Three-loop singularity structure for a non-linear sigma model

P. V. Akacevich, A. V. Ivanov and I. V. Korenev

arXiv: 2507.05923v1 · PDF · 8 July 2025
Categories: hep-th, math-ph, math.MP
Licence: CC BY 4.0 The site shows the text.

Abstract

The paper is devoted to the three-loop renormalization of the effective action for a two-dimensional non-linear sigma model using the background field method and a cutoff regularization in the coordinate representation. The coefficients of the renormalization constant and the necessary auxiliary vertices are found, as well as the asymptotic expansions of all three-loop diagrams, and their dependence on the type of regularizing function. A comparison is also made with the standard case of cutoff in the momentum representation.

Items 24

Theorem

Theorem 1

D2S2

Let all the conditions stated above be fulfilled. Then it is possible to exclude nonlocal (disproportionate to the classical action) singular contributions from the three-loop correction (?) using the following choice of auxiliary renormalization vertices

\Gamma_{r,3}=-\frac{L_1}{3\pi}\hat{\Gamma}_3,

\Gamma_{r,4}=\Lambda^2\bigg(\frac{\alpha_1}{2\pi}-\frac{5\alpha_6}{2}\bigg) \widetilde{\Gamma}_4 -\frac{15\Lambda^2c_2^2}{2\pi}\mathrm{R}_0\tau_5- \frac{5L_1}{32\pi^2}\Big(c_2^2\mathrm{R}_1\tau_1+c_2\mathrm{R}_2\tau_2+\mathrm{R}_3\tau_3+c_2^2\mathrm{R}_4\tau_4\Big),

where auxiliary finite numbers are defined by the equalities

\begin{aligned} \tau_1=&\,4\theta_3-10\theta_2-2\theta_1+26\alpha_{11},\\ \tau_2=&\,2\alpha_8+34\theta_3-154\theta_2/5+8\theta_1+14\alpha_{11},\\ \tau_3=&\, 8\alpha_8-4\theta_3-164\theta_2/5+8\theta_1-6\alpha_{11},\\ \tau_4=&\,10\alpha_8-60\theta_2+52\theta_1+44\alpha_{11},\\ \tau_5=&-\mathbf{f}(0)/2+\alpha_7-4\alpha_{10}. \end{aligned}

They depend only on the deforming function \mathbf{f}(\cdot). Singular terms proportional to the classical action S[B] or independent of the background field at all can be reduced by a suitable choice of the coefficients a_2 and \hat{\varkappa}_2. In this case, the structure of the value a_2 has the following form

a_2=\frac{a_0a_1}{2}-\frac{c_2^3L_1^2}{16(4!)^2\pi^3}\big(2\tau_1-2\tau_2+2\tau_3+\tau_4\Big)+L_1t,

where t\in\mathbb{R} and it depends only on the regularizing function. \mathbf{f}(\cdot).

Used by Theorem 2

Theorem

Theorem 2

U6CP

Let the assumptions of Theorem 1 be fulfilled. Let \{\mathrm{K}_i(\cdot)\}_{i=1}^4 be a set of continuous functions such that the relations hold

\mathrm{supp}(\mathrm{K}_i)\subset\mathrm{B}_1,\,\,\, \int_{\mathbb{R}^2}\mathrm{d}^2x\,\mathrm{K}_i(x)=\tau_i,

for all index values i. Next, using the functionals from Section ?, we define four vertices

\hat{\mathrm{V}}_i[\phi]=\mathrm{V}_i[\mathrm{K}_i,\phi],

where i\in\{1,\ldots,4\}. Then, as an auxiliary vertex \Gamma_{r,4} from Theorem 1, we can choose a functional of the form

\begin{gathered} \Lambda^2\bigg(\frac{\alpha_1}{2\pi}-\frac{5\alpha_6}{2}\bigg) \widetilde{\Gamma}_4 -\frac{15\Lambda^2c_2^2}{2\pi}\mathrm{R}_0\tau_5- \frac{5L_1}{32\pi^2}\Big(c_2^2\mathrm{R}_1\tau_1\upsilon_1+c_2\mathrm{R}_2\tau_2\upsilon_2+\mathrm{R}_3\tau_3\upsilon_3+c_2^2\mathrm{R}_4\tau_4\upsilon_4\Big)-\\-\frac{5}{32\pi} \Big(c_2\hat{\mathrm{V}}_1\vartheta_1+c_2\hat{\mathrm{V}}_2\vartheta_2+\hat{\mathrm{V}}_3\vartheta_3+c_2\hat{\mathrm{V}}_4\vartheta_4\Big), \end{gathered}

in which the coefficients \{\upsilon_i\}_{i=1}^4 and \{\vartheta\}_{i=1}^4 are subject to the following conditions

\begin{aligned} \upsilon_1-\vartheta_1+\vartheta_2-\vartheta_3=&\,1,\\ \upsilon_2-\vartheta_2+\vartheta_4/2=&\,1,\\ \upsilon_3-\vartheta_3=&\,1,\\ \upsilon_4+\vartheta_4=&\,1. \end{aligned}

