The Optimal Merge Pattern problem finds the most efficient way to merge a set of sorted files into a single sorted file, minimizing total record comparison movements.

Repeatedly extract the two files of smallest sizes from a min-priority queue, merge them (cost = sum of sizes), and insert the resulting file back.

An optimal merge pattern for N files consists of merging the two smallest files first, then solving the remaining N-1 file merges optimally.

Proof by Huffman coding equivalence: since record movements are weighted by the depth of leaf files, placing larger files closer to the root (merging them last) minimizes total weighted depth.