Spectral Geometry and the One-Loop QED $\beta$-Function on $S^3 \times S^1$

Spectral Geometry and the One-Loop QED β-Function on S3×S1 Lyudmil Antonov September 30, 2025 Abstract We compute the one-loop QED β-function coefficient directly from heat kernel data of the twisted SpincDirac operator on S3×S1. Using ζ-function regularization, the logarithmic scale dependence is encoded in the a4coefficient of the spectral expansion. We show that the Fµν Fµν contribution to a4reproduces precisely the universal coefficient β(e) = e3/(12π2). The result is independent of the radii of S3and S1and of the choice of gauge background, providing a parameter-free consistency check that spectral data on compact manifolds encode renormalization group information. This calculation demonstrates that universal quantum corrections can be extracted from purely geometric spectral invariants. 1 Introduction Spectral geometry provides a powerful dictionary between heat kernel coefficients of Laplacetype operators and effective field theory counterterms. Building on Gilkey’s invariance theory [1] and Vassilevich’s comprehensive heat kernel review [2], together with the spectral action framework of Connes and Chamseddine [3], one may ask whether elementary renormalization data can be read off from spectral invariants on compact manifolds. In this note we address a fundamental test case: computing the QED one-loop β-function from spectral data on S3(r)×S1(L) with a unit U(1) twist along the Hopf bundle. Our main result is that the F2contribution to the a4heat kernel coefficient reproduces exactly the universal one-loop QED β-function coefficient, with no adjustable parameters. 1.1 Physical motivation and interpretation The choice of S3×S1as our background manifold requires justification, as it differs topologically from physical Minkowski spacetime R3,1. Our calculation is performed in Euclidean signature on a compact manifold for several compelling reasons. First, the high symmetry of the round S3makes all curvature tensors and volume integrals explicit, while the compact geometry ensures that ζ-function regularization is well-defined without infrared divergences, providing computational tractability. Second, the one-loop β-function coefficient is a universal quantity—independent of infrared physics, manifold topology, and gauge background—because it arises from the ultraviolet structure of the theory. Our calculation exploits this universality: the logarithmic term in the heat kernel expansion captures local UV behavior that is the same on any manifold and for any gauge configuration. A perturbative expansion around a trivial (zero) gauge background would yield identical logarithmic divergences, confirming that our use of the Hopf bundle is a computational convenience that does not affect the universal result. Finally, the unit Chern class of the Hopf bundle provides a minimal non-trivial gauge configuration that probes the coupling between spinors and gauge fields without introducing perturbative complications. The quantized flux RS2F/(2π) = 1 ensures we work in a topologically stable sector, making the calculation particularly clean. 1 The independence of our result from the radii rand L, as well as from the specific choice of gauge background, demonstrates that we have isolated a genuinely universal quantity. In the language of effective field theory, the a4coefficient encodes the coefficient of the logarithmic divergence that appears in dimensional regularization, independent of the choice of background metric or gauge configuration. This justifies using the compact manifold as a computational device to extract physics that applies equally to flat space QED. 1.2 Connection to the spectral action principle Our calculation provides a concrete verification of the spectral action approach at the one-loop level. In the full spectral action framework [3, 4], all physical scales—including the UV cutoff Λ—are intrinsically tied to the spectrum of the Dirac operator through a cutoff function. The energy scale emerges naturally from spectral density rather than being imposed externally. The spectral action takes the form Tr[f(D/Λ)], where the function fencodes not merely a cutoff but contains the Standard Model action parameters themselves. Our work demonstrates that the form of renormalization group flow (encoded in the βfunction) follows from spectral geometry, while leaving the determination of absolute scales (the value of αat a given energy) to the full spectral action cosmology. The challenge in that broader program is to show that the running we have calculated is consistent with the physical values of the coupling parameters at experimentally accessible scales. This separation between universal RG structure (which we derive) and scale-fixing (which requires the full cosmological framework) is a feature, not a limitation, of the geometric approach. Our calculation verifies the consistency of the method at the perturbative level, supporting the broader spectral action research program. 