In this case, the third coefficient a_2 of the renormalization constant becomes equal to

a_2\to a_2=\frac{a_0a_1}{2}-\frac{c_2^3L_1^2}{16(4!)^2\pi^3}\big(2\tau_1\upsilon_1-2\tau_2\upsilon_2+2\tau_3\upsilon_3+\tau_4\upsilon_4\Big)+L_1t_1,

where the number t_1\in\mathbb{R} depends on the regularizing function \mathbf{f}(\cdot), the set of coefficients \{\vartheta_i\}_{i=1}^4, and the set of kernels \{\mathrm{K}_i(\cdot)\}_{i=1}^4, and it does not depend on \{\upsilon_i\}_{i=1}^4. In particular, fixing the values as follows \upsilon_1=\upsilon_2=\upsilon_3=\upsilon_4=0, we obtain coefficients for quasi-local vertices in the form

\vartheta_1=-1/2,\,\,\, \vartheta_2=-1/2,\,\,\, \vartheta_3=-1,\,\,\, \vartheta_4=1.

In this case, the coefficient for the renormalization constant takes the form

a_2\to a_2=\frac{a_0a_1}{2}+L_1t_2,

where t_2\in\mathbb{R} depends on the regularizing function \mathbf{f}(\cdot) and the set of kernels \{\mathrm{K}_i(\cdot)\}_{i=1}^4.

Uses Theorem 1

Lemma

Lemma 1

0J0T

Taking into account the type of regularization, the relations \theta_k\leqslant0 are correct for k>0. The constraint a_1\geqslant0 for the two-loop renormalization coefficient is also valid.

Lemma

Lemma 2

1Y0Z

Taking into account all of the above, the following relations are true

G_{1}^\mu(0)=0,\,\,\,G_{11}^{\nu\mu}(0)=0,\,\,\, \partial_{x_\rho}G_{11}^{\nu\mu}(x-y)\big|_{y=x}=0,\,\,\, \partial_{x_\rho}G_{2}^{\nu\mu}(x-y)\big|_{y=x}=0,

G_{2}^{\nu\mu}(0)=\frac{L\alpha_{5}(\mathbf{f})}{\Lambda^2}\delta_{\mu\nu}+\mathcal{O}\big(1/\Lambda^2\big),

\partial_{x^\mu}G_{\Lambda}^{ab}(x,y)\big|_{y=x}= \frac{L_1}{2\pi}B_\mu^{ab}(x)+2\partial_{x^\mu}PS_{\Lambda}^{ab}(x,y)\big|_{y=x} +\mathcal{O}\big(L/\Lambda^2\big),

D_{\mu}^{ac}(x)G_{\Lambda}^{cb}(x,y)\big|_{y=x}= -\frac{L_1}{2\pi}B_\mu^{ab}(x)+2D_{\mu}^{ac}(x)PS_{\Lambda}^{cb}(x,y)\big|_{y=x} +\mathcal{O}\big(L/\Lambda^2\big),

\partial_{x^\mu}G_{\Lambda}^{ab}(x,y)\big|_{y=x}+ \partial_{y^\mu}G_{\Lambda}^{ab}(x,y)\big|_{y=x}= 2\partial_{x^\mu}PS_{\Lambda}^{ab}(x,y)\big|_{y=x}+ 2\partial_{y^\mu}PS_{\Lambda}^{ab}(x,y)\big|_{y=x}+\mathcal{O}\big(L/\Lambda^2\big),

D_{\mu}^{ac}(x)G_{\Lambda}^{cb}(x,y)\big|_{y=x}+ D_{\mu}^{bc}(y)G_{\Lambda}^{ac}(x,y)\big|_{y=x}= 2 D_{\mu}^{ac}(x)PS_{\Lambda}^{cb}(x,y)\big|_{y=x}+ 2 D_{\mu}^{bc}(y)PS_{\Lambda}^{ac}(x,y)\big|_{y=x}+\mathcal{O}\big(L/\Lambda^2\big).

Proof of Lemma 2

All relations except (2) are a consequence of the spherical symmetry of the densities and the explicit decomposition from (?). The remaining equality from (2) is presented in Section ?.

