0), yielding an initial up step, a sub-Dyck path of semilength k-1 (corresponding to the left subtree), a down step, and a remaining Dyck path of semilength n-k (the right subtree). This recursive structure mirrors the definition of a full with n internal nodes, establishing a size-preserving between the two sets, both enumerated by the C_n. This connection, central to combinatorial enumeration, was formalized in early works on recursive structures. Similarly, lattice paths connect to permutations via pattern avoidance. The set of 321-avoiding permutations of $$ is equinumerous with Dyck paths of semilength n, as both are counted by C_n. A explicit bijection maps a 321-avoiding permutation to a Dyck path by recording the heights of left-to-right maxima or using the insertion tableau from the , where the shape of the tableau encodes the path via its row lengths. For instance, the permutation's decreasing subsequences are avoided, ensuring the path stays above the diagonal, analogous to the non-negativity condition in Dyck paths. This bijection, preserving statistics like fixed points and excedances, was first described by Knuth and has been refined in subsequent analyses of permutation classes. Lattice paths also relate to Young tableaux through boundary representations and the Robinson-Schensted-Knuth (RSK) correspondence. The outer boundary of a Young diagram of shape \lambda \vdash n can be interpreted as a lattice path in the plane from (0,0) to (\lambda_1, \lambda_1'), where \lambda_1' is the conjugate length, staying above the diagonal. The RSK bijection maps a permutation (or more generally, a biword or matrix) to a pair of standard Young tableaux (P, Q) of the same shape \lambda, and the growth of \lambda along the insertion process corresponds to non-intersecting lattice paths via the Lindström-Gessel-Viennot lemma, linking the number of such tableaux to determinants of path counts. This framework highlights how lattice paths delineate the possible shapes and fillings of tableaux, with seminal developments in the RSK algorithm providing the core mapping. Lattice paths also relate to non-crossing of $$ via a , both enumerated by the C_n, where the blocks of the partition correspond to the connected components defined by the path's ascents and descents in a standard mapping. For example, traversing the Dyck path identifies "valleys" that delimit non-crossing blocks, preserving the enumeration. This correspondence, rooted in early analytic , facilitates refined enumerations preserving statistics like area or .

Interpretations in Tilings and Partitions

Lattice paths offer a geometric interpretation for enumerating tilings of polyomino regions, particularly through bijections to families of non-intersecting paths on the . A prominent example is the domino tilings of the of order n, a region consisting of $2n(n+1) unit squares arranged in a diamond shape centered at the with |x| + |y| \leq n. Each such tiling corresponds bijectively to a set of n non-intersecting lattice paths from the points (0, -n + 2i - 1) for i = 1, \dots, n to the points (2n, 2i - 1) for i = 1, \dots, n, using steps (1,1), (1,-1), and (2,0). The enumeration of these paths, via the , yields exactly $2^{n(n+1)/2} tilings. These configurations also model the 6-vertex model in under domain wall boundary conditions, where each vertex satisfies the ice-rule constraint (two incoming and two outgoing s), and the paths trace the arrow directions along the edges. Integer partitions admit a natural representation via Ferrers diagrams, which are left-justified arrays of boxes where the i-th row contains \lambda_i boxes, with \lambda_1 \geq \lambda_2 \geq \cdots \geq \lambda_k > 0. The outer boundary of a Ferrers diagram for the partition \lambda \vdash m traces a from (0, \lambda_1) to (\ell(\lambda), 0) using right (east) steps (1,0) and down (south) steps (0,-1), staying weakly above the line y = -\frac{\lambda_1}{\ell(\lambda)} x + \lambda_1 but more importantly forming a decreasing skyline. The Durfee square of \lambda, defined as the largest integer d such that \lambda_d \geq d (i.e., the side length of the largest square fitting in the upper-left corner of the diagram), corresponds in this path interpretation to the largest d where the path remains above the diagonal y = \lambda_1 - x for the initial segment. This connection facilitates generating functions and identities for classified by Durfee square size, decomposing \lambda into the square plus two smaller partitions attached to its right and below. Plane partitions extend integer partitions to weakly decreasing arrays \pi_{i,j} \geq \pi_{i+1,j} \geq \pi_{i,j+1} \geq 0 for $1 \leq i \leq a, $1 \leq j \leq b, visualized as stacked in forming a corner . These structures biject to families of non-intersecting on the , where each "layer" or height corresponds to paths from starting points along one to ending points along another, analogous to the path encodings for partitions but in a higher-dimensional . For plane partitions confined to an a \times b \times c (with parts at most c), the bijection to \min(a,b) non-intersecting paths enables the application of determinantal , confirming MacMahon's classical product formula for the total count: \prod_{i=1}^{a} \prod_{j=1}^{b} \prod_{k=1}^{c} \frac{i + j + k - 1}{i + j + k - 2}. This formula arises from the path determinants via the Lindström-Gessel-Viennot lemma, providing a combinatorial proof of the generating function. A basic illustration of these tiling-path connections appears in the domino tilings of a $2 \times n board, which number the (n+1)-th Fibonacci number F_{n+1} (with F_1 = 1, F_2 = 1, F_3 = 2, etc.), satisfying the recurrence F_{n+1} = F_n + F_{n-1}. This count links directly to lattice path enumerations sharing the same recurrence, such as the number of paths from (0,0) to (n,0) composed of u up steps (1,1), d down steps (1,-1), and horizontal steps (2,0) that balance heights appropriately, mirroring the tiling's vertical and paired-horizontal placements in the path model.