1.3 Conventions and sign choices We work throughout in Euclidean signature with {γµ, γν}= 2δµν. The square of the twisted Dirac operator is written in Laplace form D2 A=−∇2+E, (1) consistent with the heat kernel literature (see Theorems 4.8.18 in Gilkey [1] and Section 3.3 of Vassilevich [2]). The classical Maxwell action is Scl =1 4e2ZM FµνFµν dV, (2) so that FµνFµν ≥0 in Euclidean signature. With these conventions, the sign of the logarithmic counterterm matches the standard QED one-loop β-function. Alternative conventions (e.g., Minkowski signature with −1 4F2) differ only by analytic continuation and yield the same βfunction coefficient. 2 Geometric Setup Let M=S3(r)×S1(L) with product metric, where ris the radius of S3and Lthe circumference of S1. We equip S3with its canonical Hopf fibration π:S3→S2and twist the spinor bundle by the associated principal U(1) bundle L. 2.1 Metrics and curvature The metric on S3is the standard round metric with scalar curvature RS3= 6/r2. The metric on S1is gS1= (L/2π)2dθ2, which is flat with RS1= 0. In an orthonormal coframe {ei}on S3 2 (given by ei=rσiwhere {σi}are left-invariant one-forms on SU(2)) extended by e4= (L/2π)dθ on S1, the curvature two-forms are standard and all components are explicitly computable. 2.2 Gauge connection and Hopf bundle The connection one-form Arepresents the Hopf bundle. We choose the normalization A=1 rσ3,(3) which gives field strength F=dA =2 r3e1∧e2.(4) This satisfies the quantization condition (see Appendix A) 1 2πZS2 F= 1,(5) corresponding to first Chern class c1(L) = 1. In the orthonormal frame, the components of Fare constant: F12 =−F21 = 2/r3, with all other components zero. This yields FµνFµν = 8/r6in the orthonormal frame. The volume form is dV = (r3sin θ dθ ∧dϕ ∧dψ)∧(L/2π dθS1), where θ, ϕ, ψ are coordinates on S3and θS1is the coordinate on S1. The integral RMFµνFµνdV evaluates to 8π2L/r3, but the βfunction derivation depends only on the coefficient of this term in the effective action, which is independent of rand L. 2.3 Twisted Dirac operator The SpincDirac operator with U(1) twist is DA=γµ(∇µ+iAµ),(6) where ∇µis the spin connection on S3×S1. Its square takes the Laplace form D2 A=−∇2+E, (7) where the endomorphism Econtains both curvature and gauge contributions. Following the standard heat kernel literature (Theorem 4.8.16–18 in [1]), for a Dirac operator we have: E=R 4+i 2γµνFµν.(8) This form ensures consistency with the general theory of Laplace-type operators and the heat kernel expansion.1 3 Heat Kernel Expansion and the a4Coefficient 3.1 General structure For a Laplace-type operator P=−∇2+Eon a four-manifold, the local heat kernel has the asymptotic expansion Tr(e−tP )∼(4πt)−2 ∞ X k=0 a2k(P)tk, t →0+.(9) For dimension n= 4, the standard Gilkey form (4πt)−n/2Paktk/2simplifies to this expression. 1The factor i/2 follows from Euclidean continuation of the Minkowski coupling ie ¯ ψγµAµψ; equivalent to 1/4[γµ, γν]Fµν up to Clifford definitions. 3 Remark 3.1. For manifolds with boundary, additional terms appear in the heat kernel expansion [2]. Here, the compactness of S3×S1without boundary ensures pure volume integrals. The relevant local invariants appearing in a4are given by the Seeley–DeWitt–Gilkey formula (Theorem 4.1.16–18 in [1]): Lemma 3.2 (Gilkey).The coefficient a4(P)for a twisted Dirac operator on a four-manifold contains the gauge contribution a4(P)⊃(4π)−2ZM1 12tr(ΩµνΩµν) + 1 2tr(E2)dV +(curvature-only terms),(10) where Ωµν is the total connection curvature on the twisted bundle. 3.2 Bundle curvature decomposition Lemma 3.3. For the SpincDirac operator DA, the total connection curvature decomposes as Ωµν =1 4Rµνρσγρσ +iFµν,(11) where the first term is the spin connection curvature and the second is the U(1) gauge curvature. Proof. The Spincbundle is the tensor product of the spinor bundle (with spin connection) and the U(1) line bundle (with gauge connection). The total curvature is the sum of the two contributions acting on the respective factors. 