Lemma

Lemma 3

4QH2

Let us consider three auxiliary matrix-valued operators

M_1^{ab}(y)=\int_{\mathbb{R}^2}\mathrm{d}^2x\, \Big(\partial_{y_\mu}G_\Lambda^{ad}(y,x)\Big) \Big(A^{de}(x)G_\Lambda^{eb}(x,y)\Big),

M_2^{ab}(y)=\int_{\mathbb{R}^2}\mathrm{d}^2x\, \Big(D_{\mu}^{ac}(y)G_\Lambda^{cd}(y,x)\Big) \Big(A^{de}(x)G_\Lambda^{eb}(x,y)\Big),

M_3^{ab}(y)=\int_{\mathbb{R}^2}\mathrm{d}^2x\, \Big(A^{ac}(y)G_\Lambda^{cd}(y,x)\Big) \Big(A^{de}(x)G_\Lambda^{eb}(x,y)\Big).

Then, taking into account all the above, the following relations are true

M^{ab}_1(y)=\frac{L_1+2\theta_2}{2\pi}B_\mu^{ab} +\partial_{y_\mu}PS_\Lambda^{ad}(y,x)\big|_{x=y}+\mathcal{O}(L/\Lambda),

M_2^{ab}(y)=-\frac{L_1-2\theta_2+2\theta_1}{2\pi}B_\mu^{ab} +D_{\mu}^{ac}(y)PS_\Lambda^{ad}(y,x)\big|_{x=y}+\mathcal{O}(L/\Lambda),

\frac{1}{2}\Big(M_3^{ab}(y)+M_3^{ba}(y)\Big)=\Lambda^2\alpha_6(\mathbf{f}) -\frac{\theta_3+\theta_2}{4\pi}B_\mu^{ac}(y)B_\mu^{cb}(y) +\mathcal{O}(L/\Lambda).

Proof of Lemma 3

Decompositions follow from substitutions of (?) and (?) into integrals and additional integration by parts.

Lemma

Lemma 4

LKF9

Taking into account all of the above, the following relation holds

\begin{aligned} \mathbb{H}_0^{\mathrm{sc}}(\Gamma_3\Gamma_5)\stackrel{\mathrm{s.p.}}{=} &-\frac{25c_2L_1}{6\pi}\mathbb{H}_0^{\mathrm{sc}}(\Gamma_3\Gamma_3)+\frac{L_1}{3\pi}\mathbb{H}_0^{\mathrm{sc}}(\Gamma_3\hat{\Gamma}_3)\\&+ \frac{30L}{\pi}\Big(\mathrm{A}_1+2\mathrm{A}_2-2\mathrm{A}_3-\mathrm{A}_4\Big) -\frac{15c_2L}{\pi} \Big(\mathrm{B}_1+2\mathrm{B}_2+2\mathrm{B}_3\Big)\\&+ \frac{15(L_1^2+2L\theta_1^{\phantom{1}})}{4\pi^2} \Big(2c_2^2J_1[B]-c_2J_2[B]+2J_3[B]\Big). \end{aligned}

Used by Proof

Lemma

Lemma 5

4L61

Let \sigma_1>0. Then, taking into account all the above, the equality is true

\mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^4\big)\stackrel{\mathrm{s.p.}}{=} -2^23^3\ln(\Lambda/\sigma_1)\mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\mathrm{X}_1^{\phantom{1}}\big)-2^23^5\ln^2(\Lambda/\sigma_1)\mathbb{H}_0\big(\mathrm{X}_1^2\big).

Proof of Lemma 5

The contribution \mathbb{H}_0\big(\Gamma_3^4\big) contains diagrams topologically equivalent to either triangular or cylindrical diagrams, see (?). The first type of diagrams does not provide a singular contribution, which is a consequence of the low dimension, that is d=2, the number of integration operators, and the fact that only single lines appear in such diagrams. Auxiliary estimates for integrals appearing in triangular diagrams are given in Section ?.

Turning to the cylindrical part, we note that the contributions are not singular if two derivatives act on the side line (left or right), which is a consequence of the convergence of the integral

\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2x\,A_0^{\phantom{1}}(x)G^{1,\mathbf{f}}(x)=1,

which in fact does not depend on the regularizing parameter \Lambda. This reasoning leads to the fact that it is convenient to cut a cylindrical diagram into two parts (upper and lower), which in diagrammatic language can be represented by the same vertex \mathbb{H}_2^{\mathrm{c}}(\Gamma_3^2), and then investigate this part separately.

