Scale-Dependent Suppression Functions and Functional Space Geometry in Renormalization

Daniel Ketels · 2025

Read Scale-Dependent Suppression Functions and Functional Space Geometry in Renormalization by Daniel Ketels (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-13196.txt.

Introduction

Ketels studies what a smooth, scale-dependent cutoff does to the geometry of the space of functions used in renormalization, in a 2025 paper. Instead of cutting off high momenta abruptly, the suppression function fades them out under the control of a moving scale. He shows this choice bends the metric of the function space itself.

The induced curvature grows more negative toward high energies, which dynamically suppresses those states and deforms the domain in a controlled way. He also derives how the suppression function flows with scale and links it to the curvature flow. The spectral changes hint at an effective drop in dimension at high energies.

Catalog record

CREATORDaniel Ketels
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-13196

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Scale-Dependent Suppression Functions and Functional Space Geometry in Renormalization Daniel Ketels Abstract We analyze the effects of a scale-dependent suppression function Ω ⁡ ( k , Λ ) \Omega(k,\Lambda) on the functional space geometry in renormalization theory. By introducing a dynamical cutoff scale Λ \Lambda , the suppression function smoothly regulates high-momentum contributions without requiring a hard cutoff. We show that Ω ⁡ ( k , Λ ) \Omega(k,\Lambda) induces a modified metric on functional space, leading to a non-trivial Ricci curvature that becomes increasingly negative in the ultraviolet (UV) limit. This effect dynamically suppresses high-energy states, yielding a controlled deformation of the functional domain. Furthermore, we derive the renormalization group (RG) flow of Ω ⁡ ( k , Λ ) \Omega(k,\Lambda) and demonstrate its role in controlling the curvature flow of the functional space. The suppression function leads to spectral modifications that suggest an effective dimensional reduction at high energies, a feature relevant to functional space deformations and integral convergence in renormalization theory. Our findings provide a mathematical framework for studying regularization techniques and their role in the UV behavior of function spaces. 1 Introduction By an optimized suppression function we mean a family of smooth maps \Omega(\cdot,\Lambda)\,\colon\,D\,\to\,(0,1],\quad\text{with }\,\,\Omega(k,\Lambda)\,:=\,\frac{1}{\,1\,+\,\bigl(\tfrac{k^{2}}{\Lambda^{2}}\bigr)^{\beta}\,}\,-\,\frac{\,\eta\,\exp\!\bigl[-\,k^{2}/\Lambda^{2}\bigr]}{\epsilon\!\bigl(k^{2}\bigr)}, where: • D=\mathbb{R}^{1,3} (or more generally \mathbb{R}^{d} ) is the momentum‐space domain, • \Lambda>0 is a dynamical cutoff scale, • \beta,\,\eta>0 control power‐law vs. exponential suppression, • \epsilon\in C^{\infty}\bigl(D,\,(0,1]\bigr) ensures a smooth, Lorentz‐invariant transition without hard cutoffs. Hence \Omega(k,\Lambda)\in(0,1] , is smooth, and suppresses high‐momentum modes without imposing a sharp momentum cutoff. Definition of \beta (Using the Gamma Function) In d dimensions, a factor k^{-2\beta} converges at large \|k\| if 2\beta>\tfrac{d}{2} . Concretely, \int_{\Lambda}^{\infty}k^{\tfrac{d}{2}-1}\,k^{-2\beta}\,dk\,\text{converges for}\,\tfrac{d}{2}-1-2\beta<-1\,\Longleftrightarrow\,2\beta>\tfrac{d}{2}. Equivalently, inserting a factor e^{-\Lambda\,k} and evaluating via \Gamma -function yields the same condition. Thus: \beta\,:=\,\inf\Bigl\{\gamma>0\,\Bigm|\,\Gamma\!\bigl(\tfrac{d}{2}-2\gamma\bigr)\text{ finite}\Bigr\},\quad\text{ensuring }2\beta>\tfrac{d}{2}\text{ for UV damping}. Definition of \eta (Additional Exponential Damping) If \eta>0 , then \int_{0}^{\infty}k^{\tfrac{d}{2}-1}\,e^{-\eta\,k^{2}}\,dk\,=\,\tfrac{1}{2}\,\eta^{-\tfrac{d}{2}}\,\Gamma\!