Adapting the mSTreg heuristic to non-complete or non-Euclidean graphs

Develop and assess adaptations of the mSTreg heuristic for settings where the underlying graph is not complete or where distances are non-Euclidean, and determine whether the algorithm performs well in these regimes.

Background

The proposed mSTreg heuristic is demonstrated primarily on complete Euclidean graphs. Many practical applications involve sparse (non-complete) graphs or non-Euclidean distance structures.

The authors explicitly state uncertainty about the algorithm’s performance under such generalizations, identifying this extension as an open question.