Ultra-light axion constraints from Planck and ACT: the role of nonlinear modelling

Lauren Gaughan, Anne M. Green, Adam Moss · 2026

Read Ultra-light axion constraints from Planck and ACT: the role of nonlinear modelling by Lauren Gaughan, Anne M. Green, Adam Moss (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-12054.txt.

Introduction

Lauren Gaughan, Anne Green, and Adam Moss study how ultra-light axion constraints from cosmic microwave background data depend on nonlinear modeling. For axion masses from 10^-25 to 10^-23 electronvolts, the Jeans scale sits in a quasi-linear zone probed by CMB lensing. That placement makes constraints highly sensitive to the nonlinear prescription. Using Planck 2018, ACT DR6, and DESI DR2 data, they find naive nonlinear modeling can falsely favor a subdominant axion dark matter component near 10^-24 electronvolts. This arises through a lensing-like power spectrum enhancement.

Catalog record

CREATORLauren Gaughan, Anne M. Green, Adam Moss
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54476
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.12054

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Ultra-light axion constraints from Planck and ACT: the role of nonlinear modelling Lauren Gaughan, Anne M. Green, Adam Moss Abstract We study how constraints on the abundance of ultralight axions (ULAs) from cosmic microwave background (CMB) data depend on their nonlinear modelling. We focus on the axion mass range 10 − 25 ≤ m / eV ≤ 10 − 23 10^{-25}\leq m/\rm{eV}\leq 10^{-23} , where the axion Jeans scale falls in the quasi-linear regime probed by CMB lensing, making constraints highly sensitive to the choice of nonlinear prescription. We show that the inferred constraints depend significantly on the choice of nonlinear model, which must therefore be treated carefully. Performing Markov Chain Monte Carlo (MCMC) analyses with Planck 2018, ACT DR6 and DESI DR2 BAO data, we find naive nonlinear modelling of non-cold matter can produce an artificial preference for a subdominant ULA dark matter component with mass m ≈ 10 − 24 m\approx 10^{-24}\, eV. This arises from a lensing-like enhancement of the CMB power spectrum. I Introduction The \Lambda CDM model has had great success in describing the evolution of the Universe 2 . Several questions, however, still persist, including the nature of dark matter and dark energy. While a wide range of candidates have been proposed, ultralight axions (ULAs) 21 have gained popularity in recent years. The QCD axion was originally proposed to solve the strong-CP problem in quantum chromodynamics. Lighter axions arise in extensions to the standard model, such as the string theory axiverse 3 ; 29 in which pseudo-Nambu-Goldstone bosons (or axion-like particles) are produced via symmetry breaking. The behaviour of ULAs depends on the relationship between the axion mass, m , and the Hubble parameter, H 12 . When H>m , the axion field is frozen and behaves like dark energy. As the Universe expands, H decreases and once m\sim H , the field begins to oscillate and subsequently behaves like cold dark matter (CDM). Crucially, the mass corresponding to the Hubble rate at matter-radiation equality, H_{\rm{eq}}\sim 10^{-28}\, eV, provides the distinction between lighter ULAs ( m\lesssim 10^{-28}\, eV), which behave like dark energy at late times, and heavier ULAs, which behave like dark matter. The full ULA mass range is 10^{-33}\lesssim m/\rm{eV}\lesssim 10^{-18} , bounded below by the current Hubble rate H_{0} (lighter axions are still frozen today) and above by the scale at which wave-like effects become astrophysically negligible. ULAs are a specific realisation of fuzzy dark matter, a class of dark matter candidates whose wave-like behaviour manifests on astrophysical scales. Due to their low mass, ULAs have large de Broglie wavelengths, \lambda=h/(mv) , which can extend to galactic or larger scales. On scales larger than the de Broglie wavelength, ULAs are almost indistinguishable from CDM. On smaller scales, quantum pressure opposes gravity, preventing collapse, and instead the density perturbation oscillates. If axions make up a non-negligible fraction of the dark matter, the matter power spectrum is suppressed 14 . Standard CDM, however, does cluster on these small scales, so mixed dark matter (MDM) cosmologies—with both fuzzy and cold components—require more careful treatment. See Refs. 21 ; 16 for more detailed reviews. Constraints on the abundance of ULAs are usually quoted in terms of