This course studies the combinatorial and algebraic structure of partially ordered sets and lattices, with an emphasis on how local order data controls global enumeration and topology. It develops the basic language of finite posets, then moves into the central invariants that measure their size and shape, such as chains, antichains, ranks, and interval structure. From there, the course introduces Möbius functions, incidence algebras, and inversion formulas as algebraic tools for counting and organizing combinatorial information.

The main themes are decomposition, duality, and structure: how posets can be broken into chains and antichains, how lattice operations and closure operators encode algebraic behavior, and how geometric lattices and matroids connect order theory to geometry and independence. Later chapters broaden the perspective to Eulerian posets, flag enumeration, order complexes, and topological properties such as shellability and Cohen-Macaulayness, showing how combinatorial posets can carry deep geometric and homological information. The final chapters bring these ideas back to enumeration through $P$-partitions and order polynomials, culminating in a synthesis of local interval behavior and global counting principles.

# Introduction

This opening chapter sets the direction for the course. We study finite partially ordered sets as objects that can be drawn, counted, multiplied, inverted, and topologised. The guiding theme is that order relations encode algebraic structure: once a set has a partial order, it has intervals, incidence functions, chains, rank data, and often a lattice operation. Chapters 1 through 5 develop these ideas through finite-poset constructions, decomposition theorems, incidence algebras, Möbius inversion, and lattices; here we fix the viewpoint and introduce the examples that will reappear throughout the course.

The course begins from undergraduate algebra, basic combinatorics, linear algebra, elementary graph theory, and familiarity with generating functions. Later chapters use a small amount of commutative algebra, simplicial topology, matroid theory, representation theory, and homological language; when those topics enter, the note treats them as structured tools for finite posets rather than as assumed background. The central objects are finite, so the emphasis is not on set-theoretic pathology. Instead, finite posets provide a common language for enumeration, representation by diagrams, inversion formulas, lattice theory, and simplicial topology.

## What Does A Poset Remember?

What information is retained when we forget the labels of a collection and keep only the comparisons among its elements? A partial order records which elements are below which others, but it also records many derived objects: chains, antichains, intervals, ideals, filters, and linear extensions. These objects are not decoration; they are the basic data from which the algebraic and enumerative invariants of the course are built.

[definition: Finite Poset] A finite poset is a pair $(P, \le)$ where $P$ is a finite set and $\le$ is a binary relation on $P$ satisfying, for all $x,y,z \in P$, \begin{align*} x \le x, \qquad x \le y \text{ and } y \le x \implies x = y, \qquad x \le y \text{ and } y \le z \implies x \le z. \end{align*} [/definition]

The three axioms are reflexivity, antisymmetry, and transitivity. In finite examples, the order is often represented by a Hasse diagram, where only cover relations are drawn and all other comparisons are recovered by transitivity.

[example: Divisibility Poset] Let $P=\{1,2,3,4,6,12\}$ and define $a \le b$ to mean $a \mid b$. The relation is reflexive because $a=1\cdot a$, so $a\mid a$ for every $a\in P$. It is antisymmetric because if $a\mid b$ and $b\mid a$, then $b=ra$ and $a=sb$ for positive integers $r,s$, hence $a=sra$, so $sr=1$, and therefore $r=s=1$ and $a=b$. It is transitive because if $a\mid b$ and $b\mid c$, then $b=ra$ and $c=tb$ for positive integers $r,t$, so $c=t(ra)=(tr)a$, hence $a\mid c$. The minimum element is $1$, since \begin{align*} 1\mid 1,\ 1\mid 2,\ 1\mid 3,\ 1\mid 4,\ 1\mid 6,\ 1\mid 12. \end{align*} The maximum element is $12$, since \begin{align*} 1\mid 12,\ 2\mid 12,\ 3\mid 12,\ 4\mid 12,\ 6\mid 12,\ 12\mid 12. \end{align*} The element $2$ lies below $4,6,12$ because $4=2\cdot 2$, $6=3\cdot 2$, and $12=6\cdot 2$. The elements $4$ and $6$ are incomparable: $4\nmid 6$ because $6/4$ is not an integer, and $6\nmid 4$ because $4/6$ is not an integer. For the interval from $2$ to $12$, an element $z\in P$ belongs to $[2,12]$ exactly when $2\mid z$ and $z\mid 12$. Checking the six elements of $P$, the possibilities are $2,4,6,12$, while $1$ fails $2\mid 1$ and $3$ fails $2\mid 3$. Thus \begin{align*} [2,12]=\{2,4,6,12\}. \end{align*} This example turns divisibility calculations into order-theoretic data: extrema, comparable pairs, incomparable pairs, and intervals are all read from the same relation $a\mid b$. [/example]

This example already illustrates the main habit of the course: translate a familiar combinatorial or algebraic situation into order language, then use the order to compute. The next basic example is more symmetric and will serve as the testing ground for many theorems.

