Choosing Tractable Partitions for Search-Then-Optimize Planning

Determine which partition should be used in search-then-optimize planning and whether the conditioned problem induced by that partition is tractable.

Background

The paper discusses partitioned optimization methods in which a discrete quantity, such as a mode assignment in a mixed-integer linear program, is fixed before solving a conditioned continuous problem. Existing approaches primarily focus on accelerating the search over an already declared discrete structure. The unresolved issue is to determine which partition is appropriate and whether fixing that partition actually produces a tractable conditioned problem. This issue motivates the paper’s solver-sufficient symbolization framework, which seeks partitions whose conditioned problems admit efficient and faithful solution procedures.

References

In each, the discrete structure is declared by the formulation and the effort goes into searching it faster; which partition to use, and whether the conditioned problem is tractable under it, remain open \citep{audet2026partitioned}.

— What Makes a Good Symbol for Hybrid Discrete-Continuous Planning?  (2610.23010 - Pan et al., 16 Oct 2026) in Appendix, Section 1, subsection “Search-then-Optimize: From MILP to Energy-Based Models” (Appendix Section \ref{app:rw_partition})