Unitary and Analytic Renormalisation of Cosmological Correlators

This paper resolves discrepancies in the literature regarding ultraviolet divergences in cosmological correlators by demonstrating that various renormalization schemes yield consistent, unitary results and establishing a practical framework where the imaginary part of one-loop wavefunction coefficients is universally fixed by the logarithmic running of the real part.

The Gist

Imagine the universe as a giant, expanding balloon. A long time ago, this balloon was tiny and hot, and tiny quantum fluctuations (ripples) on its surface grew into the galaxies and stars we see today. Physicists try to predict the shape of these ripples using a mathematical tool called a "wavefunction." Think of the wavefunction as a recipe that tells us exactly how the universe should look.

For a long time, physicists only used the "tree-level" part of this recipe—the simplest, most obvious ingredients. It worked great for explaining what we see. But recently, they realized that to get the perfect recipe, they need to account for the "loops."

The Problem: Infinite Soup

In quantum physics, "loops" represent particles popping in and out of existence, interacting with each other. When you try to calculate these loops, you run into a mathematical disaster: infinity.

Imagine you are baking a cake, but every time you add a cup of flour, the recipe demands you add a cup of sugar, which demands a cup of eggs, which demands a cup of flour... and so on, forever. The bowl overflows. In physics, this overflow is called a "divergence." To fix it, you need a regulator—a mathematical sieve that catches the infinite overflow so you can work with a manageable amount.

The Conflict: Different Sieves, Different Cakes

The paper by Jain, Pajer, and Tong investigates three different "sieves" (regulators) that physicists have been using to catch these infinities in the early universe:

1. Dimensional Regularization (The "Shape-Shifter"): This method tries to change the number of dimensions of the universe (like turning a 3D cube into a 4D hypercube) just for the calculation, then changes it back. It's like trying to measure a 3D object by pretending it's 4D to make the math easier.
2. Mass-Dimensional Regularization (The "Tweaker"): This is a clever variation where, instead of just changing dimensions, they also tweak the "mass" of the particles so the math stays simple.
3. $\eta$ Regularization (The "Window"): A newer method that uses a smooth "window function" to gently fade out the infinite parts of the calculation, rather than changing the rules of the universe.

The Big Mystery:
When the authors used these different sieves, they got different results for the imaginary part of the recipe. In physics, the "real part" tells you the strength of the signal, but the "imaginary part" is like a secret code that tells you about parity violation (whether the universe has a "handedness," like a left hand vs. a right hand) and how the universe loses information (decoherence).

For a while, it looked like the answer depended entirely on which sieve you picked. If you used Sieve A, you got a "left-handed" universe. If you used Sieve B, you got a "right-handed" one. This was a crisis: Nature shouldn't depend on our choice of math tools.

The Solution: The "Cosmological Optical Theorem"

The authors realized that not all sieves are created equal. They introduced a strict set of rules for a sieve to be valid, based on two fundamental principles of physics:

1. Unitarity: Information cannot be destroyed. The universe is a closed system where probabilities must add up to 100%.
2. Analyticity: The math must be smooth and continuous, without sudden jumps.

They called the sieves that obey these rules "Unitary and Analytic Regulators."

Here is the magic trick they discovered:

• If you use a "bad" sieve (one that breaks the rules), you get nonsense results.
• If you use a "good" sieve (one that respects Unitarity and Analyticity), all three methods give the exact same answer.

The Universal Secret

The paper proves that no matter which "good" method you use, the imaginary part of the recipe is universally fixed. It is directly tied to the "logarithmic running" (how the strength of forces changes with scale).

The Analogy:
Imagine you are trying to measure the height of a mountain.

• Bad Method: You use a ruler that stretches when it gets hot. You get a different height depending on the weather.
• Good Method: You use a ruler that is perfectly rigid. No matter who measures it, or what tool they use, as long as the tool is rigid, they all get the exact same height.

