Tiny inhomogeneities, seeded by inflation in the first moments after the Big Bang, froze into the expanding universe and got imprinted in the cosmic microwave background. Among those imprints, non-Gaussian correlations are especially valuable because they directly depend on which quantum fields played an active role in that epoch. Usually a single field—the inflaton—dominates inflation, but in fundamental theories like the Standard Model and its extensions, extra degrees of freedom inevitably appear: candidates for dark matter particles or carriers of new interactions. Until now, their impact on observable non-Gaussianity was modeled approximately, assuming weak coupling to the inflaton. A new leap becomes possible thanks to exact solutions for the case of arbitrarily strong mixing.
The starting point was a simple but rich model: two scalar fields on an inflating background, one responsible for curvature ζ and the other for entropy perturbations σ. Their linear coupling is set by a dimensionless parameter λ and the entropy mass μ. Instead of treating this coupling as a small perturbation, the authors incorporated its exact solutions into the "dressed" propagators of the Schwinger–Keldysh formalism—in essence, a quantum-mechanical system respecting the uncertainty principle and the superposition principle. Thus, the transfer of perturbations between sectors is summed to infinite order. Then three canonical cubic interactions were added and the tree-level bispectra were computed. All calculations were reduced to elegant integral representations, which were then analyzed in the squeezed kinematic limit (one wavelength much larger than the other two).
The main result is closed-form expressions for the squeezed bispectrum, displaying a non-monotonic dependence of the form ($k_L$/$k_S$)^{1/2 ± ν}, where ν = √(9/4 – µ²/H²) for the light entropy mode, or logarithmic oscillations for the heavy one. In the weak-coupling limit, the classic results of "quasi-single-field" inflation and the cosmological collider program are recovered. However, when λ ≫ 1 and µ/H is not too large, both the power spectrum and the bispectrum experience exponential enhancement ∝ $e^{πλ}$. Notably, the frequency scale of oscillations is set directly by µ—a key difference from the perturbative approach, which features an effective mass m²_eff = µ² – H²λ². In practical terms, this means that strong mixing regimes can generate enormous inhomogeneities capable of collapsing into primordial black holes, and the signal shape in the CMB will be qualitatively different from earlier predictions.
This work provides, for the first time, analytical control over non-Gaussianity in the non-perturbative λ regime. It paves the way for a full scan of masses and coupling constants of particles that existed during inflation, using data from the cosmic microwave background and large-scale structure. Going beyond the weak-coupling approximation is especially crucial for theories where fields interact strongly—for instance, in models with curved field space or spontaneous supersymmetry breaking. Moreover, it becomes clearer how to disentangle the contributions of different particles without confusing them with artifacts of the perturbative expansion.
The machinery built around the conformal point ν = 1/2 and the sl(2) ladder algebra strongly suggests a connection with the cosmological bootstrap method, where correlators are constructed without dynamical calculations. The hypergeometric "dressing" operator acting on seed solutions appears as a natural tool to formulate bootstrap rules at strong coupling. Technically, the next steps are to include time-dependent mixing, compute four-point functions and loop corrections, where dressed propagators should considerably simplify the calculations.
The results are important for cosmology, particle physics, and quantum field theory because they offer a unified language to describe inflationary correlators at any coupling strength. They also set new benchmarks for observational tests of fundamental theories via non-Gaussianity signatures.
Direct comparison with Planck and future data (e.g., CMB-S4) for the whole parameter plane (λ, µ/H); develop software to accurately approximate the bispectrum in arbitrary configurations; extend to models with more than two fields.
The problem directly touches on the hierarchy problem: why could particle masses in the early Universe be of order the Hubble scale rather than the Planck scale? The method allows testing strongly coupled theories, which often appear in string landscapes or composite inflaton models. The enhancement of the power spectrum simultaneously addresses the puzzle of primordial black hole origin and may provide a clue to dark matter's nature if it consists of the same black holes.
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
This paper examines a model of two scalar fields on an inflationary background with arbitrarily strong mixing, parameterized by λ. Using exact 'dressed' propagators in the Schwinger–Keldysh formalism, we compute tree-level bispectra for three cubic interactions, obtain closed-form expressions in the squeezed limit, and find exponential amplification of the signal for λ ≫ 1.
Where its strength is
For the first time, we achieve analytical control over the nonperturbative-in-λ regime, where the spectrum and bispectrum grow exponentially, and the oscillation scale is set directly by the entropic mass μ, rather than the effective mass—a qualitative distinction between strong-coupling physics and perturbative predictions.
Where it could go next
Worth trying
Why this may matter
Accurate accounting for strong mixing could radically change the interpretation of future observational data: instead of constraints on weakly coupled particles, we gain a window into regimes where new fields are inextricably intertwined with the inflaton. This is critical for testing fundamental theories such as supergravity or string theory, where weak coupling is not guaranteed.
Nearby works in our archive
The archive is crowded around this work: 10 reviews nearby, closest at 0.71. This ground is well walked — and there is plenty to compare against.
as written by the authors arXiv:2607.14529
We develop an analytic treatment of primordial non-Gaussianity in multifield inflation that is nonperturbative in the constant curvature--isocurvature mixing strength $$\lambda$$. Using exact linear solutions for the coupled curvature perturbation $$\zeta$$ and isocurvature perturbation $$\sigma$$, we construct dressed propagators and derive exact integral representations for the tree-level bispectra generated by the interactions $$\dot\zeta^2\sigma$$, $$\dot\zeta\,\sigma^2$$, and $$\sigma^3$$. This formalism resums curvature--isocurvature transfer to all orders in $$\lambda$$ and applies for arbitrary values of the entropy mass $$\mu$$. We obtain closed-form expressions for the leading squeezed limit of the bispectrum contributions and recover the well-known cosmological-collider and quasi-single-field results in the weak-mixing limit. In the strong-mixing regime, where conventional transfer perturbation theory breaks down, we find that both the power spectrum and the bispectrum can be dramatically enhanced, giving rise to distinctive nonperturbative scaling laws for the reduced non-Gaussian amplitude. Together, these results open a new analytic window onto multifield inflation beyond the weak-coupling approximation and establish a framework for studying cosmological-collider signals, primordial-black-hole production, and loop corrections in strongly mixed inflationary dynamics.
To find solutions for fields of arbitrary mass, physicists used an elegant trick: they first solved the equations for a conformally coupled field (ν=1/2), then "multiplied" them using ladder operators of the sl(2) algebra, which acts like a quantum elevator between different masses. This same algebra appears in problems of supersymmetry and exactly solvable potentials in quantum mechanics.