The scheme independent 3-sphere free energy is not a monotone F-function

This paper demonstrates that while the scheme-independent three-sphere free energy decreases locally under relevant scalar deformations at conformal fixed points, it fails to be a globally monotone F-function along the full renormalization-group flow, as evidenced by its non-monotonic behavior in the massive scalar theory.

The Gist

Here is an explanation of the paper, translated into everyday language with some creative analogies.

The Big Question: Can We Measure "Time" in Physics?

Imagine you are watching a movie of a physical system evolving. In the world of Quantum Field Theory (the rules that govern tiny particles), there is a concept called the Renormalization Group (RG) flow. Think of this as the "movie" of the universe changing as you zoom in and out.

• High Energy (UV): You are zoomed in very close, seeing the tiniest, most chaotic details.
• Low Energy (IR): You are zoomed out, seeing the smooth, calm, large-scale behavior.

Physicists have long wanted a "thermometer" or a "counter" that proves this movie can only play forward, never backward. In 2D (flat space), we have a famous rule called the c-theorem that does this. In 3D (our space), there is a similar rule called the F-theorem. It says that a specific number, called F, must always get smaller as you go from the chaotic high-energy world to the calm low-energy world.

If you find a number that always goes down, you have proven that time (in the physics sense) has a direction.

The Problem: The "Noisy" Measurement

The authors of this paper looked at a very natural way to measure this number F. They looked at the "partition function" of a sphere (a fancy way of calculating the total energy of a quantum system living on a 3D ball).

However, there was a big problem. Imagine trying to weigh a delicate diamond on a scale that is covered in dust and has a wobbly leg.

• The "dust" represents local counterterms. These are mathematical artifacts that depend on how you choose to do your calculations (the "scheme").
• Because of this dust, the raw weight of the diamond (the sphere free energy) changes depending on how you clean the scale. It's not a reliable measurement.

The Attempted Fix: The "Double Filter"

The authors asked: Can we build a special filter to wipe off all that dust and get a clean, scheme-independent number?

They built a mathematical tool they call a "Double Filter."

• Think of the raw data as a messy signal with static noise.
• This filter is designed to subtract the noise perfectly.
• They proved that if you use this filter, the resulting number does go down when you make tiny, small changes to the system (perturbations). It looked like they had found the perfect monotone F-function!

The Twist: The "Overshoot"

Here is where the paper gets interesting. The authors decided to test their filter on a very simple, exact system: a free massive scalar field (think of it as a single, non-interacting particle moving on a sphere).

They expected the number to smoothly slide down from the high-energy value to the low-energy value, like a ball rolling down a hill.

It didn't.

Instead, the number behaved like a yo-yo or a dive-bombing plane:

1. It started high.
2. It went down (as expected).
3. It dipped below the final destination.
4. Then, it had to climb back up to reach the final destination.

Because it went down and then came back up, it is not monotone. It failed the test.

The "Why": The Order of the Filter

Why did this happen? The authors explain it with a structural reason involving the "order" of their math.

• The Entanglement Method (The Winner): Other physicists proved the F-theorem using "entanglement entropy" (how much two parts of a system are connected). Their math used a first-order filter (like a simple slope). Because of a deep mathematical rule called "Strong Subadditivity," this slope is guaranteed to always point down.
• The Sphere Method (The Loser): The sphere free energy has two types of "noise" (dust) to remove. To remove two things, you need a second-order filter (like a curve or a parabola).
  • The authors show that any filter complex enough to remove two types of noise must change direction. It's mathematically impossible for a second-order filter to be a straight line that only goes down. It has to curve, dip, and come back up.

The Takeaway

The paper concludes with a humbling lesson for physicists:

You cannot prove that time has a direction just by looking at the thermodynamics of a sphere.

Even though the sphere is a beautiful, natural object, its "free energy" is too messy to be a perfect clock. To prove the F-theorem, you need something "extra" that the sphere doesn't have—like the rules of entanglement (quantum connections) or spectral positivity.

In short:

• The Idea: Let's use a sphere to measure the flow of time in physics.
• The Plan: Clean the sphere's data with a mathematical filter.
• The Result: The data went down, then up, then down again. It wasn't a straight line.
• The Lesson: The universe is too complex for a simple thermometer. We need more sophisticated tools (like entanglement) to prove that physics is irreversible.

Technical Summary

Here is a detailed technical summary of the paper "The scheme independent 3-sphere free energy is not a monotone F-function" by Giacomo Santoni and Francesco Scardino.

1. Problem Statement

In $(2+1)$-dimensional Quantum Field Theory (QFT), the F-theorem states that the quantity $F \equiv -\log |Z(S^3)|$, evaluated at conformal fixed points, decreases along the Renormalization Group (RG) flow ($F_{UV} \geq F_{IR}$). This is the odd-dimensional analogue of the 2D $c$-theorem.

A long-standing hypothesis suggested that the sphere partition function itself, after removing scheme-dependent ambiguities, could serve as a monotone interpolating function (an "F-function") for the entire RG flow, similar to how entanglement entropy provides a monotone function in the proof of the F-theorem.

The Core Obstacle: The bare sphere free energy $W_{S^3}(R) = \log |Z(S^3_R)|$ is contaminated by local counterterms allowed by diffeomorphism invariance. On a sphere of radius $R$, these scale as $R^3$ and $R$. Because these terms shift the value and derivatives of $W_{S^3}$, the bare quantity is not scheme-independent, and no meaningful monotonicity statement can be made without removing them.

The authors investigate whether a scheme-independent quantity, constructed by systematically removing these local ambiguities, can serve as a monotone F-function.

2. Methodology

A. Construction of the Scheme-Independent Quantity

The authors define a "double filter" to eliminate the local counterterms ($R^3$ and $R$).

