Could axion string filaments remain unresolved even with adaptive refinement if the mesh can't shrink enough?

# On Mesh Resolution and Axion String Filaments

Yes, this is a real computational challenge. When adaptive mesh refinement (AMR) cannot shrink finer than some minimum cell size, thin string filaments can indeed slip through unresolved—meaning the simulation misses them entirely or smears their physics across several grid cells.

What the simulations show

The most directly relevant work here is the axion mass prediction study [AG-2024.12-838], which used AMR specifically to avoid the expense of a static lattice with 262,144³ sites. The key insight: AMR allowed them to achieve sub-percent precision on the spectral index of axion radiation from strings without hitting the memory wall of a uniformly fine grid.

However, AMR has limits. In the ultra-high-resolution fixed-grid study [AG-2024.10-1401], researchers pushed to 16,384 points per side to measure axion string network scaling. They found the network evolves toward a stable density (~1.22) and velocity (~0.57). This long fixed-grid campaign was necessary precisely *because* you need sustained fine resolution across the entire domain—strings can hide or behave anomalously if local refinement isn't aggressive enough.

The practical problem

A mesh cell that's too coarse relative to the string core width will fail to resolve:

• The detailed dynamics of string reconnection (where two strings cross and exchange partners)
• Subtle kink propagation along filaments
• The small-scale radiation spectrum that feeds into dark matter abundance predictions

The [AG-2024.12-838] team mitigated this by making AMR *follow* string cores dynamically, refining wherever the field gradient was steep. But if your refinement budget runs out, or if filaments fragment into genuinely sub-grid structures, you're stuck.

Why it matters

Axion mass predictions depend on measuring how much energy strings radiate away as axion particles. If thin filaments remain unresolved, you'll either overestimate their persistence (they look more robust than they are) or underestimate radiation loss, throwing off the dark matter relic abundance calculation that experiments rely on [AG-2024.12-838].

The archive doesn't directly address a pathological minimum-cell-size limit, so I can't cite a concrete study of what happens when AMR refinement saturates.