the axion fraction, defined as the fraction of the total dark matter that is axions: f_{\rm{ax}}\equiv\frac{\Omega_{\rm{ax}}}{\Omega_{\rm{D}}}\,, (1) where \Omega_{\rm{ax}} and \Omega_{\rm{D}} are the axion and total dark matter density parameters respectively. Cosmic microwave background (CMB) data constrain ULAs because the suppression of small-scale structure growth by ULAs alters the angular power spectrum. Hlozek et al. 12 ; 13 used the standard effective fluid approximation (EFA) in axionCAMB and Planck 2015 data to place the constraint f_{\rm{ax}}\leq 0.05 at 95% confidence 1 1 1 All subsequent limits are also quoted at 95% confidence. in the mass range 10^{-32}\leq m/\rm{eV}\leq 10^{-25.5} . Rogers et al. 27 found stronger constraints at lower masses using Planck 2018, Atacama Cosmology Telescope (ACT) DR4 and South Pole Telescope (SPT-3G) CMB data: f_{\rm{ax}}<0.1 for m\lesssim 10^{-26}\,\rm{eV} , tightening to f_{\rm{ax}}<0.01 for 10^{-30}\leq m/\rm{eV}\leq 10^{-28} . ULAs with m\gtrsim 10^{-24}\,\rm{eV} are effectively degenerate with standard CDM on CMB length scales; thus the CMB alone does not constrain \Omega_{\rm{ax}}h^{2} , and large axion fractions remain allowed at these masses. The rapid oscillations of the ULA field become computationally prohibitive because small timesteps are required to accurately track the evolution. The standard EFA 14 ; 12 addresses this by averaging over the oscillations and treating the axion field as a fluid. In Ref. 24 we presented an implementation of the improved Passaglia and Hu (PH) EFA 25 in AxiCAMB (publicly available at https://github.com/adammoss/AxiCAMB ). The PH EFA factors out the rapid axion-field oscillations by introducing auxiliary fields, and calibrates the effective fluid equation of state and sound speed, providing a more reliable approximation of the axion behaviour. Using Planck PR4 and DESI DR1 baryon acoustic oscillation (BAO) data, we obtained improved constraints compared to the standard EFA implementation in Ref. 27 : f_{\rm{ax}}<0.044 for 10^{-26}\,\rm{eV} , tightening to f_{\rm{ax}}<0.0082 for 10^{-28}\,\rm{eV} . The recent ACT DR6 analysis 4 used the standard EFA in axionCAMB and also the axionEMU emulator, a machine-learning emulator trained on axionCAMB outputs, in the mass range 10^{-28}\leq m/\rm{eV}\leq 10^{-25} . Combining Planck 2018 and ACT CMB data, they found tighter constraints for m\geq 10^{-26}\,\rm{eV} than from Planck alone. However, they noted that for m\geq 10^{-25}\,\rm{eV} more detailed nonlinear modelling is required, and deferred this to future work. While the CMB alone is limited by the angular scales it probes, small-scale structure data can significantly tighten the constraints. Lyman- \alpha forest data rule out f_{\rm{ax}}=1 for m<2\times 10^{-20}\,\rm{eV} 28 , though ULAs in this mass range can still make up a subdominant component at the {\cal O}(10)\% level. Winch et al. 32 find f_{\rm{ax}}<0.28 from HST UV luminosity function (UVLF) data alone and f_{\rm{ax}}<0.15 from HST UVLF combined with Planck 2018 for m=10^{-24}\, eV. The reliability of small-scale structure constraints depends crucially on the nonlinear modelling. CMB experiments that measure gravitational lensing are also sensitive to nonlinear scales, and this sensitivity increases with the precision of the lensing reconstruction. In this work, we investigate the effect of nonlinear modelling for axions in the mass range 10^{-25}\leq m/\rm{eV}\leq 10^{-23} . For masses in this range, the axion Jeans cutoff lies in the quasi-linear regime, k\sim(0.1 – 1)\,h Mpc -1 , that contribute significantly to CMB lensing. The CMB-lensing kernel is broad, with sensitivity over z\sim 0.5 – 5 and a peak near z\simeq 2 . For lighter masses, the linear suppression of power is already large on CMB scales, so nonlinear corrections are a secondary effect. For heavier masses, the axion becomes nearly degenerate with CDM in CMB observables. This mass window is also well motivated phenomenologically. Rogers et al. 27 suggested that axions with a mass of 10^{-25}\, eV can reduce the S_{8} tension between CMB and galaxy clustering data. In addition, the formation of axion soliton cores may alleviate the cusp-core problem 15 . Finally, upcoming CMB lensing measurements from experiments such as Simons Observatory will probe the heavier end of this mass window, particularly on nonlinear scales. We integrate the axionHMcode 31 nonlinear model for ULAs into the Boltzmann solver AxiCAMB , including both

Read the complete text