When both Grounding and not Grounding are Bad -- A Partially Grounded Encoding of Planning into SAT (Extended Version)

Collection of recent AI/ML research on planning, agentic systems, and formal methods. Papers introduce: a partially-grounded SAT encoding for classical planning that scales linearly with plan length; DGM-Hyperagents, a self-referential self-improving agent architecture; methods for part-level vector sketch generation; fine-tuning LLMs to generate formally verifiable counterexamples (Lean 4); ItinBench, a multi-dimension itinerary/route benchmark; PA2D-MORL for better Pareto frontier approximation in multi-objective RL; PowerLens, an LLM-driven, constraint-checked personalized mobile power管理/DP