\bigl(\tfrac{d}{2}\bigr)\,<\,\infty, so exponential decay can overcome strong divergences. We pick \eta>0 as a positive parameter for extra damping strength in \Omega . Definition of \epsilon(k^{2}) The function \epsilon:\,D\to(0,1] manages the IR–UV transition smoothly. A canonical choice: \epsilon\!\bigl(k^{2}\bigr)\,=\,\frac{1}{1+\bigl(\tfrac{k^{2}}{k_{c}^{2}}\bigr)^{\alpha}},\quad\alpha>0, so that \epsilon(k^{2})\approx 1 for k^{2}\ll k_{c}^{2} and \approx 0 for k^{2}\gg k_{c}^{2} . No sharp cutoff arises, ensuring Lorentz invariance and continuity across momentum scales. Proposition 1.1 (Positivity and Boundedness of \Omega ) . Assume there exist constants \eta>0 and a smooth function \epsilon:\,\mathbb{R}_{\geq 0}\to(0,1] such that for all \Lambda>0 and all k\in\mathbb{R}^{d} , \eta\,e^{-\,\tfrac{k^{2}}{\Lambda^{2}}}\,\leq\,\frac{\epsilon\!\bigl(k^{2}\bigr)}{1+\bigl(\tfrac{k^{2}}{\Lambda^{2}}\bigr)^{\beta}}\quad. (1) Then the scale-dependent suppression function \Omega(k,\Lambda)\,=\,\frac{1}{\,1+\bigl(\tfrac{k^{2}}{\Lambda^{2}}\bigr)^{\beta}}\,-\,\frac{\,\eta\,e^{-\,\tfrac{k^{2}}{\Lambda^{2}}}}{\epsilon\!\bigl(k^{2}\bigr)} satisfies 0\,\leq\,\Omega(k,\Lambda)\,\leq\,1\quad\text{for all }k\in\mathbb{R}^{d},\,\Lambda>0. In particular, \Omega(\cdot,\Lambda) remains in the interval (0,1] and never becomes negative. Proof. By definition \Omega(k,\Lambda)\in(0,1] , and by using the definition: \Omega(k,\Lambda)\,=\,\frac{1}{\,1+(k^{2}/\Lambda^{2})^{\beta}}\,-\,\frac{\,\eta\,e^{-\,k^{2}/\Lambda^{2}}}{\epsilon(k^{2})}. The condition rearranges to 0<\frac{\eta\,e^{-\,k^{2}/\Lambda^{2}}}{\epsilon(k^{2})}\,\leq\,\frac{1}{\,1+(k^{2}/\Lambda^{2})^{\beta}}. Which follows dirrectly from \Omega(k,\Lambda)>0 and both terms are strictly decreasing for increasing \Lambda>0 . For \|k\|\ll\Lambda , we get \tfrac{1}{\,1+(k^{2}/\Lambda^{2})^{\beta}}<1 and \eta\,e^{-\tfrac{k^{2}}{\Lambda^{2}}}<\eta , so as \epsilon(\cdot)\leq 1 , the expression does not exceed 1 (or can be bounded by a constant factor). Hence 0<\Omega(k,\Lambda)\leq 1 . ∎ Remark 1.2 (Analytic Properties of \Omega ) . By design, \Omega(k,\Lambda) is: • strictly decreasing in \|k\| for each \Lambda , preserving IR modes ( \Omega\approx 1 at small \|k\| ) and suppressing UV modes, • bounded in (0,1] , avoiding both negativity and blow‐ups, • smoothly differentiable both k and \Lambda , introducing no discontinuities. 2 Momentum-Space Integral Convergence Before turning to functional integrals, let us see how \Omega regulates typical momentum‐space integrals in quantum field theory (QFT). A general loop integral looks like I\,=\,\int_{\mathbb{R}^{d}}f(k)\,d^{d}k, which might diverge for large \|k\| . Replacing d^{d}k by \Omega(k,\Lambda)\,d^{d}k yields I_{\Omega}\,=\,\int_{\mathbb{R}^{d}}\Omega(k,\Lambda)\,f(k)\,d^{d}k, and we want to show I_{\Omega} is finite under certain conditions on \beta , \eta , etc. Proof of Convergence in Different Regimes Theorem 2.1 (Convergence of Suppressed Integrals) . Let \Omega(k,\Lambda) be the function in defined above and suppose f(k)\sim k^{\alpha} as \|k\|\to\infty . Then: • If 2\beta>d+\alpha, the integral \int\Omega\,f(k)\,d^{d}k converges by power‐law suppression. • If \eta>0 and \epsilon does not vanish too fast, exponential damping ensures convergence regardless of \alpha . Proof. By definition, for large k , we have \Omega(k,\Lambda)~\approx~k^{-2\beta} thus \Omega\,f(k)\approx k^{-2\beta}\,k^{\alpha}=k^{\alpha-2\beta} . Spherical coordinates show that k^{\alpha+d-1-2\beta} is integrable at \infty if 2\beta>d+\alpha . Additional exponential factor: if \eta>0 in \tfrac{\eta\,e^{-k^{2}/\Lambda^{2}}}{\epsilon(k^{2})} , the decay beats any polynomial growth. Hence the integral converges for all \alpha . ∎ Implications for Renormalization. In QFT, loop integrals typically behave like \int k^{d-1-2n}\dots , requiring 2\beta>d-2n or exponential damping. Meanwhile IR physics is preserved since \Omega\approx 1 at small k . Thus divergences are removed without imposing a hard cutoff. 