MAPLE Agent for Dynamic Optimization via Memory and Evolutionary Search

MAPLE uses language-based problem construction and evolutionary search to maintain optimization models through successive natural-language updates. It retains previous optimization programs and accepted plans to handle dynamic changes in constraints and objectives without restarting the search process.