ProofComplete

Proof of Dilworth Theorem via Induction

We present a direct inductive proof of Dilworth's theorem that constructs the chain decomposition explicitly. This approach provides insight into the structure of posets and offers an algorithm for finding optimal decompositions.

TheoremDilworth's Theorem (Restated)

For any finite poset PP with maximum antichain size ww, there exists a partition of PP into ww chains.

Proof

We proceed by strong induction on ∣P∣|P|.

Base case: If ∣P∣=1|P| = 1, then w=1w = 1 and PP itself is a single chain. βœ“

Inductive step: Assume the theorem holds for all posets with fewer than nn elements. Let ∣P∣=n|P| = n and let AA be a maximum antichain in PP with ∣A∣=w|A| = w.

Case 1: A=PA = P (all elements are pairwise incomparable). Then PP is itself an antichain, and we can partition it into ww chains of length 1 (singletons). βœ“

Case 2: A≠PA \neq P (there exist comparable elements). Define:

  • U={x∈P:x>aΒ forΒ someΒ a∈A}U = \{x \in P : x > a \text{ for some } a \in A\} (elements above AA)
  • L={x∈P:x<aΒ forΒ someΒ a∈A}L = \{x \in P : x < a \text{ for some } a \in A\} (elements below AA)

Claim: Every maximal antichain in UU has size at most ww.

Proof of claim: Suppose BB is an antichain in UU with ∣B∣>w|B| > w. For each b∈Bb \in B, choose some ab∈Aa_b \in A with ab<ba_b < b. Since ∣B∣>w=∣A∣|B| > w = |A|, by pigeonhole principle, some a∈Aa \in A satisfies a<b1a < b_1 and a<b2a < b_2 for distinct b1,b2∈Bb_1, b_2 \in B.

But then Aβˆͺ{b1,b2}βˆ–{a}A \cup \{b_1, b_2\} \setminus \{a\} contains at least w+1w + 1 elements, and we claim it's an antichain:

  • Elements of Aβˆ–{a}A \setminus \{a\} are pairwise incomparable
  • b1,b2∈Ub_1, b_2 \in U are incomparable (both in antichain BB)
  • No element of Aβˆ–{a}A \setminus \{a\} is comparable to b1b_1 or b2b_2 (else b1,b2∉Ub_1, b_2 \not\in U)

This contradicts maximality of AA. Thus BB has size ≀w\leq w. βœ“

By symmetry, every antichain in LL also has size ≀w\leq w.

Applying induction:

  • By induction, UU can be partitioned into ≀w\leq w chains C1U,…,CkUC_1^U, \ldots, C_k^U where k≀wk \leq w
  • By induction, LL can be partitioned into ≀w\leq w chains C1L,…,CmLC_1^L, \ldots, C_m^L where m≀wm \leq w

Constructing the decomposition of PP: For each a∈Aa \in A, we extend chains:

  • Find a chain CiLC_i^L with maximum element <a< a (if exists in LL)
  • Find a chain CjUC_j^U with minimum element >a> a (if exists in UU)
  • Combine: CiLβˆͺ{a}βˆͺCjUC_i^L \cup \{a\} \cup C_j^U forms a single chain through aa

Since ∣A∣=w|A| = w and we use at most ww chains from UU and ww from LL, we can construct a chain decomposition of PP using exactly ww chains.

The matching of chains from LL and UU through elements of AA uses the fact that k,m≀wk, m \leq w and ∣A∣=w|A| = w, allowing us to merge chains through antichain elements to achieve exactly ww total chains.

β– 
Remark

This proof is constructive and yields an algorithm:

  1. Find a maximum antichain AA
  2. Recursively decompose UU (elements above AA) and LL (elements below AA)
  3. Connect chains through elements of AA

The algorithm runs in polynomial time when maximum antichain finding is efficient.

This inductive approach reveals the hierarchical structure underlying Dilworth's theorem and provides intuition for poset decomposition algorithms.