1. Counterterms: The local ambiguities are of the form $W_{loc} = \lambda_0 \int \sqrt{g} + \lambda_1 \int \sqrt{g} \mathcal{R}$, which scale as $R^3$ and $R$ respectively.
2. Differential Operator: They introduce the operator $D \equiv R \partial_R$.
3. The Filter: They construct a second-order differential operator $D^{(3)}_{ren}$ that annihilates $R^3$ and $R$ while preserving the constant term (the physical F-value):
   $$D^{(3)}_{ren} \equiv \left(1 - D\right)\left(1 - \frac{1}{3}D\right) = 1 - \frac{4}{3}D + \frac{1}{3}D^2$$
   This operator satisfies $D^{(3)}_{ren}[R^3] = 0$, $D^{(3)}_{ren}[R] = 0$, and $D^{(3)}_{ren}[1] = 1$.
4. Definition: The proposed thermodynamic F-function is defined as:
   $$F_E(R) \equiv -D^{(3)}_{ren} W_{S^3}(R)$$
   By construction, $F_E(R)$ is invariant under the addition of local counterterms and reduces to the standard $F$ at conformal fixed points.

B. Perturbative Analysis (Conformal Perturbation Theory)

The authors analyze the behavior of $F_E(R)$ near a UV fixed point under a relevant scalar deformation with coupling $g$ and dimension $\Delta < 3$.

• They calculate the leading $O(g^2)$ correction to the partition function, which involves an integrated two-point function scaling as $R^{6-2\Delta}$.
• They apply the filter $D^{(3)}_{ren}$ to this term. The filter introduces a polynomial factor $Q(\Delta) = \frac{4}{3}(\frac{3}{2}-\Delta)(\frac{5}{2}-\Delta)$.
• They demonstrate that while $Q(\Delta)$ changes sign in the window $3/2 < \Delta < 5/2$, the renormalized coefficient of the integral (involving the Gamma function$\Gamma(3/2-\Delta)$) also changes sign.
• Result: The product remains negative, implying $dF_E/d\log R < 0$ at leading order. Thus, $F_E$ appears monotonic near the UV fixed point.

C. Exact Non-Perturbative Computation

To test global monotonicity, the authors perform an exact calculation for a free massive scalar field on $S^3$.

• They compute the partition function $W$ using the eigenvalues of the Laplacian on $S^3$.
• They apply the filter $D^{(3)}_{ren}$ to derive the exact flow of $F_E(x)$ where $x = mR$ (mass $\times$ radius).
• They analyze the asymptotic behavior for large $x$ (IR limit).

3. Key Results

Failure of Monotonicity

The exact computation reveals that $F_E(R)$ is not monotone along the full RG flow.

• Behavior: Starting from the UV ($x \to 0$), $F_E$ decreases initially (consistent with perturbative analysis). However, it dips below the infrared value ($F_{IR} = 0$ for a massive scalar), reaching a minimum at $x^* \approx 1.58$.
• Recovery: After the minimum, $F_E$ increases back toward zero from below as $x \to \infty$.
• Derivative Sign: The flow derivative $dF_E/d\log R$ is negative for small $x$ but becomes positive for $x > x^*$.
• Asymptotics: At large $x$, the derivative behaves as $\frac{\pi}{96x} > 0$, confirming the non-monotonic rise.

Structural Obstruction

The authors identify the root cause of this failure:

• Order of the Filter: The entanglement entropy F-function ($F_{EE}$) uses a first-order filter because the entanglement entropy of a disk has only one UV divergence (area law). Its derivative depends on the second derivative of entropy, which is constrained by Strong Subadditivity (SSA) to have a fixed sign.
• Second-Order Requirement: The sphere free energy has two UV divergences ($R^3$ and $R$), necessitating a second-order differential filter.
• Lack of Positivity: The derivative of the filtered quantity involves higher-order derivatives ($D^2W$ and $D^3W$). There is no known positivity condition (like SSA) that simultaneously constrains these higher-order variations to maintain a fixed sign.
• Polynomial Sign Change: The filter polynomial $Q(\Delta)$ required to annihilate two power laws must necessarily change sign between its roots. While this is compensated perturbatively by the renormalization of coefficients, this compensation fails non-perturbatively (at $mR \sim O(1)$).

4. Significance and Conclusions

1. Negative Resolution: The paper definitively settles the question of whether sphere thermodynamics alone can encode RG irreversibility in odd dimensions. The answer is no. A scheme-independent quantity constructed solely from the sphere partition function is not a monotone F-function.
2. Necessity of Additional Structure: To construct a monotone F-function, one cannot rely solely on counterterm removal. Additional dynamical structure is required, such as:
   • Entanglement Entropy: Which utilizes Strong Subadditivity (SSA).
   • Spectral Positivity: As seen in the dilaton effective action for the $a$-theorem in $d=4$.
3. Practical Implications:
   • The quantity $F_E(R)$ remains a useful, UV-finite, scheme-independent diagnostic for identifying fixed points ($F_{UV} \geq F_{IR}$ holds).
   • It can be implemented on the lattice via finite differences without model-dependent subtractions.
   • However, researchers using sphere free energies to track degrees of freedom along RG flows (in lattice simulations, bootstrap, or holography) must be aware that monotonicity is not guaranteed, even in free theories.
4. General Lesson: The number of UV subtractions determines the order of the differential filter. Filters of order $\geq 2$ are generically sign-indefinite, preventing the construction of monotone interpolants from the partition function alone.

In summary, the paper demonstrates that while the sphere partition function captures the correct fixed-point values, the thermodynamic data alone is insufficient to prove the irreversibility of the RG flow in 3D without invoking entanglement or spectral properties.