This paper introduces a new family of simple inflation models that can move continuously between two well‑known types of “attractor” behavior. By changing a single parameter called μ, the models can behave like exponential α‑attractors on one end and like polynomial attractors on the other. That lets the models produce a wide range of values for the spectral index ns, the number that describes how the initial density ripples in the universe vary with scale.

The authors build these models by taking polynomial potentials and embedding them in the α‑attractor framework. The key parameters are α, which controls the shape of the inflationary plateau, k, which sets the power in the polynomial piece, and μ, the interpolation parameter. In the large‑μ limit the models reproduce the standard exponential α‑attractor predictions: ns ≈ 1 − 2/N and a tensor‑to‑scalar ratio r that scales like 12α/N2 (N is the number of e‑foldings, typically 50–60). In the small‑μ limit they reproduce the polynomial attractor results, for example ns = 1 − 2(k+1)/((k+2)N). The paper gives concrete examples: for α = 1, k = 2 and N = 55, values of μ greater than about 3 give ns ≈ 0.9633, while μ below about 0.3 gives ns ≈ 0.9723.

This interpolation matters because recent measurements of the spectral index disagree at a subtle level. The cosmic microwave background (CMB) alone prefers ns ≈ 0.9682, while a joint CMB plus DESI (Dark Energy Spectroscopic Instrument) analysis shifts the best value upward to ns ≈ 0.9728. The new family can cover the range of ns between the exponential and polynomial attractors. In numbers, the models can describe ns between about 0.9636 and 0.9818 for N = 55, which is wide enough to include the CMB and CMB+DESI central values.

There are important caveats. The higher ns from DESI is still under debate. The paper notes that DESI’s result is in some tension with CMB data if one assumes the standard ΛCDM cosmological model, and that combining datasets like ACT (Atacama Cosmology Telescope) and SPT‑3G (South Pole Telescope 3G) increases that tension. The model predictions also depend on where the last ≈60 e‑foldings of inflation occur in field space: if they happen at small field values the polynomial regime applies, and if they happen at large field values the exponential α‑attractor regime applies. For the polynomial behavior to hold one needs μ below model‑dependent thresholds (for example μ ≪ 0.33 for the α = 1, k = 2, N = 55 case quoted in the paper).