⚛️ high-energy theory

Notes on LSZ, i epsilon Prescriptions and Perturbation Theory, in QFT and Cosmology

This paper clarifies the LSZ reduction formula and the origin of $i\epsilon$ prescriptions in flat space quantum field theory without assuming a free vacuum at asymptotic times, and extends these arguments to inflationary cosmology to demonstrate that unitarity-violating contour deformations are unnecessary.

Original authors: S. P. de Alwis

Published 2026-02-25

📖 5 min read🧠 Deep dive

CC0 1.0

Original authors: S. P. de Alwis

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: What is this paper about?

Imagine you are trying to predict the outcome of a complex game of billiards. In physics, this "game" is the collision of subatomic particles. Physicists have a standard rulebook (called Perturbation Theory) to calculate these outcomes. However, the rulebook has a few "cheat codes" or "loopholes" that people have been using for decades to make the math work.

This paper, written by S. P. de Alwis, argues that we don't need those cheat codes.

The author shows that we can get the correct answers using the "real" rules of the game, without having to pretend that the game stops or that time behaves strangely. He fixes two specific "cheat codes" that physicists usually rely on:

Pretending the interaction between particles magically turns off in the distant past or future.
Pretending time flows into a "parallel universe" (imaginary numbers) to make the math converge.

Part 1: The LSZ Theorem (The "Ghost" of the Past)

The Old Way (The Cheat Code):
To calculate how particles scatter, textbooks often say: "Let's assume that a long time ago, the particles were just floating around doing nothing, and they only started interacting right before the collision."

The Problem: This is physically impossible. If particles have mass and interact, they are always interacting, even when they are far apart. It's like saying two magnets only feel each other's pull when they are about to crash, but not when they are miles apart. This breaks the laws of physics (specifically, time symmetry).

The New Way (The Real Deal):
The author uses a mathematical tool called the Riemann-Lebesgue Theorem.

The Analogy: Imagine you are listening to a chaotic crowd of people shouting (the interaction). You are trying to hear a specific, steady tone (the particle).
The math shows that if you wait long enough, the chaotic shouting naturally fades into the background noise. You don't need to silence the crowd (turn off the interaction); you just need to wait until the random noise averages out to zero.
The Result: The particles behave as if they are free, not because we forced them to be free, but because the "noise" of their interactions naturally washes out over time. We don't need to break the laws of physics to make the math work.

Part 2: The $i\epsilon$ Prescription (The "Imaginary" Shortcut)

The Old Way (The Cheat Code):
When calculating these interactions, physicists often use a trick called the $i\epsilon$ prescription.

The Analogy: Imagine you are walking on a tightrope (the real timeline). To keep from falling, you imagine a safety net that exists in a "parallel dimension" (imaginary time). You pretend you can step into this dimension to stabilize your walk, calculate your path, and then step back.
The Problem: Stepping into an imaginary dimension violates Unitarity. In physics, Unitarity means "probability is conserved." If you step into a parallel dimension, you might lose track of where the particles went. It's like calculating a bank account balance by temporarily moving money to a fake bank account that doesn't exist. It works for the math, but it feels "illegal."

The New Way (The Real Deal):
The author shows that you don't need to step into the parallel dimension.

The Analogy: Instead of stepping off the tightrope, you just realize that the tightrope itself has a tiny bit of "friction" or "damping" built into it naturally.
By looking at the wave function (the mathematical description of the vacuum state) of the actual interacting universe, the math naturally produces the same "stabilizing effect" that the imaginary dimension was supposed to provide.
The Result: The "safety net" was never needed. The universe naturally dampens the wild fluctuations, allowing us to calculate the results while staying strictly on the real timeline.

Part 3: Cosmology (The Universe as a Billiard Table)

The Context:
In the early universe (Cosmology), things are different. The universe is expanding, and there is no "far future" where particles stop interacting. The old "cheat codes" are even more dangerous here because they violate the rules of how the universe expands.

The Solution:
The author applies his new method to the early universe.

The Analogy: Imagine trying to predict the weather. The old method said, "Let's pretend the atmosphere stops moving so we can calculate the storm." This is bad because the atmosphere is always moving.
The new method says: "Let's look at the actual moving atmosphere. Even though it's chaotic, if we look at the statistical average of the waves, the math naturally settles down."
The Result: We can calculate how the early universe evolved and how structures (like galaxies) formed without having to pretend the universe froze or stepped into an imaginary time.

Summary: Why does this matter?

For decades, physicists have been using "mathematical fudge factors" to make their equations work. They pretended interactions turned off or stepped into imaginary time.

This paper says: "Stop fudging it."

By using a more rigorous mathematical approach, the author proves that the universe is consistent. The interactions don't need to magically disappear, and time doesn't need to become imaginary. The math works perfectly fine if we just trust the full, messy, interacting reality of the universe.

In a nutshell: We don't need to break the rules of the game to win; we just need to understand the game better.

1. Problem Statement

The paper addresses foundational ambiguities and unphysical assumptions often found in standard textbook derivations of the LSZ (Lehmann-Symanzik-Zimmermann) reduction formula and the $i\epsilon$ prescription in perturbation theory. Specifically, it targets two common justifications used to derive scattering amplitudes and cosmological correlation functions:

Adiabatic Switching: The assumption that interactions are "adiabatically turned off" at asymptotic times ( $t \to \pm \infty$ ) to define free in/out states. The author argues this violates time-translation invariance and is physically unnecessary.
Contour Deformation: The practice of rotating the time integration contour into the complex plane (imaginary time) to select the vacuum state and derive the $i\epsilon$ prescription. The author argues this violates unitarity.