The authors found that the "imaginary part" of the universe's wavefunction is like that rigid ruler. It is a fundamental, unchangeable feature of the universe, not an artifact of our math.

Why Does This Matter?

1. Solving the Disagreement: It settles a debate in the physics community. We now know that the "imaginary" signals (like parity-odd correlations) are real, predictable, and independent of the math we choose.
2. New Observables: This gives astronomers a specific target. They can now look for these specific "imaginary" signals in the Cosmic Microwave Background (the afterglow of the Big Bang) to test if our theory of the early universe is correct.
3. A New Tool: They championed the $\eta$ regulator as the easiest and most transparent way to do these calculations in the future, making it easier for other scientists to study the quantum history of the cosmos.

In short: The authors built a better filter for the universe's quantum soup. They proved that if you filter it correctly, the universe always tastes the same, revealing a hidden, universal "flavor" (the imaginary part) that was previously hidden by bad math.

Technical Summary

1. Problem Statement

The paper addresses the challenge of handling ultraviolet (UV) divergences in cosmological correlators and the associated wavefunction of the universe within the framework of Quantum Field Theory (QFT) in curved spacetime (specifically de Sitter space).

• Context: While tree-level (classical) calculations have successfully described cosmological observations, loop corrections are theoretically necessary for consistency and may offer unique observational signatures (e.g., parity-odd correlators).
• The Conflict: Existing literature presents conflicting results regarding the imaginary part of one-loop wavefunction coefficients. Different regularization schemes (variations of dimensional regularization) yield different imaginary parts, some of which violate unitarity (specifically the Cosmological Optical Theorem, COT) or scale invariance.
• The Core Question: Is the imaginary part of the renormalized wavefunction coefficient an arbitrary artifact of the regularization scheme, or is it a universal, physically determined quantity constrained by fundamental principles like unitarity and analyticity?

2. Methodology

The authors investigate a toy model: a shift-symmetric massless scalar field in exact de Sitter space with a cubic interaction ($\dot{\phi}^3$). They focus on the one-loop correction to the quadratic wavefunction coefficient ($\psi_2$).

The study employs a comparative analysis of four distinct regularization schemes:

1. Dimensional Regularization (Dim Reg): Following Senatore and Zaldarriaga, where both the spacetime dimension $d$ and the mode functions (Hankel functions) are analytically continued.
2. Mass-Dimensional Regularization (m&d Reg): Following Melville and Pajer, where the field mass is analytically continued alongside the dimension to keep mode functions simple (exponential/polynomial).
3. Partial Dimensional Regularization: Following Weinberg, where only the momentum integral is continued to $d$ dimensions while mode functions remain 3-dimensional.
4. $\eta$ Regularization: A new scheme adapted from flat-space work (Padilla and Smith), introducing a smooth window function $\eta(x)$ inside the loop integral to cut off UV divergences without changing spacetime dimensions.

Key Analytical Tools:

• Cosmological Optical Theorem (COT): Used as a consistency check. The COT relates the discontinuity of the wavefunction to the product of lower-point amplitudes, enforcing unitarity.
• Wick Rotation: Used to analyze the phase properties of propagators and integrands in the complex plane.
• Mellin Transforms: Used to analyze the divergence structure of $\eta$ regulators.

3. Key Contributions

A. Resolution of Dimensional Regularization Discrepancies

The authors demonstrate that different implementations of dimensional regularization (Dim Reg vs. m&d Reg) yield identical final renormalized results for the wavefunction coefficient $\hat{\psi}_2$, despite having different intermediate steps:

• Dim Reg: The loop contribution violates 3D scale invariance (producing a $\log k$ term), which is cancelled by a counterterm. The imaginary part arises entirely from the counterterm.
• m&d Reg: The loop contribution preserves 3D scale invariance but introduces time-dependent secular growth ($\log \eta_0$). The imaginary part arises from the loop itself, while the counterterm cancels the time dependence.
• Result: Both schemes converge to the same renormalized result, confirming that the physical prediction is regulator-independent if the scheme is consistent.

