ProofComplete

Tree Characterizations - Proof

Trees admit multiple equivalent characterizations. We prove that six natural conditions are equivalent, each providing different insights and proof techniques for tree properties.

TheoremEquivalent Characterizations of Trees

For a graph TT with nn vertices, the following are equivalent:

  1. TT is connected and acyclic
  2. TT is connected and has n1n-1 edges
  3. TT is acyclic and has n1n-1 edges
  4. TT is minimally connected
  5. TT is maximally acyclic
  6. Every pair of vertices has a unique path
Proof

We prove the cycle (1)(2)(3)(1)(1) \Rightarrow (2) \Rightarrow (3) \Rightarrow (1) and (1)(4)(5)(6)(1)(1) \Rightarrow (4) \Rightarrow (5) \Rightarrow (6) \Rightarrow (1).

(1) \Rightarrow (2): We prove by induction on nn that a connected acyclic graph with nn vertices has n1n-1 edges.

Base case: n=1n=1 gives 0 edges. ✓

Inductive step: Let TT be connected and acyclic with n2n \geq 2 vertices. Since TT is connected, every vertex has degree at least 1. By the handshaking lemma, deg(v)=2m\sum \deg(v) = 2m where mm is the number of edges. If all degrees were at least 2, we'd have 2m2n2m \geq 2n, so mnm \geq n.

But a connected graph with nn vertices and at least nn edges must contain a cycle (by pigeonhole on any spanning tree). This contradicts acyclicity. Therefore, some vertex vv has degree 1.

Removing vv and its incident edge yields T=TvT' = T - v, which is connected and acyclic with n1n-1 vertices. By induction, TT' has (n1)1=n2(n-1)-1 = n-2 edges. Thus TT has (n2)+1=n1(n-2) + 1 = n-1 edges.

(2) \Rightarrow (3): Suppose TT is connected with n1n-1 edges. If TT contained a cycle CC, removing any edge from CC would leave TT connected (can route around the cycle). This contradicts minimality: a connected graph requires at least n1n-1 edges. Thus TT is acyclic.

(3) \Rightarrow (1): Let TT be acyclic with nn vertices and n1n-1 edges. Suppose TT has kk connected components, each a tree. If component ii has nin_i vertices, it has ni1n_i - 1 edges (by (1)\Rightarrow(2) applied to each component). Summing: n1=i=1k(ni1)=nkn-1 = \sum_{i=1}^k (n_i - 1) = n - k Thus k=1k = 1, so TT is connected.

(1) \Rightarrow (4): Suppose TT is connected and acyclic. Removing any edge ee disconnects TT: otherwise, there would be a path avoiding ee between ee's endpoints, creating a cycle. Thus TT is minimally connected.

(4) \Rightarrow (5): Let TT be minimally connected. Then TT is acyclic (else removing an edge from a cycle maintains connectivity). Adding any edge {u,v}\{u,v\} creates exactly one cycle: the existing path from uu to vv plus the new edge. Thus TT is maximally acyclic.

(5) \Rightarrow (6): If some pair u,vu,v had no path, adding edge {u,v}\{u,v\} wouldn't create a cycle, contradicting maximality. If some pair had two distinct paths, these paths form a cycle, contradicting acyclicity. Thus every pair has exactly one path.

(6) \Rightarrow (1): Unique paths imply connectivity. If a cycle existed, its vertices would have two distinct paths (clockwise and counterclockwise around the cycle), contradicting uniqueness. Thus TT is connected and acyclic.

Remark

These characterizations are fundamental tools. Property (2) gives edge counting arguments, (4) shows trees are critical for connectivity, (5) shows they're extremal for acyclicity, and (6) provides uniqueness for shortest paths and routing algorithms.