Thus, the analysis of the cylindrical diagram reduces to the asymptotic decomposition of the integral \mathbb{H}_2^{\mathrm{c}}(\Gamma_3^2) with respect to the parameter \Lambda as a functional considered on smooth fields. Note that it can be represented as a linear combination of auxiliary vertices, see (?). Taking into account the result from Lemma 14 and the Jacobi identity (?) for structure constants, repeating the calculations from the fifth stage of the proof of Lemma 4, we can verify the validity of the following equalities

\mathrm{C}_1\stackrel{\mathrm{s.p.}}{=}-2\ln(\Lambda/\sigma_1)\mathrm{X}_1,\,\,\, \mathrm{C}_2\stackrel{\mathrm{s.p.}}{=}-4\ln(\Lambda/\sigma_1)\mathrm{X}_1,\,\,\, \mathrm{C}_3\stackrel{\mathrm{s.p.}}{=}2\ln(\Lambda/\sigma_1)\mathrm{X}_1,

\mathrm{C}_4\stackrel{\mathrm{s.p.}}{=}4\ln(\Lambda/\sigma_1)\mathrm{X}_1,\,\,\, \mathrm{C}_5\stackrel{\mathrm{s.p.}}{=}2\ln(\Lambda/\sigma_1)\mathrm{X}_2,\,\,\, \mathrm{C}_6\stackrel{\mathrm{s.p.}}{=}-2\ln(\Lambda/\sigma_1)\mathrm{X}_1+2\ln(\Lambda/\sigma_3)\mathrm{X}_2,

\mathrm{C}_i\stackrel{\mathrm{s.p.}}{=}0,

where i\in\{7,\ldots,12\}. Next, for convenience, we define two functionals \mathrm{C}=\mathbb{H}_2^{\mathrm{c}}(\Gamma_3^2) and \mathrm{C}_0=-18\mathrm{X}_1. Then, based on the last calculations, we can make sure that the equality \mathrm{C}\stackrel{\mathrm{s.p.}}{=}\ln(\Lambda/\sigma_1)\mathrm{C}_0 holds and the chain of relations is true

\begin{aligned} \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^4\big)&\stackrel{\mathrm{s.p.}}{=} 3\mathbb{H}_0^{\mathrm{sc}}\big(\mathbb{H}_2^{\mathrm{c}}(\Gamma_3^2)\mathbb{H}_2^{\mathrm{c}}(\Gamma_3^2)\big)= 3\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{C}^2\big)\\&\,=\, 3\ln^2(\Lambda/\sigma_1)\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{C}_0^2\big)+6\ln(\Lambda/\sigma_1) \mathbb{H}_0^{\mathrm{sc}}\big((\mathrm{C}-\ln(\Lambda/\sigma_1)\mathrm{C}_0^{\phantom{1}})\mathrm{C}_0^{\phantom{1}}\big) +3\mathbb{H}_0^{\mathrm{sc}}\big((\mathrm{C}-\ln(\Lambda/\sigma_1)\mathrm{C}_0^{\phantom{1}})^2\big)\\&\stackrel{\mathrm{s.p.}}{=} 6\ln(\Lambda/\sigma_1)\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{C}\mathrm{C}_0^{\phantom{1}}\big)-3\ln^2(\Lambda/\sigma_1)\mathbb{H}_0^{\mathrm{c}}\big(\mathrm{C}_0^2\big). \end{aligned}

From the last equalities the main statement follows.

Uses Lemma 14, Lemma 4

Lemma

Lemma 6

6FJR

Taking into account all the above, the following decomposition is true for the first part of (?)

\begin{aligned} \mathrm{H}_{334}^{1} \stackrel{\mathrm{s.p.}}{=}\,& 18\ln(\Lambda/\sigma_2)\Lambda^2\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{X}_1^{\phantom{1}} \widetilde{\Gamma}_2^{\phantom{1}}\big)+\Lambda^2 \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\widetilde{\Gamma}_2^{\phantom{1}}\big)+\frac{12c_2L_1}{\pi}\mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\big)\\+\,&\,3\ln(\Lambda/\sigma_3) \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\mathrm{X}_1^{\phantom{1}}\big)-18\ln(\Lambda/\sigma_2)\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{X}_1^{\phantom{1}}\Gamma_4^{\phantom{1}}\big) +54\ln(\Lambda/\sigma_2)\ln(\Lambda/\sigma_3) \mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{X}_1^{2}\big) \\-\,&\,\frac{144L}{\pi} \Big(\mathrm{A}_1-2\mathrm{A}_2+\mathrm{A}_4\Big)+ \frac{36c_2^2L_1\big(\ln(\Lambda/\sigma_2)-2\theta_2^{\phantom{1}}\big)}{\pi^2} J_1[B]\\+\,&\, c_2\frac{18\big(L_1^2+2L\theta_1^{\phantom{1}}-2\ln(\Lambda/\sigma_2)L_2\big)}{\pi^2}J_2[B]- \frac{36\big(L_1^2+2L\theta_1^{\phantom{1}}+2L\theta_2^{\phantom{1}}\big)}{\pi^2}J_3[B] +\hat{\kappa}_1S[B]. \end{aligned}