B. Development of Unitary and Analytic $\eta$ Regulators

The authors introduce a new class of regulators, Unitary and Analytic $\eta$ Regulators, defined by three conditions:

1. Normalization: $\eta(0) = 1$.
2. Convergence: $\lim_{|x|\to\infty} \eta(x) = 0$.
3. Hermitian Analyticity: $\eta(x) = [\eta(-x^*)]^*$.

They prove that generic $\eta$ regulators (e.g., real functions like $e^{-x}$) can lead to ambiguous or unphysical imaginary parts. However, imposing Hermitian analyticity ensures the regulator commutes with the discontinuity operation required by the COT.

C. Proof of Universality

The paper establishes a universality theorem: Under the constraints of unitarity (COT) and analyticity, the imaginary part of any one-loop renormalized wavefunction coefficient is universally fixed relative to the logarithmic running of its real part.

4. Key Results

The Universal Renormalized Result

For the quadratic wavefunction coefficient $\hat{\psi}_2$ at one-loop order, all consistent schemes (Dim Reg, m&d Reg, and Unitary $\eta$ Reg) yield the same result:

$$\hat{\psi}_2 = \psi_2^{\text{tree}} + \hat{\psi}_2^{\text{1L}} = \frac{1}{H^2}\left(\frac{ik^2}{\eta_0} - k^3\right) + \frac{1}{15}\frac{\lambda^2 H^2}{(4\pi)^2} k^3 \left( \frac{i\pi}{2} + \log\frac{\mu}{H} \right)$$

• Imaginary Part: The imaginary part is universally fixed to $+i\pi/2$ (relative to the coefficient of the logarithmic divergence).
• Real Part: Contains a logarithmic running term $\log(\mu/H)$, where $\mu$ is the renormalization scale.

The Universality Relation

The authors derive a general relation for any $n$-point wavefunction coefficient $\hat{\psi}_n$ at one loop:

$$\left( \mu \frac{\partial}{\partial \mu} - \frac{2}{\pi} \text{Im} \right) \hat{\psi}_n^{\text{1L}} = 0$$

This equation links the Renormalization Group (RG) running of the real part directly to the imaginary part. It implies that the emergence of an imaginary part is not arbitrary but is a necessary consequence of the RG flow required to absorb UV divergences while maintaining unitarity.

Parity-Odd Correlators

Since parity-odd correlators are purely imaginary in momentum space, this result provides a parameter-free prediction for the leading quantum contribution to parity-odd observables (like the parity-odd trispectrum), which vanish at tree level in exact de Sitter space.

5. Significance

1. Theoretical Consistency: The work resolves apparent contradictions in the literature regarding loop calculations in de Sitter space, establishing that unitarity and analyticity are sufficient to fix the imaginary part uniquely.
2. New Computational Framework: The introduction of $\eta$ regularization offers a technically simpler alternative to dimensional regularization. It avoids the complexity of special functions (Hankel/Bessel) in curved spacetime by keeping calculations in 3+1 dimensions, while still capturing the correct physics if the "unitary and analytic" conditions are met.
3. Connection to RG Flow: The paper uncovers a deep structural link between the spontaneous generation of imaginary parts in cosmological correlators and the renormalization group flow, analogous to the trace anomaly in flat-space QFT.
4. Observational Implications: By fixing the imaginary part, the paper provides robust, regulator-independent predictions for parity-odd signals in the Cosmic Microwave Background (CMB) and Large Scale Structure (LSS), offering a concrete target for future observational tests of quantum gravity effects in the early universe.
5. Parity Anomaly: The emergence of non-zero parity-odd correlators at one-loop (despite vanishing at tree level) is identified as a quantum parity anomaly, arising from the necessity of regulators that break classical symmetries to preserve unitarity.

In summary, the paper provides a rigorous, regulator-independent framework for computing quantum corrections in cosmology, proving that the imaginary part of the wavefunction is a physical observable fixed by the Cosmological Optical Theorem.