Improved Epstein--Glaser renormalization in coordinate space. I: Euclidean framework (Q1870258)

Epstein-Glaser (EP) renormalization is a method to regularize a function on \(\mathbb{R}^d\) with polynomial singularity at the origin so that the regularized ``function'' has finite integral on whole \(\mathbb{R}^n\) [\textit{H. Epstein} and \textit{V. Glaser}, Ann. Inst. Henri Poincaré 19, 211-295 (1973)]. In this paper, a refinement of EG renormalization from an analytic point of view is investigated. The author notes that other refinement from the Hopf algebra approach and combinatorial-geometrical aspects will follow as Parts II and III of this series of papers. Let \(f\) be a smooth function on \(\mathbb{R}^n\setminus\{0\}\) and \(O(|x|^{-a})\) as \(x\to 0\), and let \(k= a-d\geq 0\). Let \(\phi\) be a rapidly decreasing testing function. Epstein and Glaser introduced a weight function \(w\); \(w(0)= 1\), \(w^{(\alpha)}(0)= 0\), \(0<|\alpha|\leq k\), and proposed to renormalize \((f,\phi)\) by \[ (T_w, \phi)= (f, T_w\phi),\quad T_w\phi= (1- w) R^{k-1}_0 \phi+ wR^k_0\phi, \] where \(R^k_0\phi\) is the Taylor remainder term corresponding to the \(k\)-jet \(j_0\phi\). The weight function \(w\) should decay at infinity. In this paper, \(w\) is shown to be sufficient to take in \({\mathcal K}'\), the dual of \({\mathcal K}\); the inductive limit of \({\mathcal K}_\gamma\), the space of those smooth functions \(\phi\) such that \(\partial^\alpha\phi(x)= O(|x|^{\gamma- |\alpha|})\), \(|x|\to\infty\). Since \(e^{iqx}\in{\mathcal K}'\), \(q\neq 0\), we can take \(e^{iqx}\), \(q\neq 0\) as the weight function [cf. \textit{R. Estrada}, \textit{J. M. Garcia-Bondía} and \textit{J. C. Várilly}, Commun. Math. Phys. 191, 219-248 (1998; Zbl 0897.60040)]. Taking \(w_\mu(x)= H(\mu^{-1}-|x|)\), \(H\) the Heaviside function, \(T_\mu= T_{w_\mu}\) is computed to be \[ \begin{aligned} T_\mu f(x) &= (-1)^k(k+ 1) \sum_{|\beta|= k+1} \partial^\beta\Biggl[{x^\beta f(x)\over \beta!}\log\mu|x|\Biggr]+ H_k \sum_{|\alpha|= k} c^\alpha \delta^\alpha(x),\\ H_k &= \sum^k_{l=1} {(-1)^{l+1}\over l} {k!\over l!(k- l)!},\quad c^\alpha= \int_{|x|= 1} fx^\alpha.\end{aligned} \] A general formula of \(T_w\) is also given EG renormalization of \(x^{-a}\), \(a= 4,6\) are computed in Sect. 3. They are the extension of these functions as distributions. The answers in Sect. 3 and previous results [e.g. \textit{L. Hörmander}, The analysis of partial differential operators, I, Berlin (1983; Zbl 0521.35001), \textit{V. A. Smirnov} and \textit{O. I. Zavyalov}, Theor. Math. Phys. 96, 974-918 (1993; Zbl 0812.46074)] are compared in Sect. 4-5. In Sect. 4, realizations of refined EG renormalization and Hadamard's finite part theory is also discussed and it is shown that \[ T_\mu(x^{-l}_+)= \text{PF} {H(x)\over x^l}+ \delta^{l-1}(x) \log\mu. \] Physically, examples in Sect. 3, are taken from massless \(\phi^4_4\) theory. Massive cases are discussed in Sect. 6. In Conclusion, the author says the refined EG renormalization may justify the conjecture that Hadamard's finite part theory is in principle enough to deal with quantum field divergence.