Lemma

Lemma 7

1BOB

Taking into account all the above, the following decomposition is true for the second part of (?)

\begin{aligned} \mathrm{H}_{334}^{1}\stackrel{\mathrm{s.p.}}{=}& \frac{2c_2}{\pi}\big(\ln(\Lambda/\sigma_4)+9\ln(\Lambda/\sigma_{5})\big) \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\big) \\ -\,&\frac{288L}{\pi}\Big(2\mathrm{A}_2-\mathrm{A}_3-\mathrm{A}_4\Big) +\frac{72c_2L}{\pi}\Big(\mathrm{B}_2+\mathrm{B}_3\Big) \\ +\,&\frac{9 c_2^2J_1[B]}{\pi^2} \Big(-6L_1^2-12L\theta_1^{\phantom{1}}-4L\theta_2^{\phantom{1}}+2\big(\ln(\Lambda/\sigma_4)+3\ln(\Lambda/\sigma_5)\big)L_2+6\ln^2(\Lambda/\sigma_5)\Big)\\ +\,&\hat{\kappa}_3S[B], \end{aligned}

where \hat{\kappa}_3 is a singular coefficient that does not depend on the background field.

Lemma

Lemma 8

YQ1K

Let \{\sigma_1,\sigma_2,\sigma_3,\sigma_4\} be a set of positive parameters. The numbers L_1, L_2, and \theta_2 are defined according to the formulations from Section ?. Then, for the sum of the diagrams \mathbb{H}_0^{\mathrm{sc}}(\Gamma_3^2\Gamma_4^{\phantom{1}}) the following decomposition into singular components is valid

\begin{aligned} \mathbb{H}_0^{\mathrm{sc}}&(\Gamma_3^2\Gamma_4^{\phantom{1}})\stackrel{\mathrm{s.p.}}{=} 18\ln(\Lambda/\sigma_2)\Lambda^2\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{X}_1^{\phantom{1}} \widetilde{\Gamma}_2^{\phantom{1}}\big)+ \Lambda^2 \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\widetilde{\Gamma}_2^{\phantom{1}}\big)+ \frac{2c_2(\ln(\Lambda/\sigma_4)+9\ln(\Lambda/\sigma_{5})+6L_1)}{\pi} \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\big) \\ -&18\ln(\Lambda/\sigma_2)\mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{X}_1^{\phantom{1}}\Gamma_4^{\phantom{1}}\big)+ 3\ln(\Lambda/\sigma_3) \mathbb{H}_0^{\mathrm{sc}}\big(\Gamma_3^2\mathrm{X}_1^{\phantom{1}}\big) +54\ln(\Lambda/\sigma_2)\ln(\Lambda/\sigma_3) \mathbb{H}_0^{\mathrm{sc}}\big(\mathrm{X}_1^{2}\big) \\ -& \frac{144L}{\pi}\Big(\mathrm{A}_1+2\mathrm{A}_2-2\mathrm{A}_3-\mathrm{A}_4\Big)+ \frac{72c_2L}{\pi}\Big(\mathrm{B}_2+\mathrm{B}_3\Big) \\ +&\frac{18 c_2^2J_1[B]}{\pi^2} \Big(-3L_1^2-6L\theta_1^{\phantom{1}}-6L\theta_2^{\phantom{1}}+2L_1\ln(\Lambda/\sigma_2)+\big(\ln(\Lambda/\sigma_4)+3\ln(\Lambda/\sigma_5)\big)L_2+3\ln^2(\Lambda/\sigma_5)\Big)\\+& c_2\frac{18(L_1^2+2L\theta_1^{\phantom{1}}-2\ln(\Lambda/\sigma_2)L_2)}{\pi^2}J_2[B]- \frac{36(L_1^2+2L\theta_1^{\phantom{1}}+2L\theta_2^{\phantom{1}})}{\pi^2}J_3[B] +(\hat{\kappa}_1+\hat{\kappa}_3)S[B], \end{aligned}

where \hat{\kappa}_1 and \hat{\kappa}_3 are singular coefficients by the parameter \Lambda that are independent of the background field.

