⚛️ high-energy theory

Off-shell equivalence in quantum field theory and gravity

This paper establishes an operational criterion for off-shell equivalence in quantum field theory and gravity using the Vilkovisky–DeWitt effective action and scalar observables, demonstrating that apparent quantum inequivalences in theories like metric $f(R)$ gravity often stem from comparing distinct quantum objects rather than a failure of actual equivalence.

Original authors: Iberê Kuntz, Stefano Liberati

Published 2026-07-15

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Iberê Kuntz, Stefano Liberati

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ 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

Imagine you are trying to describe a complex video game level. You could describe it using the raw code (the "Lagrangian"), or you could describe it by changing the camera angle, the color palette, or even the coordinate system you use to map the terrain. In physics, these different descriptions are called "field redefinitions."

For a long time, physicists had a golden rule called the Equivalence Theorem. It said: "If you change the coordinates or the variables, the final score—the scattering of particles—stays exactly the same." It was like saying, "It doesn't matter if you measure the distance in miles or kilometers; the finish line is in the same spot."

But here's the catch: that rule only works when you look at the finish line (the "on-shell" state). It doesn't tell you if the journey looks the same if you change the map. In the messy, dynamic middle of the game—like in gravity, cosmology, or systems that aren't in equilibrium—physicists needed to know if two different maps described the exact same reality, not just the same finish line.

Enter Iberê Kuntz and Stefano Liberati, who have built a new, ultra-precise ruler to measure this.

The Problem: The "Naive" Map Trap

Imagine you have a map of a city. You decide to redraw it using a new grid system. If you just redraw the streets but forget to update the compass or the scale, your new map might look like a different city entirely, even though it's supposed to be the same place.

In quantum physics, when scientists tried to compare different versions of gravity (like metric $f(R)$ gravity and its "scalar-tensor" cousin), they often made this mistake. They would take two theories that looked like they were related by a simple change of variables, quantize them (turn them into quantum theories), and then compare the results. They found differences and concluded, "Aha! These theories are different!"

The authors argue that this is a trap. It's like comparing a map drawn in miles to a map drawn in kilometers, but then blaming the city for being different because the numbers don't match. The difference wasn't in the city; it was in how they measured it.

The Solution: The Vilkovisky–DeWitt "Universal Compass"

To fix this, the authors use a special tool called the Vilkovisky–DeWitt (VDW) effective action. Think of this as a "Universal Compass" that always points North, no matter how you rotate your map.

In standard physics, the "compass" (the mathematical action) gets messed up when you change coordinates. It depends on how you label things. The VDW construction fixes this by treating the field variables like points on a curved surface (a manifold). It ensures that the "compass" is a scalar—a single number that doesn't change just because you rotated your head.

Using this compass, the authors propose a new way to check for equivalence:

Pick a Probe: Decide what you are actually measuring. Are you looking at particle collisions (the finish line)? Or are you looking at how the system reacts to a push (response functions)?
Pair the Data: Combine the theory's data with your probe.
Compare: If the numbers match for your specific probe, the theories are equivalent for that experiment.

The Three Levels of Equivalence

The paper breaks down equivalence into three distinct flavors, using a clever analogy of exploring a landscape:

Local Equivalence: You are standing on a small patch of grass. You can walk around, and the map works perfectly. This is true for small changes in variables.
Branchwise Equivalence: Imagine a river that splits into three paths. You can follow one path and say, "This map works here!" But if you try to merge all three paths back into one single map, it breaks. The authors show that some field redefinitions only work on specific "branches" of the solution. If you try to force them to work everywhere, you get a contradiction.
Global Equivalence: This is the holy grail. It means the map works perfectly everywhere, from the start of the river to the ocean, without any splits or breaks. The authors show that this is much harder to achieve than people thought. Just because two theories look similar locally doesn't mean they are the same globally.

