What happens if you try building stealth solutions in theories where the coupling function changes sign?

Your question asks about what happens when you try to construct stealth (presumably meaning hidden or protected) solutions in theories where the coupling function changes sign—but I need to be direct: the preprints provided don't address stealth solutions or their construction in sign-changing coupling regimes.

However, the archive *does* contain relevant material on negative coupling theories more broadly. Here's what we know about the regime:

Negative couplings are surprisingly well-behaved quantum mechanically. Despite having classically unstable potentials (like an upside-down bowl), theories with negative quartic self-interactions develop stable quantum ground states through loop corrections [AG-2026.03-1028]. This matters because it suggests negative coupling doesn't automatically break the theory—it rewires quantum behavior in non-obvious ways.

Sign reversals in operators don't automatically cause logical disasters. While some Lorentz-invariant theories with negative-coefficient higher-dimensional operators do permit superluminality and even closed timelike curves classically, quantum backreaction from low-energy quanta prevents these pathologies from developing [AG-2024.06-694]. The intuition: classically "wrong" configurations become unstable once you include quantum effects.

Coupling constant anomalies constrain phase structure. When coupling constants themselves carry topological obstructions, they limit which gapped phases a theory can reach without phase transitions or symmetry breaking [AG-2024.07-797]—relevant if sign-changing couplings alter the allowed low-energy phases.

To properly answer what happens to stealth solutions specifically, you'd need preprints directly studying stealth mechanisms (acoustic cloaking, metamaterial designs, or cloaking via effective metric engineering) in these negative-coupling backgrounds—which aren't in this batch.