Lemma

Lemma 9

IYZF

Taking into account all the above, the following decomposition is true for the first part of (?)

\begin{aligned} \mathrm{H}_{44}^{1}\stackrel{\mathrm{s.p.}}{=}& -\Lambda^4\mathbb{H}_0^{\mathrm{sc}}(\widetilde{\Gamma}_2^2)+ 2\Lambda^2\mathbb{H}_0^{\mathrm{sc}}(\Gamma_4\widetilde{\Gamma}_2) -6L_1\Lambda^2\mathbb{H}_0^{\mathrm{sc}}\big(\widetilde{\Gamma}_2\mathrm{X}_1\big) -9L_1^2\mathbb{H}_0^{\mathrm{sc}}(\mathrm{X}_1^2)+ 6L_1\mathbb{H}_0^{\mathrm{sc}}(\Gamma_4\mathrm{X}_1) \\&+2^5\Lambda^2\alpha_6(\mathbf{f}) \bigg(J_6[B]+J_5[B]\frac{c_2^2L_1}{\pi}\bigg) \\& +\frac{L}{\pi}\Big(32\mathrm{A}_2-40\mathrm{A}_3+8\mathrm{A}_4 -80\mathrm{A}_5 -80\mathrm{A}_6 +80\mathrm{A}_7 +80\mathrm{A}_8\Big) \\& +\frac{c_2L}{\pi}\Big(8\mathrm{B}_3-72\mathrm{B}_4-72\mathrm{B}_5+192\mathrm{B}_6-192\mathrm{B}_7\Big) \\& +\frac{6c_2^2}{\pi^2}J_1[B] L_1\big(4\theta_2+2L_1\big) -\frac{12c_2^2}{\pi^2}J_4[B]L_1\big(4\theta_2-2\theta_1\big) \\&+\frac{8Lc_2\big(2\theta_2+\theta_3\big)}{\pi^2}J_2[B]+\Big(2^6\alpha_5(\mathbf{f})\alpha_6(\mathbf{f})L-8L\theta_2/\pi^2\Big)J_3[B] +\hat{\kappa}_4S[B], \end{aligned}

where \hat{\kappa}_4 is a singular coefficient that does not depend on the background field.

Lemma

Lemma 10

T6ZC

Taking into account all of the above, the following decomposition is true

\begin{aligned} \mathrm{H}_{44}^2\stackrel{\mathrm{s.p.}}{=}&\,\frac{L}{\pi}\Big( 16\mathrm{A}_1 +8\mathrm{A}_3 -24\mathrm{A}_4 +120\mathrm{A}_5 +40\mathrm{A}_6 -80\mathrm{A}_7 -80\mathrm{A}_8 \Big) \\+&\frac{Lc_2}{\pi}\Big( 16\mathrm{B}_1 +8\mathrm{B}_2 +72\mathrm{B}_4 +92\mathrm{B}_5 +256\mathrm{B}_6 -256\mathrm{B}_7 \Big) \\+& 16\Lambda^2\alpha_6\Big(c_2J_7+4J_6+\frac{3L_1c_2^2}{\pi}J_5\Big)+\frac{48c_2^2}{\pi}\Lambda^2J_5(-\mathbf{f}(0)/2+\alpha_7-4\alpha_{10}) \\+& \frac{Lc_2^2J_1}{\pi^2}\Big(2\alpha_9+4\theta_3+2\theta_2-14\theta_1+26\alpha_{11}\Big) \\+& \frac{c_2J_2}{\pi^2}\Big(10L_1^2+L\big[32\pi^2\alpha_5\alpha_6+2\alpha_8-4\alpha_9+26\theta_3-10\theta_2+20\theta_1+14\alpha_{11}\big]\Big) \\+& \frac{J_3}{\pi^2}\Big(4L_1^2+L\big[128\pi^2\alpha_5\alpha_6+8\alpha_8-16\alpha_9-4\theta_3-4\theta_2+8\theta_1-6\alpha_{11}\big]\Big) \\+& \frac{Lc_2^2J_4}{\pi^2}\Big(10\alpha_8-20\alpha_9-12\theta_2+28\theta_1+44\alpha_{11}\Big) +\hat{\kappa}_{1,1}S[B]+\hat{\kappa}_{2,1}, \end{aligned}

where \hat{\kappa}_{1,1} and \hat{\kappa}_{2,1} are singular coefficients that do not depend on the background field.