The Big Reveal: $f(R)$ Gravity and the "Parent" Theory

The authors use metric $f(R)$ gravity as their main test case. This is a theory where gravity depends on a function of the curvature of space, rather than just the curvature itself. To make it easier to study, physicists often introduce an "auxiliary" field (a helper variable) to turn the complex math into a simpler "scalar-tensor" form.

The paper reveals a crucial distinction:

Scenario A (The Right Way): You start with the complex gravity theory, introduce the helper variable with a constraint (a rule that says "you must equal the curvature"), and then do the quantum math. This is like building a house with a blueprint that includes a strict rule: "The roof must match the walls."
Scenario B (The Wrong Way): You take the simplified scalar-tensor theory and treat the helper variable as a completely independent, free-floating field. You do the quantum math without the rule. This is like building a house where the roof and walls are unrelated.

The authors prove that Scenario A and Scenario B are different quantum theories. They are not equivalent. If you compare the results of Scenario A (the constrained parent) with Scenario B (the free-floating theory), you will see differences. But this doesn't mean the original gravity theory is broken; it just means you compared two different quantum objects.

When you use the Vilkovisky–DeWitt compass and enforce the constraint correctly (Scenario A), the theory matches the original metric $f(R)$ gravity perfectly. The apparent "inequivalence" found in other papers was just a mismatch in how the quantum rules were applied.

What This Means for You

This paper doesn't claim to have solved all of gravity or discovered a new force. Instead, it provides a strict operational criterion for deciding when two quantum descriptions are actually the same.

It tells us:

Don't trust your eyes: Two theories can look different but be the same if you measure the right thing with the right tool.
Context matters: Equivalence isn't a yes/no switch for the whole universe; it depends on what you are measuring (scattering vs. response) and where you are looking (local patch vs. global map).
Beware of the "Free" Variable: If you add a helper variable to a theory, you must keep the rules that tie it to the original physics. If you let it run wild, you aren't studying the same universe anymore.

In short, Kuntz and Liberati have handed physicists a new, unbreakable ruler. It ensures that when we say two theories are "equivalent," we aren't just talking about the finish line, but the entire journey, measured with a compass that never lies.

Technical Summary: Off-shell Equivalence in Quantum Field Theory and Gravity

Problem Statement
Field redefinitions are ubiquitous in quantum field theory (QFT) and gravity, used to simplify actions, remove redundant operators in Effective Field Theories (EFT), and relate different formulations such as the Jordan and Einstein frames in scalar-tensor theories. The standard justification for these transformations relies on the equivalence theorem, which guarantees that local, invertible field redefinitions leave the S-matrix (scattering amplitudes) unchanged. However, this theorem is strictly an on-shell result. It fails to provide a general criterion for the equivalence of quantum theories when the relevant observables are off-shell, such as response functions, expectation values, in-in correlators, and dynamical quantities central to gravity, cosmology, and non-equilibrium physics.

Furthermore, in gauge theories and gravity, the standard effective action is not only sensitive to field parametrization off-shell but also gauge-dependent. This creates ambiguity when comparing different formulations (e.g., $f(R)$ gravity vs. its scalar-tensor representation) or frames, as naive comparisons often yield different results not because the theories are inequivalent, but because the comparison involves different quantum objects (e.g., comparing a constrained path integral to a free quantization of an enlarged space).

Methodology
The authors develop an operational framework for defining equivalence based on the Vilkovisky–DeWitt (VDW) effective action and a probe-based observable formalism.

