Cosmic voids — the vast underdense regions between galaxy filaments — are usually treated as the negative space of large-scale structure. Galaxies cluster into filaments and walls; voids are what's left over. The interesting physics lives in the dense regions where galaxies form, merge, and evolve. The voids are background.
Spencer London, Rogers, Laguë, Hložek, and Zaman (arXiv 2602.22990, February 2026) argue that voids are a better probe of dark matter's nature than the dense regions are. The logic is structural: voids are sensitive to the absence of small halos, and the absence of small halos is exactly what alternative dark matter models predict.
Ultra-light axions — hypothetical particles with masses around 10⁻²² eV, so light that their de Broglie wavelength spans kiloparsecs — suppress the formation of small dark matter halos below a mass threshold set by the axion mass. Fewer small halos means the small voids that would otherwise form between small halos merge into larger voids. The void size function — the distribution of void sizes — shifts toward larger voids in a universe with ultra-light axion dark matter compared to standard cold dark matter.
The effect is clean. Galaxy clustering statistics (power spectra, correlation functions) also change when small-scale structure is suppressed, but the signal is contaminated by galaxy bias, nonlinear evolution, and baryonic effects. Voids are simpler: their sizes are set primarily by the halo mass function at the scales where they form, and the mapping from halo suppression to void merging is relatively direct.
Simulations confirm the predictions. Forecasts for upcoming surveys — the Dark Energy Spectroscopic Instrument (DESI) and Euclid — show that void statistics from a Euclid-like survey could constrain the ultra-light axion energy density to less than 4.6% of total dark matter, roughly twice as strong as current limits. The constraint comes from counting emptiness, not from measuring density.
The void size function and the galaxy power spectrum provide complementary information because they respond to different aspects of structure suppression. The power spectrum measures the amplitude of density fluctuations at a given scale. The void size function measures the topology of underdense regions — not how much matter is missing, but how the missing matter is organized. Two dark matter models can produce similar power spectra but different void populations, because the voids depend on the threshold behavior of halo formation, not just its amplitude.
The emptiest regions of the universe carry the clearest signal of what fills them.