Lemma

Lemma 11

UHKU

Taking into account all the above, for the sum of the diagrams \mathbb{H}_0^{\mathrm{sc}}(\Gamma_4\Gamma_4) the following asymptotic expansion is valid

\begin{aligned} \mathbb{H}_0^{\mathrm{sc}}(\Gamma_4^2)\stackrel{\mathrm{s.p.}}{=}& -\Lambda^4\mathbb{H}_0^{\mathrm{sc}}(\widetilde{\Gamma}_2^2)+ 2\Lambda^2\mathbb{H}_0^{\mathrm{sc}}(\Gamma_4\widetilde{\Gamma}_2) -6L_1\Lambda^2\mathbb{H}_0^{\mathrm{sc}}\big(\widetilde{\Gamma}_2\mathrm{X}_1\big) -9L_1^2\mathbb{H}_0^{\mathrm{sc}}(\mathrm{X}_1^2)+ 6L_1\mathbb{H}_0^{\mathrm{sc}}(\Gamma_4\mathrm{X}_1) \\&+8\alpha_6\Lambda^2\mathbb{H}_0^{\mathrm{sc}}(\widetilde{\Gamma}_4)+\frac{48c_2^2}{\pi}\Lambda^2J_5(-\mathbf{f}(0)/2+\alpha_7-4\alpha_{10}) \\& +\frac{16L}{\pi}\Big(\mathrm{A}_1+2\mathrm{A}_2-2\mathrm{A}_3-\mathrm{A}_4\Big) +\frac{c_2L}{\pi}\Big(16\mathrm{B}_1 +8\mathrm{B}_2+8\mathrm{B}_3+448\mathrm{B}_6-448\mathrm{B}_7\Big) \\&+ \frac{c_2^2J_1}{\pi^2}\Big(12L_1^2+L\big[2\alpha_9+4\theta_3+26\theta_2-14\theta_1+26\alpha_{11}\big]\Big) \\&+ \frac{c_2J_2}{\pi^2}\Big(10L_1^2+L\big[2\alpha_8-4\alpha_9+34\theta_3+6\theta_2+20\theta_1+14\alpha_{11}\big]\Big) \\&+ \frac{J_3}{\pi^2}\Big(4L_1^2+L\big[8\alpha_8-16\alpha_9-4\theta_3-12\theta_2+8\theta_1-6\alpha_{11}\big]\Big) \\&+ \frac{Lc_2^2J_4}{\pi^2}\Big(10\alpha_8-20\alpha_9-60\theta_2+52\theta_1+44\alpha_{11}\Big) +(\hat{\kappa}_4+\hat{\kappa}_5)S[B]+\hat{\kappa}_6, \end{aligned}

where \{\hat{\kappa}_4,\hat{\kappa}_5,\hat{\kappa}_6\} are singular coefficients and do not depend on the background field.

Lemma

Lemma 12

ISOC

Taking into account the representation from (?) and the decomposition from (?), the formula is valid

\begin{gathered} \mathbb{H}_0^{\mathrm{sc}}(\Gamma_4)=\Lambda^2\mathbb{H}_0^{\mathrm{sc}}(\widetilde{\Gamma}_2) +12\Big(\mathrm{B}_6-\mathrm{B}_7\Big)+\frac{3L_1c_2}{2\pi}J_1[B]+\frac{\theta_2}{\pi}J_2[B]+\\+S[B]\bigg(-\frac{3L_1^2c_2^2}{8\pi^2}+\frac{c_2^2L_1\theta_2}{2\pi^2}\bigg) +\hat{\kappa}_7+o\big(1/\Lambda\big), \end{gathered}

where the coefficient \hat{\kappa}_7 is singular and independent of the background field.

Lemma

Lemma 13

92XD

Taking into account all of the above, the following decomposition is true

\begin{aligned} \mathbb{H}_0^{\mathrm{sc}}(\Gamma_6)\stackrel{\mathrm{s.p.}}{=} -&\frac{\Lambda^2\alpha_1(\mathbf{f})}{2\pi}\mathbb{H}_0^{\mathrm{sc}}(\widetilde{\Gamma}_4) +\frac{L}{\pi}\Big( -5\mathrm{A}_1 -10\mathrm{A}_2 +10\mathrm{A}_3 +5\mathrm{A}_4 \Big) \\+&\frac{Lc_2}{\pi}\Big( \frac{5}{2}\mathrm{B}_1 +5\mathrm{B}_2 +5\mathrm{B}_3 -50\mathrm{B}_6 +50\mathrm{B}_7 \Big) \\+& \frac{L_1^2}{\pi^2}\Big(-c_2^2\frac{15J_1}{2}+c_2\frac{5J_2}{8}-\frac{5J_3}{4}\Big) \\+& \frac{L\theta_2}{\pi^2}\Big(-4c_2J_2+J_3\Big)+\hat{\kappa}_8S[B]+\hat{\kappa}_9, \end{aligned}

where \hat{\kappa}_8 and \hat{\kappa}_9 are singular coefficients and do not depend on the background field.

