2.1 × 10⁻⁹ — that's the temperature variation across the sky, about the same as measuring ripples a fraction of a millimeter high on the ocean surface. This microscopic non-uniformity is imprinted in the cosmic microwave background and tells the story of the very beginning of time. It all started a tiny fraction of a second after the big bang, when the universe inflated like a giant particle accelerator, boosting Quantum Field energies to levels unreachable by terrestrial colliders. This period of inflation was predicted by Alan Guth and Andrei Linde, and Juan Maldacena proposed using it as a cosmological collider — a tool to hunt for new particles.
The inflaton — the field that drove the expansion — didn't have to work alone. Massive scalar partner fields could have existed alongside it. Their quantum fluctuations, born in the rapid expansion of the universe under the influence of gravity, would have left traces in the form of non-Gaussianity — deviations in the statistics of primordial perturbations from a normal distribution. If such particle exchange took place at the dawn of time, its signal would show up in three-point correlations (the bispectrum) of hot and cold spots on the sky map. In essence, it's like looking for a fingerprint on glass: a tiny but distinctive pattern.
For the first time, the analysis was not limited to weak field interactions. The scientists used the full dynamics of the extended Standard Model sector, accounting for up to seven independent interaction channels. The key parameter — mixing ρ — changes the effective mass of the particle, making it visible or hidden. Bayesian inference compared predictions with the real Planck satellite data. The maximum improvement in χ² was 5.3 with five extra parameters — less than a random fluctuation. Simply put, there is no signal. Constraints on mixing reached ρ/H ≲ 170.
This result deals a blow to a whole class of multi-field inflation models popular among theorists. It shows that the common weak coupling approximation is inadequate: all allowed parameter values contribute too weakly to be noticed by Planck. You could say the cosmic collider turned up empty for this range of particles — but that's not a failure, it's a valuable clue. If inflaton partners exist, they must be either very light or interact differently.
The next generation of experiments — CMB-S4, Simons Observatory — will improve sensitivity by only 2–5 times. That's unlikely to be enough to hear the faint whisper of exchange particles. Much greater hopes are pinned on maps of the 21-cm line from the cosmic dark ages, which carry information about millions of independent modes. For now, those very same 2.1 × 10⁻⁹ written across the sky remain silent about the first trillionths of a second. The cosmic collider awaits new listeners — and new instruments capable of discerning its subtlest notes.
Calculating the bispectra took more than 4 million runs of the CosmoFlow program — as if a single computer had been running non-stop for 11 years. Yet each individual calculation after interpolation took milliseconds.