How does surrogate accuracy degrade when detector dimensions change versus when sensor spacing patterns are altered?

The preprints provided don't directly address how surrogate accuracy degrades when detector *dimensions* change versus when *sensor spacing patterns* are altered—that specific comparison isn't explored in these papers.

That said, I can point to relevant context: Two papers use neural surrogates to optimize detectors. Paper [AG-2025.11-530] trains a surrogate of a gravitational wave simulator and notes that "small changes in physical parameters can change the output by orders of magnitudes," suggesting surrogates must handle high sensitivity to geometric changes. Paper [AG-2025.03-1551] optimizes calorimeter layer thicknesses (a dimensional parameter) using mutual information surrogates but doesn't isolate how accuracy depends on *type* of geometric change (dimension vs. spacing pattern).

A related finding: In quantum sensing, Paper [AG-2026.02-125] shows that the overlap reduction function of detector networks is invariant under certain geometric transformations—a configuration with "small separation and high noise correlation" maps to one with "larger separation and reduced noise correlation"—suggesting that spacing geometry and its effects on information extraction are deeply linked, but again not through the lens of surrogate degradation.

For a direct answer to your question, you'd need papers explicitly benchmarking surrogate fidelity across different types of geometric perturbations.