3.3 Isolating the F2contribution Only terms quadratic in the gauge field Fcontribute to the gauge kinetic term renormalization. We systematically extract these from both the Ω2and E2terms. Lemma 3.4. The gauge contribution to the trace of ΩµνΩµν is tr(ΩµνΩµν)F2=−4FµνFµν.(12) Proof. Using Lemma 3.3, the product ΩµνΩµν contains three types of terms: •Spin–spin: 1 16 RµνρσRµνρσ(γρσγρσ) (curvature-only), •Spin–gauge: i 4RµνρσγρσFµν (vanishes under spinor trace since tr(γρσ) = 0 by the standard Clifford algebra trace identities [7]), •Gauge–gauge: (iFµν)(iFµν) = −FµνFµν. The gauge–gauge block contributes −FµνFµν times the spinor trace factor trspin(1) = 4, giving the stated result. Lemma 3.5. The gauge contribution to the trace of E2is tr(E2)F2=−2FµνFµν.(13) Proof. Using the expression E=R 4+i 2γµνFµν, we expand: E2=R 42 +R 4·i 2γµνFµν +i 2γµνFµν ·R 4−1 4FµνFρσγµνγρσ.(14) 4 The first term is curvature-only, the second and third terms vanish under spinor trace (since tr(γµν) = 0 by the standard Clifford algebra trace identities [7]), leaving the fourth term. Using the standard Clifford trace identity tr(γµνγρσ) = 4(gµρgνσ −gµσgνρ),(15) we contract: tr(FµνFρσγµνγρσ) = 4FµνFρσ(gµρgνσ −gµσgνρ) (16) = 4(FµνFµν −FµνFνµ) (17) = 8FµνFµν.(18) Thus tr(E2)F2=−1 4·8FµνFµν =−2FµνFµν.2 Theorem 3.6. The gauge contribution to the a4coefficient is a4F2= (4π)−2−4 3ZM FµνFµν dV. (19) Proof. Combining Lemmas 3.4 and 3.5 with their prefactors from Lemma 3.2: a4F2= (4π)−2ZM1 12(−4FµνFµν) + 1 2(−2FµνFµν)dV (20) = (4π)−2ZM−1 3−1FµνFµν dV (21) = (4π)−2−4 3ZM FµνFµν dV. (22) Remark 3.7. Higher heat kernel coefficients a6, a8, . . . contribute power-suppressed terms (proportional to 1/µ2,1/µ4, etc.) involving higher derivatives or more curvature factors. For instance, according to the general formulas in Vassilevich [2] and Avramidi [6], the a6coefficient includes terms such as RFµνFµν,(∇ρFµν)(∇ρFµν), RµνFµρFνρ,(23) while a8includes terms like R2FµνFµν, RµνRµνFρσFρσ,(FµνFµν)2.(24) These are finite, non-universal corrections to the effective action that do not affect the logarithmic running encoded in a4. This clean separation between universal (logarithmic, a4) and non-universal (power-suppressed, ak>4) contributions is a key feature of the heat kernel approach. 4 Mapping to the β-Function via ζ-Regularization 4.1 Effective action from the spectral zeta function The ζ-regularized one-loop effective action is defined by Γ[A] = −1 2ζ′ D2 A(0),(25) 2The minus sign arises from squaring the factor of iin the Euclideanized gauge coupling; see Lawson–Michelsohn Appendix D for conventions. 5 where the spectral zeta function is ζD2 A(s) = Tr[(D2 A)−s] = 1 Γ(s)Z∞ 0 ts−1Tr(e−tD2 A)dt. (26) Substituting the heat kernel expansion, we find ζD2 A(s) = (4π)−2 Γ(s)Z∞ 0 ts−1a4dt + (terms regular at s= 0).(27) The integral R∞ 0ts−1dt has a pole at s= 0. Upon analytic continuation and taking the derivative at s= 0, this pole becomes a logarithm. Introducing a renormalization scale µto make the ζ-function dimensionless, we obtain Γ[A]⊃1 2ln(µ2)a4(D2 A) (28) where a4(D2 A) is the standard Seeley–DeWitt coefficient. The overall factor of 1 2reflects both the use of D2 Arather than DAdirectly and the fermionic minus sign in the functional determinant. Our normalization matches the treatments of Avramidi [6] and Vassilevich [2]. Different sign conventions for the Euclidean action may shift this prefactor, but the final β–function coefficient is universal. This procedure is equivalent to minimal subtraction (MS) in dimensional regularization for the present calculation, as both methods isolate the same logarithmic divergence structure.3 While the finite parts of the effective action can be scheme-dependent, the coefficient of the logarithmic divergence—and hence the β-function—is a universal quantity. This universality ensures that our result is valid across all standard renormalization schemes. 4.2 One-loop correction to the gauge coupling The classical Maxwell action is Scl[A] = 1 4e2ZM FµνFµν dV. (29) By Theorem 3.6, the one-loop quantum correction is Γ1-loop[A] = 1 2ln(µ2)·(4π)−2−4 3ZM FµνFµν dV. (30) The total effective action at one loop is Γtotal[A] = Scl[A]+Γ1-loop[A] = 1 4e2−2 3(4π)2ln µ ΛZM FµνFµν dV, (31) where Λ is an arbitrary reference scale. Thus the running coupling satisfies 1 4e2(µ)=1 4e2(Λ) −2 3(4π)2ln µ Λ.