Introduction to Graph Theory · Problem 2.23

Simple graphs — Darij Grinberg

Mathematics — Graph Theory · hamiltonian-cycles-hypercube-variants

Problem

Let nn and kk be two integers such that n>k>0n > k > 0. Define the simple graph Qn,kQ_{n,k} as follows: Its vertices are the bitstrings (a1,a2,…,an)∈{0,1}n(a_1, a_2, \ldots, a_n) \in \{0,1\}^n; two such bitstrings are adjacent if and only if they differ in exactly kk bits (in other words: two vertices (a1,a2,…,an)(a_1, a_2, \ldots, a_n) and (b1,b2,…,bn)(b_1, b_2, \ldots, b_n) are adjacent if and only if the number of i∈{1,2,…,n}i \in \{1, 2, \ldots, n\} satisfying ai≠bia_i \neq b_i equals kk). (Thus, Qn,1Q_{n,1} is the nn-hypercube graph QnQ_n.) (a) Does Qn,kQ_{n,k} have a hamc when kk is even? (Recall that ``hamc'' is short for ``Hamiltonian cycle''.) (b) Does Qn,kQ_{n,k} have a hamc when kk is odd? [Hint: One way to approach part (b) is by identifying the set {0,1}\{0,1\} with the field F2\mathbb{F}_2 with two elements. The bitstrings (a1,a2,…,an)∈{0,1}n(a_1, a_2, \ldots, a_n) \in \{0,1\}^n thus become the size-nn row vectors in the F2\mathbb{F}_2-vector space F2n\mathbb{F}_2^n. Let e1,e2,…,ene_1, e_2, \ldots, e_n be the standard basis vectors of F2n\mathbb{F}_2^n (so that eie_i has a 11 in its ii-th position and zeroes everywhere else). Then, two vectors are adjacent in the nn-hypercube graph QnQ_n (resp. in the graph Qn,kQ_{n,k}) if and only if their difference is one of the standard basis vectors (resp., a sum of kk distinct standard basis vectors). Try to use this to find a graph isomorphism from QnQ_n to a subgraph of Qn,kQ_{n,k}.]
Intuitive explanation

The key insight is to use a parity argument combined with a bipartite structure hidden in the graph. Think of each bitstring as a vector in F2n\mathbb{F}_2^n, and notice that moving from one vertex to an adjacent vertex changes the Hamming weight (number of 1's) by exactly kk. When kk is even, this preserves parity of the weight; when kk is odd, it flips parity. A Hamiltonian cycle must visit all 2n2^n vertices, and the parity constraint becomes a topological obstruction that either allows or forbids such a cycle depending on whether kk is even or odd.

Intuitive explanation

The key insight is to use a parity argument combined with a bipartite structure hidden in the graph. Think of each bitstring as a vector in F2n\mathbb{F}_2^n, and notice that moving from one vertex to an adjacent vertex changes the Hamming weight (number of 1's) by exactly kk. When kk is even, this preserves parity of the weight; when kk is odd, it flips parity. A Hamiltonian cycle must visit all 2n2^n vertices, and the parity constraint becomes a topological obstruction that either allows or forbids such a cycle depending on whether kk is even or odd.

Browse the problem library