Probabilistic Models

Lattice paths provide a natural framework for modeling probabilistic processes, particularly through their connection to and stochastic paths in spaces. The simple symmetric on the integer line \mathbb{Z}, which can be viewed as the projection of a one-dimensional path, starts at the and at each step moves left or right with equal probability $1/2. This model is recurrent, meaning it returns to the starting point infinitely often with probability 1, a property established by analyzing the expected number of returns using generating functions or . In contrast, extensions to higher dimensions reveal transience for d \geq 3, where the walk escapes to infinity without returning, highlighting how dimensionality affects path behavior in probabilistic models. The ballot theorem extends to probabilistic interpretations, quantifying the likelihood that one maintains a lead throughout in an scenario. Specifically, if A receives a votes and B receives b votes with a > b, the probability that A is always ahead is (a - b)/(a + b), assuming uniform randomness over vote sequences. This result, derived via reflection principles on paths, models the path staying above the diagonal and has been applied to assess lead probabilities in sequential processes. In the scaling limit, uniform random Dyck paths of length $2n, which are lattice paths from (0,0) to (2n,0) staying non-negative, converge to the after appropriate . The height function of such paths, scaled by \sqrt{n}, approaches the standard , a continuous starting and ending at zero with non-negative values, as shown through in the Skorokhod space. This limit shape connects discrete to continuous probability and is analogous to the Vershik-Kerov-Logan-Shepp curve observed in random Young diagrams, though specifically for the excursion measure here. Lattice paths also underpin applications in queueing theory and statistical physics. In the M/M/1 queue, the workload process corresponds to a reflected simple on the non-negative integers, where arrivals and services drive up and down steps, respectively, and at zero models idle periods. The and busy period probabilities derive from path enumeration techniques, such as cycle lemmas, to compute return times. In statistical physics, directed lattice paths model linear , with path weights incorporating interactions; for instance, self-avoiding walks simulate polymer chains, and their functions yield free energies via methods. Non-intersecting path ensembles, or "vicious walkers," describe multi-polymer systems, where repulsion enforces exclusion, leading to determinantal point processes in the limit.

References

  1. [1]
    Lattice Path -- from Wolfram MathWorld
    A path composed of connected horizontal and vertical line segments, each passing between adjacent lattice points.
  2. [2]
  3. [3]
    [PDF] Lattice Path Combinatorics
    One suitable definition for lattice paths, is the use of one variable as the length of the path, and the second one as the final height of the path, i.e. ...
  4. [4]
    None
    Summary of each segment:
  5. [5]
    [PDF] Chapter 10 - Lattice Path Enumeration - Universität Wien
    at the outset. — reasonably simple combinatorial objects, the study of physical, probabilistic, or statistical ...
  6. [6]
    Basic analytic combinatorics of directed lattice paths - ScienceDirect
    Basic analytic combinatorics of directed lattice paths. To Maurice Nivat, with many thanks for so many things! Author links open overlay panel. Cyril Banderier
  7. [7]
    lattice paths and ballot numbers - PlanetMath
    lattice paths and ballot numbers. A lattice path is a path in Rn ℝ n whose vertices are lattice points in some lattice ...