Probe-Based Observables: The paper posits that equivalence is not absolute but relative to a class of observables. Observables are defined as scalar functionals obtained by pairing theory tensors (e.g., the VDW effective action and its covariant derivatives) with "probes" (e.g., detector trajectories, source profiles, asymptotic states).
- Separating vs. Physical Probes: A distinction is made between mathematical "separating" probes (which can uniquely identify theory tensors) and "physically admissible" probes (which correspond to actual experimental setups, such as scattering amplitudes or response functions).
- Tensorial Pairings: Observables are constructed as contractions of theory tensors with probe data, ensuring coordinate invariance.
The Vilkovisky–DeWitt Effective Action: To handle off-shell equivalence, the authors utilize the VDW effective action. Unlike the standard effective action, the VDW construction is a scalar on the configuration space manifold. It incorporates:
- A field-space metric ( $G_{AB}$ ) to define covariant derivatives.
- A geodesic source coupling ( $\sigma^A$ ) instead of linear source coupling, ensuring the source term transforms as a scalar.
- An invariant measure.
  This ensures the effective action transforms covariantly under field redefinitions, providing a consistent off-shell quantum object for comparison.
Classification of Equivalence: The framework distinguishes between:
- Direct Equivalence: A direct diffeomorphism between field spaces.
- Parent-Mediated Equivalence: Equivalence established via an enlarged configuration space (introducing auxiliary variables) followed by a reduction (e.g., via equations of motion or constraints).
- Local, Branchwise, and Global Equivalence: Equivalence may hold only on specific domains (patches) or branches of the field space map, particularly when maps are non-invertible or multi-valued (e.g., cubic field redefinitions or non-monotonic Legendre transforms).

Key Contributions and Results

Operational Criterion for Off-Shell Equivalence: The paper establishes that two quantum descriptions are equivalent if their renormalized VDW effective actions agree (up to finite parameter redefinitions) when transported by the field-space map, and if the field-space metrics and measures are consistently transported.
Resolution of Frame Equivalence Debates: The authors apply this framework to the Jordan-Einstein frame debate in $f(R)$ $f (R)$ gravity. They demonstrate that apparent quantum inequivalences often arise from comparing mismatched objects:
- Comparing the direct quantization of metric $f(R)$ with the unconstrained quantization of the auxiliary scalar-tensor action yields different theories.
- Correct equivalence is recovered only when the parent scalar-tensor theory is quantized as a constrained path integral (enforcing the auxiliary constraint via a delta functional or Lagrange multiplier) and the field-space geometry is consistently transported.
Distinction Between Redundancy and Inequivalence: The formalism clarifies that operators proportional to equations of motion are redundant for on-shell scattering (standard EFT) but can alter off-shell response functions if the field-space metric is not transformed. The VDW formalism ensures that physical response functions remain invariant under admissible field redefinitions.
Branchwise vs. Global Equivalence: Using a scalar EFT example with a cubic field redefinition, the paper shows that a field redefinition may define a valid local equivalence on specific branches of the field space but fail to define a single global equivalence if the map is not globally invertible. Similarly, in $f(R)$ gravity, the transition to the scalar-tensor form is only valid on branches where $f''(\chi) \neq 0$ .
Measure Compatibility: A critical result is that for parent-mediated equivalence to hold at the quantum level, the measure of the reduced theory must match the measure induced by the constrained parent path integral. This imposes specific conditions on the auxiliary components of the parent field-space metric (e.g., requiring an "endpoint-adapted" extension).

Significance
The paper argues that the question of whether two actions describe the same theory cannot be answered solely by comparing bare Lagrangians. The answer depends on:

The class of observables being tested (on-shell vs. off-shell).
The domain of the field-space map (local, branchwise, or global).
The quantum construction used (direct quantization vs. constrained parent quantization).

By utilizing the VDW effective action and a probe-based definition of observables, the authors provide a precise, geometric language to distinguish between genuine failures of equivalence and apparent discrepancies caused by comparing different quantum objects or neglecting the transformation of the field-space metric and measure. This framework is particularly significant for gravity and cosmology, where off-shell dynamics and frame choices are central, and for non-equilibrium QFT where response functions are the primary observables. It resolves long-standing ambiguities regarding the physical equivalence of $f(R)$ gravity and its scalar-tensor representations, showing they define the same quantum theory only when the auxiliary constraints are properly enforced in the path integral.