[example: Boolean Lattice] Let $B_n$ be the set of all subsets of $\{1,\dots,n\}$ ordered by inclusion. For a subset $A\subseteq \{1,\dots,n\}$, its rank is $\rho(A)=|A|$, so the rank-$k$ elements are exactly the subsets with $k$ elements: \begin{align*} \{A\in B_n:\rho(A)=k\}=\{A\subseteq \{1,\dots,n\}:|A|=k\}. \end{align*} There are $\binom{n}{k}$ such subsets, because choosing an element of rank $k$ is the same as choosing which $k$ of the $n$ elements belong to it. A maximal chain from $\varnothing$ to $\{1,\dots,n\}$ must increase the subset size by one at each step: \begin{align*} \varnothing=A_0\subset A_1\subset A_2\subset \cdots \subset A_n=\{1,\dots,n\}, \end{align*} with $|A_i|=i$ for every $i$. Since $A_i$ has one more element than $A_{i-1}$, there is a unique element $a_i\in \{1,\dots,n\}$ such that \begin{align*} A_i=A_{i-1}\cup\{a_i\}. \end{align*} Thus the chain determines the ordered list $(a_1,\dots,a_n)$, and every element of $\{1,\dots,n\}$ appears exactly once in that list. Conversely, any permutation $(a_1,\dots,a_n)$ defines a maximal chain by \begin{align*} A_i=\{a_1,\dots,a_i\}. \end{align*} Therefore the maximal chains in $B_n$ are in bijection with the $n!$ permutations of $\{1,\dots,n\}$. This single poset records binomial coefficients through its rank sizes and permutations through its maximal chains, which is why it reappears in symmetric chain decompositions and in the Sperner problem. [/example]

The Boolean lattice shows why posets are not merely diagrams. Its ranks form Pascal's triangle, its maximal chains encode permutations, and its antichains lead to extremal set theory. Many later results will be first proved or tested in $B_n$ before being transported to more general graded posets.

## Why Intervals Matter?

How can local order data determine global algebraic formulas? The answer begins with intervals. Most invariants in the course are computed not from the whole poset at once, but from the subposets lying between two comparable elements.

[definition: Interval In A Poset] Let $(P, \le)$ be a poset and let $x,y \in P$ with $x \le y$. The interval from $x$ to $y$ is \begin{align*} [x,y] := \{z \in P : x \le z \le y\}. \end{align*} [/definition]

Intervals are the local pieces on which incidence functions live. To turn this local viewpoint into algebra, we need a basic incidence function that records exactly where comparisons occur and that can later be inverted.

[definition: Zeta Function Of A Finite Poset] Let $(P, \le)$ be a finite poset. The zeta function of $P$ is the function \begin{align*} \zeta_P : \{(x,y) \in P \times P : x \le y\} \to \mathbb{Z} \end{align*} given by \begin{align*} \zeta_P(x,y) := 1 \quad \text{for } x \le y. \end{align*} [/definition]

The zeta function is the order-theoretic analogue of the constant function $1$ on intervals. Once summation over lower elements is written using $\zeta_P$, the natural next question is whether that summation operation can be reversed in the same order-theoretic language.

For a finite poset, this reversal is multiplication in the incidence algebra of functions on comparable pairs, where convolution is the product obtained by summing over intermediate elements of an interval. The result below is the finite-poset analogue of inverting a triangular summation matrix: it says that every cumulative function on lower ideals has a unique local inverse, and that inverse is governed by a new invariant $\mu_P$ attached to intervals of the order. The theorem below says that the inverse exists and is again controlled interval by interval. Thus Möbius inversion is not an extra structure imposed on a poset; it is the algebraic inverse of the most basic incidence function attached to the order.