Folding lattice proteins confined on minimal grids using a quantum encoding (arxiv.org)

The authors recast the problem of folding a single lattice protein confined to a minimal (no-free-site) grid—a tightly packed, sterically constrained NP-hard optimization—into a quadratic unconstrained binary optimization (QUBO) formulation and benchmark solvers. Using that encoding they reliably find minimum-energy conformations for chain length 48, comparing classical simulated annealing, linear/quadratic programming, and a hybrid quantum-classical annealer on a D-Wave system. The hybrid runs solved instances in about 10 seconds, and all heuristic methods were validated against expensive exhaustive enumeration used as ground truth. For the AI/ML and optimization community this is a compact, practical demonstration that dense, constraint-heavy biological packing problems can be mapped to QUBO and tackled by both classical and near-term quantum annealing hardware. Key technical takeaways: the QUBO mapping preserves chain constraints that foil simpler linear/quadratic programming approaches (which perform well only on unconstrained lattice-gas variants), classical simulated annealing remains competitive, and hybrid quantum annealing offers notable wall-clock speed for these instance sizes. The work highlights promising avenues for applying QUBO-based solvers to scheduling-like, sterically constrained optimization problems in biophysics, while underscoring the need to study scaling beyond minimal grids and to compare solution quality and resource trade-offs for larger, more realistic systems.