The compaction problem is to partition the vertices of an input graph G onto the vertices of a fixed target graph H, such that adjacent vertices of G remain adjacent in H, and every vertex and non‐loop edge of H is covered by some vertex and edge of G respectively, i.e., the partition is a homomorphism of G
onto
H (except the loop edges). Various computational complexity results, including both NP‐completeness and polynomial time solvability, have been presented earlier for this problem for various classes of target graphs H. In this paper, we pay attention to the input graphs G, and present polynomial time algorithms for the problem for some class of input graphs, keeping the target graph H general as any reflexive or irreflexive graph. Our algorithms also give insight as for which instances of the input graphs, the problem could possibly be NP‐complete for certain target graphs. With the help of our results, we are able to further refine the structure of the input graph that would be necessary for the problem to be possibly NP‐complete, when the target graph is a cycle. Thus, when the target graph is a cycle, we enhance the class of input graphs for which the problem is polynomial time solvable. We also present analogous results for a variation of the compaction problem, which we call the vertex‐compaction problem. Using our results, we also provide important relationships between compaction, retraction, and vertex‐compaction to cycles.
