Adapting continuous metaheuristics to discrete problems: A taxonomy-oriented review

Continuous metaheuristics are difficult to apply directly to discrete optimization because their real-valued operations are incompatible with structured discrete solution spaces. Existing studies also lack a unified framework for classifying and comparing discretization strategies. Following a systematic multi-database search and mechanism-focused screening, this review constructs a purposively curated corpus of 299 studies published between 2008 and 2025. It develops a two-dimensional taxonomy of discretization logic and discrete problem structure, formally defines three mechanisms—Operator Replacement, Mapping, and Encoding Reconstruction—and clarifies their boundaries and strictly defined hybrid combinations. The review also identifies five non-exclusive solution structures and an additional Mixed-type label, and descriptively examines recurring method–problem associations within the curated corpus. The framework standardizes terminology and offers practice-informed, non-prescriptive considerations for selecting and designing discretization strategies for continuous metaheuristics.