  8. [8]
    Lattice grids and prisms are antimagic - ScienceDirect.com
    Apr 20, 2007 · In this article, we prove that these two classes of graphs are antimagic, by constructing such antimagic labellings.<|separator|>
  9. [9]
    Combinatorial Generation Algorithms for Some Lattice Paths Using ...
    A North-East lattice path is a lattice path in the plane which begins at ( 0 , 0 ) , ends at ( n , m ) , and consists of steps ( 0 , 1 ) and ( 1 , 0 ) [19]. The ...
  10. [10]
    [PDF] Enumeration of Lattice Paths with Restrictions
    This thesis delves into enumerations of some of the most common lattice paths – north-east paths, up-down paths, and Dyck paths – with restrictions applied.
  11. [11]
    [PDF] MAT344 Lecture 4 - University of Toronto Mathematics
    May 16, 2019 · A lattice path in the plane is a curve made up of line segments that either go from a point (i, j) to the point (i + 1,j) or from a point (i, j) ...
  12. [12]
    [PDF] Enumerative Combinatorics Volume 1 second edition - Mathematics
    The basic problem of enumerative combinatorics is that of counting the number of elements of a finite set. Usually we are given an infinite collection of ...
  13. [13]
    Dyck Path -- from Wolfram MathWorld
    {1,1,1,1,-1,-1,-. See also. Catalan Number, Lattice Path, Narayana Number, Staircase Walk. Explore with Wolfram|Alpha. WolframAlpha. More things to try: aleph0^ ...
  14. [14]
    [PDF] An invitation to analytic combinatorics and lattice path counting
    Moreover, lattice paths have many applications, for instance in physics and computer science, where they are used for the solution of integer programming ...
  15. [15]
    [PDF] Counting Dyck Paths - Jeremy Martin
    A Dyck path of length 2n is a diagonal lattice path from (0, 0) to (2n, 0), consisting of n up-steps (along the vector (1, 1)) and n down-steps (along the ...
  16. [16]
    [PDF] Dyck Paths and Up-Down Walks - MIT Mathematics
    May 12, 2016 · A Dyck Path is a series of up and down steps. The path will begin and end on the same level; and as the path moves from left to right it will ...
  17. [17]
    [PDF] On binary trees and Dyck paths - Numdam
    Two constructions are given which enable to pass from a Dyck path to a binary tree and from a binary tree to a Dyck path. 1. INTRODUCTION. It is well known that ...
  18. [18]
    Catalan Number -- from Wolfram MathWorld
    The Catalan numbers on nonnegative integers n are a set of numbers that arise in tree enumeration problems of the type,
  19. [19]
    [PDF] 18.212 S19 Algebraic Combinatorics, Lecture 1: Catalan numbers II
    Feb 8, 2019 · Proof by reflection principle. Recall that Cn counts the number of Dyck paths that are above the x-axis from (0, 0) to. (2n, 0). First of all ...
  20. [20]
    [PDF] 1 Catalan numbers - Durham University
    We want to show that good paths constitute precisely 1/(2n + 1) part of all paths. Let us switch to sequences of ±1.
  21. [21]
    [PDF] Pattern-Avoidance and Fuss–Catalan Numbers
    n, with n = 1,2,...,8 and p = 1,2,...,5. so the ak(n) form a sub-family of the Raney numbers, which includes the Fuss–Catalan numbers. ak(km + j)tm for 0 ≤ j ≤ k. ...
  22. [22]
    [PDF] Four Proofs of the Ballot Theorem 1
    The most famous and elegant of these variations is the “reflection method” (often misattributed to André) in which ballot permutations are represented as ...
  23. [23]
    Ballot Lemma - Interactive Mathematics Miscellany and Puzzles
    The Ballot Theorem.​​ The solution to the ballot problem is [(u - kv) / (u + v)] C(u + v, u). For a more detailed history of the problem and a collection of ...
  24. [24]
    [PDF] The ballot theorems - SJÄLVSTÄNDIGA ARBETEN I MATEMATIK
    Jan 16, 2017 · In this paper, we will focus on the original ballot theorem and its first generalization, that is the probability for a candidate to have ...
  25. [25]
    [PDF] André's Actual Method and its Application to the Generalized Ballot ...