(32) 4.3 The β-function Differentiating with respect to ln µ: µd dµ 1 e2=−8 3(4π)2=−1 6π2.(33) 3The factor 1/2 accounts for the Dirac operator being first-order; for scalars, it would be 1. 6 The factor −8/3 arises from −4/3 in a4multiplied by 2 from the zeta-function regularization of the Dirac operator. Since d dµ 1 e2=−2 e3 de dµ,(34) we obtain the β-function: β(e) = µde dµ =e3 12π2.(35) This is precisely the standard QED one-loop result for a single Dirac fermion of charge 1 (see equation (12.61) in Peskin and Schroeder [5]). 5 Discussion and Physical Interpretation 5.1 Universality and parameter independence The central result—that spectral data on S3×S1encode the universal one-loop β-function coefficient—demonstrates remarkable independence from the radius rof S3, the circumference L of S1, and the choice of gauge background. This triple independence is not accidental but reflects the fundamental nature of the β-function as a universal, UV quantity determined entirely by the local structure of the quantum field theory. The heat kernel coefficient a4captures precisely this local UV information through its role as the coefficient of the logarithmic divergence. Our use of the Hopf bundle provides a concrete, topologically non-trivial configuration for the calculation, but the universality of the result ensures that a perturbative expansion around zero gauge field (or any other background) would yield the same logarithmic coefficient. This behavior is a direct consequence of the general structure of renormalization: UV divergences depend only on the local operator content, not on global topology or boundary conditions. Our calculation establishes several important points. The spectral action approach of Connes and Chamseddine correctly encodes renormalization group physics at the one-loop level, demonstrating the viability of this geometric framework. The choice of background manifold and gauge configuration is immaterial for universal quantities—only the local operator structure matters. Most significantly, no adjustable parameters or fitting procedures are required; the result follows purely from geometric spectral data and the Spinctwist, providing a parameter-free derivation of a fundamental quantum field theory quantity. 5.2 Limitations and the UV scale problem While our calculation successfully reproduces the β-function coefficient, it does not determine the absolute value of the coupling α(µ) = e2/(4π) at any particular scale. Such a determination would require additional input in the form of a geometric prescription for the UV boundary condition e(Λ). In the full spectral action framework, the physical UV scale Λ is intrinsically tied to the energy scale of the Dirac operator through a cutoff function f(D2/Λ2). The spectral density of the Dirac operator, rather than an externally imposed cutoff, determines the effective energy scale. Moreover, the function fin the spectral action Tr[f(D/Λ)] is not merely a regulator but encodes the Standard Model action parameters themselves. This challenge is inherent to the spectral action program, where the scale Λ is ultimately tied to the gravitational sector and the spectrum of the Dirac operator on a cosmological background. Our work verifies that the form of the renormalization group flow (the β-function) emerges correctly from spectral geometry, demonstrating the consistency of the approach at the perturbative level. The determination of absolute coupling values requires the full spectral action machinery, including gravitational sector couplings and cosmological boundary conditions. The 7 broader research program then seeks to show that the running we have calculated is consistent with the physical values of these coupling parameters at experimentally accessible energy scales. Promising future directions for this program include computing higher-loop corrections and summing renormalization group equations on spectral backgrounds, connecting the UV scale to Planck-scale physics through unified spectral models, and seeking consistency conditions from anomaly cancellation across all Standard Model sectors in a Spincframework. Chiral extensions may require additional anomaly cancellation considerations, as in the full spectral Standard Model construction [3]. 