2.1 Functional Integrability Extending to functional integrals over fields \phi(x) , define \mathcal{D}_{\Omega}\,\phi\,:=\,\lim_{\Lambda\to\infty}e^{-S_{\Omega}[\phi,\Lambda]}\,\mathcal{D}\phi, where S_{\Omega}[\phi,\Lambda]\,=\,\tfrac{1}{2}\int_{\|k\|\leq\Lambda}\bigl[1-\Omega(k,\Lambda)\bigr]~|\tilde{\phi}(k)|^{2}\,d^{d}k. If \Omega\approx 0 (large k ), the UV portion of \phi is strongly damped, i.e. the measure is finite except for superexponential growth, which we assume is not present. Theorem 2.2 (Well-Defined Functional Measure) . Assume that either 1. the power-law decay parameter satisfies 2\beta>\frac{d}{2}, or 2. an additional exponential damping is present (i.e. \eta>0 and \epsilon(\cdot) does not grow too fast). Then the regulated functional measure \int_{F}D_{\Omega}\phi\,e^{-S[\phi]} remains finite in the cutoff limit \Lambda\to\infty . In other words, D_{\Omega}\phi defines a proper Gaussian-type measure on the space of field configurations. Proof. We begin by expressing the field \phi(x) in momentum space: \phi(x)=\int_{\mathbb{R}^{d}}\tilde{\phi}(k)\,e^{ik\cdot x}\,\frac{d^{d}k}{(2\pi)^{d}}. The regulated quadratic action is defined as S_{\Omega}[\phi,\Lambda]=\frac{1}{2}\int_{\|k\|\leq\Lambda}\Bigl[1-\Omega(k,\Lambda)\Bigr]\,|\tilde{\phi}(k)|^{2}\,d^{d}k, where the suppression function \Omega(k,\Lambda) satisfies 0<\Omega(k,\Lambda)\leq 1. Step 1: Gaussian Integration Over Modes For each momentum mode k , the contribution to the path integral is a one-dimensional Gaussian integral: I(k)=\int_{-\infty}^{+\infty}d\tilde{\phi}(k)\,\exp\!\Bigl[-\frac{1}{2}\Bigl(1-\Omega(k,\Lambda)\Bigr)|\tilde{\phi}(k)|^{2}\Bigr]. This integral evaluates to I(k)=\sqrt{\frac{2\pi}{1-\Omega(k,\Lambda)}}. Thus, the full partition function is expressed as an (infinite) product: Z=\prod_{\|k\|\leq\Lambda}\sqrt{\frac{2\pi}{1-\Omega(k,\Lambda)}}. Taking the logarithm yields \ln Z=\frac{1}{2}\int_{\|k\|\leq\Lambda}\Bigl[\ln(2\pi)-\ln\Bigl(1-\Omega(k,\Lambda)\Bigr)\Bigr]\,d^{d}k. For the measure to be well-defined as \Lambda\to\infty , this integral must converge. Step 2: Analysis in the Infrared (IR) and Ultraviolet (UV) Regimes Infrared (IR) Region: For small momenta ( \|k\|\ll\Lambda ), the suppression function is designed so that \Omega(k,\Lambda)\approx 1 . Here, even though 1-\Omega(k,\Lambda) is small and therefore \ln\bigl(1-\Omega(k,\Lambda)\bigr) is large in magnitude, the IR integration is known to converge. Ultraviolet (UV) Region: For large momenta, we distinguish two cases: 1. Polynomial (Power-Law) Suppression: If \Omega(k,\Lambda)\sim\left(\frac{\Lambda}{\|k\|}\right)^{2\beta}\quad\text{for large }\|k\|, then for \|k\|\gg\Lambda we have 1-\Omega(k,\Lambda)\approx 1 . However, the density of momentum states grows as \sim k^{d-1} . Convergence is then ensured if 2\beta>\frac{d}{2}. 2. Exponential Damping: If an additional exponential factor is present (i.e. terms behaving as \sim e^{-k^{2}/\Lambda^{2}} contribute), the high-momentum modes are suppressed much more strongly. Hence, exponential decay guarantees convergence of the integrals independent of any polynomial growth. Step 3: Convergence of the Functional Integral Since the measure in momentum space is constructed as an infinite product of standard Gaussian integrals, its overall convergence depends on the summability of the logarithms. Under the conditions (assuming an approbiate choice of \epsilon ) 2\beta>\frac{d}{2}\quad\text{or}\quad\eta>0, the high-momentum contributions are sufficiently suppressed, so that the sum (or integral) in the logarithm converges. Thus, the regulated functional measure D_{\Omega}\phi\,e^{-S[\phi]} is finite as \Lambda\to\infty , meaning that D_{\Omega}\phi defines a proper Gaussian-type measure. ∎ 3 Scale-Dependent Suppression of \Omega(k,\Lambda) We now view \Lambda as a running scale, so \Omega(k,\Lambda) evolves with \Lambda . To define this explicitly for any variable \Lambda>0 , we set: \Omega(k,\Lambda)~:=~\frac{1}{1+\bigl(\tfrac{k^{2}}{\Lambda^{2}}\bigr)^{\beta}}~+~\frac{\eta\,e^{-\,k^{2}/\Lambda^{2}}}{\epsilon\bigl(k^{2}\bigr)}. (2) As \Lambda increases, the ratio (k/\Lambda) grows for any fixed k . Hence, the UV

Read the complete text