    Many excellent articles and reputable sources cite André [2] claiming that he used the “reflection method” (or “reflection principle”) to solve the ballot.Missing: statement | Show results with:statement
  26. [26]
    On The Ballot Theorems | SpringerLink
    This paper deals with the historical background and the development of these theorems, analyzes various proofs and gives some applications.
  27. [27]
    [PDF] Maintaining the Spirit of the Reflection Principle when the Boundary ...
    Aug 18, 2003 · This solu- tion to the Generalized Ballot Problem is in the spirit of the reflection principle for the Ballot Problem (the case k = 1), but it ...
  28. [28]
    André's Actual Method and Its Application to the Generalized Ballot ...
    To be fair, the reflection method is a variation of Andre's method of proof. Comtet [6, p. 22] writes when introducing the method, "We first formulate the.Missing: Theorem | Show results with:Theorem
  29. [29]
    On Generalized Delannoy Paths - SIAM.org
    A Delannoy path is a minimal path with diagonal steps in ℤ 2 between two arbitrary points. We extend this notion to the n dimensions space ℤ 𝑛 and identify ...Missing: multidimensional | Show results with:multidimensional
  30. [30]
    The Art of Computer Programming (TAOCP) - CS Stanford
    These books were named among the best twelve physical-science monographs of the century by American Scientist, along with: Dirac on quantum mechanics, Einstein ...Missing: Dyck | Show results with:Dyck
  31. [31]
    Bijections from Dyck paths to 321-avoiding permutations revisited
    Nov 16, 2007 · There are three bijections (B, K, M) from Dyck paths to 321-avoiding permutations. M = B \circ L = K \circ L', where L and L' are involutions.
  32. [32]
    The run transform - ScienceDirect.com
    Oct 6, 2012 · There is a well known bijection from Dyck paths to noncrossing partitions, whose origin we have been unable to trace. Traverse the Dyck path ...
  33. [33]
    Non-intersecting paths, random tilings and random matrices
    May 17, 2002 · We investigate certain measures induced by families of non-intersecting paths in domino tilings of the Aztec diamond, rhombus tilings of an abc- ...
  34. [34]
    [PDF] Partitions with constrained ranks and lattice paths - arXiv
    Nov 16, 2022 · The key is to interpret these maps in terms of lattice paths and to prove that the Durfee square of the partition corresponds to the number of ...
  35. [35]
    [PDF] Determinants, Paths, and Plane Partitions - Brandeis
    We shall first apply Theorem 1 to the digraph in which the vertices are lattice points in the plane and the edges go from (i, j) to (i, j + 1) and (i +1,j). ...Missing: 3D | Show results with:3D
  36. [36]
    [PDF] Domino Tilings of 2 × n Grids (or Perfect Matchings of Grid Graphs ...
    Abstract. It is well known that the number of edge-labeled perfect matchings of a 2×n planar grid graph is the (n + 1)st Fibonacci number.<|control11|><|separator|>
  37. [37]
    [PDF] Random walks
    Theorem 2.22 The simple symmetric random walk on Zd is recurrent in dimensions d = 1, 2 and transient in dimensions d ≥ 3. Proof. To apply the previous lemma, ...
  38. [38]
    [PDF] dyck-paths.pdf - LaBRI
    2n+1 × [1,2n + 1]. We state the cycle Lemma as follows: Lemma 2 (Cycle Lemma) There exists a one-to-one correspondence Ψ2n ...
  39. [39]
    [PDF] arXiv:1406.5156v2 [math.PR] 12 Jun 2015
    Jun 12, 2015 · Dyck paths and Brownian excursion. All of our results are derived from bijections be- tween Dyck paths and pattern-avoiding permutations.
  40. [40]
    Lattice path counting and the theory of queues - ScienceDirect
    In this paper we will show how recent advances in the combinatorics of lattice paths can be applied to solve interesting and nontrivial problems in the ...
  41. [41]
    Statistical mechanics of directed models of polymers in the square ...
    Apr 3, 2003 · In this review the methods for obtaining generating functions and determining free energies in directed lattice path models of linear polymers ...
  42. [42]
    Lattice paths: vicious walkers and friendly walkers - ScienceDirect.com
    Friendly walkers refers to a model in which any number P of directed lattice paths, starting at adjacent lattice sites, simultaneously proceed in one of the ...