5.3 Comparison with related work Our result complements and extends previous work on spectral methods in quantum field theory in several important ways. Avramidi’s comprehensive treatment develops heat kernel techniques for coupled gravitational and gauge systems using the background field method; our calculation provides an explicit worked example in the pure gauge sector with full technical detail. The spectral action principle proposes that all of particle physics emerges from spectral data; our verification of the QED β-function at one loop supports this program while clarifying the distinction between universal RG structure (which we derive) and absolute scale-fixing (which requires additional input). Vassilevich’s comprehensive review catalogs heat kernel coefficients in full generality; we have applied these formulas to extract a specific physical observable with clear field-theoretic interpretation, demonstrating the practical utility of these general results for concrete physical calculations. Appendix A: The Hopf Bundle and Flux Quantization The Hopf fibration π:S3→S2is the principal U(1) bundle over the two-sphere with total space S3. Viewing S3as the unit sphere in C2, S3={(z1, z2)∈C2:|z1|2+|z2|2= 1},(36) the Hopf map is given by π(z1, z2) = 2z1¯z2,|z1|2− |z2|2∈S2⊂R3.(37) The connection one-form αon S3satisfies dα =π∗(ωS2), where ωS2is the area form on S2 normalized so that RS2ωS2= 4π. With this normalization, 1 2πZS2 F= 1,(38) confirming that the U(1) bundle has first Chern class c1(L) = 1 (see Definition II.1.3 and Remark II.1.8 in Lawson and Michelsohn [7]). In our setup, we take F=dA with A= (1/r)σ3on the round S3of radius r. The factor of 1/r ensures the correct normalization as rvaries. In the orthonormal frame, the non-zero components are F12 =−F21 = 2/r3, giving FµνFµν = 8/r6. The volume element is dV = r3sin θ dθ ∧dϕ ∧dψ ∧(L/2π)dθS1, and the integral RMFµνFµνdV yields 8π2L/r3, but does not affect the universal β-function coefficient. References [1] Peter B. Gilkey. Invariance Theory, the Heat Equation, and the Atiyah-Singer Index Theorem, first edition. Publish or Perish, Inc., Wilmington, Delaware, 1984. (See especially Theorems 4.8.16–18 for the a4coefficient formula.) 8 [2] Dmitri V. Vassilevich. Heat kernel expansion: User’s manual. Physics Reports, 388(56):279–360, 2003. arXiv:hep-th/0306138. DOI: 10.1016/j.physrep.2003.09.002. (Section 3.3 covers Dirac operators; Section 5 catalogs coefficients.) [3] Alain Connes and Ali H. Chamseddine. The spectral action principle. Communications in Mathematical Physics, 186(3):731–750, 1997. arXiv:hep-th/9606001. DOI: 10.1007/s002200050133 (Springer CMP) [4] Ali H. Chamseddine and Alain Connes. The spectral action principle in particle physics and cosmology. In Handbook of Pseudoriemannian Geometry and Supersymmetry, IRMA Lectures in Mathematics and Theoretical Physics, Vol. 16, pages 1–42. European Mathematical Society, 2007. arXiv:hep-th/0608226. ISBN 978-3-03719-027-2 [5] Michael E. Peskin and Daniel V. Schroeder. An Introduction to Quantum Field Theory. Addison-Wesley, Reading, MA, 1995. ISBN 978-0-201-50397-5 (Equation (12.61) gives the one-loop QED β-function.) [6] Ivan G. Avramidi. Heat Kernel and Quantum Gravity. Lecture Notes in Physics Monographs, vol. 64. Springer-Verlag, Berlin, 2000. DOI: 10.1007/3-540-46523-5. (Comprehensive treatment of heat kernel methods using background field techniques.) [7] H. Blaine Lawson, Jr. and Marie-Louise Michelsohn. Spin Geometry. Princeton Mathematical Series, vol. 38. Princeton University Press, Princeton, NJ, 1989. ISBN 978-0-69108039-8. (Definition II.1.3 and Remark II.1.8 cover Chern classes of line bundles; Appendix D.2 provides Clifford algebra trace identities.) [8] Bryce S. DeWitt. Dynamical Theory of Groups and Fields. Gordon and Breach, New York, 1965. ISBN 978-0-677-30000-3. (Background field method and heat kernel techniques.) [9] Robert T. Seeley. Complex powers of an elliptic operator. In Singular Integrals (Proc. Sympos. Pure Math., Chicago, Ill., 1966), pages 288–307. Amer. Math. Soc., Providence, RI, 1967. ISBN 978-0-8218-1318-0. (Foundational work on ζ-function regularization.) [10] A. O. Barvinsky and G. A. Vilkovisky. The generalized Schwinger–DeWitt technique in gauge theories and quantum gravity. Physics Reports, 119:1–74, 1985. DOI: 10.1016/03701573(85)90148-6. (Heat kernel methods in gauge theories.) 9