These issues are particularly relevant in inflationary cosmology, where the background is time-dependent, making the "turning off" of interactions impossible, and where unitarity violations are unacceptable.

2. Methodology

The author employs a rigorous functional integral approach, generalizing arguments originally made by Weinberg, to derive perturbation theory without relying on the aforementioned unphysical assumptions.

Re-evaluation of LSZ: The paper re-examines the LSZ asymptotic condition. Instead of assuming the field operator $\hat{\phi}(x)$ converges strongly to a free field operator, it treats the condition as a weak limit (limits of matrix elements between normalizable states). The author utilizes the Riemann-Lebesgue (RL) theorem to demonstrate that the interaction-dependent parts of the field operators vanish in the asymptotic limit for physical states, thereby justifying the LSZ reduction formula without assuming the interaction turns off.
Functional Integral Derivation:
- The author constructs the generating functional for vacuum-to-vacuum correlation functions by integrating the transition amplitude between field eigenstates weighted by the full interacting vacuum wave functions ( $\Omega[\phi]$ ), rather than assuming the vacuum is the free vacuum.
- The vacuum wave function is split into a free quadratic part ( $\omega_0$ ) and an interacting higher-order part ( $\omega_I$ ).
- By introducing a convergence factor $e^{-\epsilon|t|}$ , the author shows that the free part of the wave function naturally generates the standard Feynman propagator with the $i\epsilon$ prescription.
- Crucially, the interacting part of the wave function ( $\omega_I$ ) is shown to vanish in the $\epsilon \to 0$ limit, meaning the full interacting vacuum behaves effectively like the free vacuum for the purpose of generating perturbation theory, but without requiring the interaction to physically switch off.
Cosmological Generalization (In-In Formalism):
- The methodology is extended to the Closed Time Path (CTP) or "in-in" formalism used in cosmology to calculate equal-time correlation functions.
- The derivation uses a density matrix $\rho_0$ and a contour $C$ that runs from $-\infty$ to a late time $t$ and back to $-\infty$ .
- The author demonstrates that the same logic applies: the vacuum density matrix (constructed from interacting wave functions) provides the necessary $i\epsilon$ prescription for the propagators on the CTP contour, without needing to rotate the time contour into the imaginary axis.

3. Key Contributions

Rigorous Justification of LSZ without Adiabatic Switching: The paper proves that the LSZ reduction formula holds based solely on Poincaré invariance, micro-causality, and the Riemann-Lebesgue theorem. It clarifies that the asymptotic condition is a property of matrix elements (weak limits), not operator limits, removing the need for the unphysical assumption that interactions vanish at infinity.
Unitarity-Preserving $i\epsilon$ Derivation: It provides a derivation of the $i\epsilon$ prescription that relies on the structure of the interacting vacuum wave function rather than complex time contour rotations. This preserves unitarity, which is often compromised by imaginary time rotations in standard textbook treatments.
Resolution of Cosmological Perturbation Theory Issues: The paper establishes a solid theoretical foundation for calculating cosmological correlators (inflationary perturbations) using the interacting vacuum. It dispels the necessity of the "Bunch-Davies vacuum" justification via imaginary time rotation, showing that the standard $i\epsilon$ prescription arises naturally from the interacting vacuum state in a time-dependent background.

4. Results

Equation (9) & (10): The author explicitly derives the asymptotic creation/annihilation operators using the RL theorem, showing that the interaction terms vanish for normalizable states as $t \to \pm \infty$ , leaving only the free-field-like operators.
Equation (18) & (19): In the flat space functional integral, the inclusion of the full vacuum wave function leads to the standard generating functional with the Feynman propagator $K^{-1}_\epsilon$ containing the $i\epsilon$ term. The interaction part of the wave function ( $\omega_I$ ) is proven to be irrelevant in the $\epsilon \to 0$ limit.
Equation (26) - (31): In the cosmological (in-in) context, the author derives the generating functional for connected correlation functions. The resulting free propagators ( $G_{++}, G_{--}, G_{+-}, G_{-+}$ ) naturally incorporate the $i\epsilon$ prescription derived from the interacting vacuum, valid for time-dependent backgrounds (like de Sitter space).

5. Significance

This paper is significant for both theoretical clarity and practical application in high-energy physics and cosmology:

Theoretical Consistency: It removes logical inconsistencies in standard QFT pedagogy where "turning off interactions" is used as a mathematical trick despite violating fundamental symmetries (time translation).
Cosmological Robustness: In inflationary cosmology, where the background metric evolves and interactions cannot be adiabatically switched off, this derivation validates the standard perturbative calculations used to predict the Cosmic Microwave Background (CMB) power spectrum and non-Gaussianities. It confirms that these calculations are valid even when the vacuum is a fully interacting state.
Unitarity: By avoiding complex time contour deformations, the paper ensures that the derived perturbation theory strictly respects unitarity, a critical requirement for any physical theory.

In summary, de Alwis demonstrates that the standard tools of perturbation theory (LSZ, $i\epsilon$ , Feynman rules) are robust consequences of the interacting vacuum structure and do not require the unphysical assumptions of adiabatic switching or imaginary time rotations.