If negative masses pair with positive masses in cosmological simulations, what prevents them from eventually separating and becoming isolated again?

Your question hinges on a real tension: if negative and positive masses repel each other gravitationally, what glues them together in cosmological models?

The answer appears to be instability-driven pairing, not stable equilibrium [AG-2026.01-618]. In the Bondi universe model, mixed positive-negative mass configurations are fundamentally unstable at all scales, with shorter wavelengths growing faster [AG-2026.01-618]. This instability doesn't lead to indefinite separation, though. Instead, the system evolves through three phases: a ballistic regime where particles move freely, then a random-walk phase with sporadic close encounters, and finally a uniformly accelerating phase driven by the formation of stable positive/negative mass pairs [AG-2026.01-618].

The key insight is that during close encounters—triggered by the very instability that would seem to tear them apart—positive and negative masses can become bound into tight pairs. This happens precisely when gravitational interactions become strong enough (the "coupling transition") for the paired state to dominate over the tendency to separate [AG-2026.01-618]. Once paired this tightly, the pair acts as a coherent accelerating structure rather than as isolated particles trying to flee each other.

Separately, it's worth noting that negative-mass stars in isolation are dynamically unstable [AG-2024.07-629], and bound systems with a positive mass and negative mass *can* form despite the mutual repulsion—the positive mass suffers a repulsive force but stays bound [AG-2026.02-280]. So pairing is possible in principle, though the mechanism ensuring it happens cosmologically is the growth of instabilities that paradoxically drives convergence.

The provided preprints don't fully explain why, once paired, these objects don't later re-separate due to perturbations.