Lemma

Lemma 14

8M95

Let v_1(\cdot) and v_2(\cdot) be continuous functions on \mathbb{R}^2 with the Hölder exponent(Note that the introduction of the Hölder exponent is related to the fact that the functions in the work are not smooth and may have logarithmic behavior. In the statements below, the Hölder index is introduced for the same reasons.) \alpha>0, see [He-1], decreasing well enough at infinity (such that the integrals below converge). Then, taking into account the explicit form of regularization (?), the following asymptotic expansions are valid for large values of the regularizing parameter \Lambda

\int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1(x)v_2(y)\Big(\partial_{x^{\nu}}G^{\Lambda,\mathbf{f}}(x-y)\Big)\Big(\partial_{y^{\mu}}G^{\Lambda,\mathbf{f}}(x-y)\Big)\stackrel{\mathrm{s.p.}}{=} -\frac{L}{4\pi}\int_{\mathbb{R}^{2}}\mathrm{d}^2x\,v_1(x)v_2(x)\delta_{\nu\mu}+\mathcal{O}(1),

\int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1(x)v_2(y)\Big(\partial_{x^{\nu}}\partial_{y^{\mu}}G^{\Lambda,\mathbf{f}}(x-y)\Big)G^{\Lambda,\mathbf{f}}(x-y) \stackrel{\mathrm{s.p.}}{=} \frac{L}{4\pi}\int_{\mathbb{R}^{2}}\mathrm{d}^2x\,v_1(x)v_2(x)\delta_{\nu\mu}+\mathcal{O}(1).

Used by Proof

Proof of Lemma 14

Note that the integration domain \mathbb{R}^2\times\mathbb{R}^2 can be replaced by \mathrm{B}_{1/\sigma}(y)\times\mathbb{R}^2, where \sigma is the mentioned above parameter. It is a finite fixed value, which, neglecting the finite integral, does not contain singularities with respect to the regularizing parameter \Lambda. In the remaining integral, we decompose one more time, then replace v_1(x)\to v_1(y), again discarding the final parts, and also use a chain of relations

\int_{\mathrm{B}_{1/\sigma}(y)}\mathrm{d}^2y\,\Big(\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x-y)\Big) \Big(\partial_{y^\mu}G^{\Lambda,\mathbf{f}}(x-y)\Big)\stackrel{\mathrm{s.p.}}{=} \frac{-1}{4\pi^2}\int_{1/\Lambda\leqslant|x|\leqslant1/\sigma} \mathrm{d}^2x\,\frac{x^\nu x^\mu}{|x|^4}\stackrel{\mathrm{s.p.}}{=}- \frac{L}{4\pi}\delta_{\mu\nu},

where the replacement of the variable x\to x+y was additionally used. This leads to the stated result. The second integral is analyzed similarly.

Lemma

Lemma 15

SR1K

Let v_1^{ab}(\cdot) and v_2^{ab}(\cdot) be matrix-valued continuously differentiable functions on \mathbb{R}^2, decreasing well enough at infinity (such that the integrals below converge). Additionally, we require that the first partial derivatives have the Hölder exponent \alpha>0, see [He-1]. Then, taking into account the explicit form of regularization (?), the following asymptotic expansions are valid for large values of the regularizing parameter \Lambda

\begin{gathered} \int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(x) \Big(\partial_{y^\mu}G_\Lambda^{bc}(x,y)\Big) \Big(\partial_{x_\nu}D_\nu^{ad}(x)G_\Lambda^{de}(x,y)\Big) v_2^{ec}(y) \stackrel{\mathrm{s.p.}}{=}\\\stackrel{\mathrm{s.p.}}{=} \frac{L}{\pi} \int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(y)B_\mu^{bc}(y)v_2^{ac}(y)+\mathcal{O}(1), \end{gathered}

\begin{gathered} \int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(x) \Big(G_\Lambda^{bg}(x,y)B_\mu^{gc}(y)\Big) \Big(\partial_{x_\nu}D_\nu^{ad}(x)G_\Lambda^{de}(x,y)\Big) v_2^{ec}(y) \stackrel{\mathrm{s.p.}}{=}\\\stackrel{\mathrm{s.p.}}{=} -\frac{2L}{\pi} \int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(y)B_\mu^{bc}(y)v_2^{ac}(y)+\mathcal{O}(1), \end{gathered}

\begin{gathered} \int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(x) \Big(G_\Lambda^{bg}(x,y)\overleftarrow{D}_\mu^{gc}(y)\Big) \Big(\partial_{x_\nu}D_\nu^{ad}(x)G_\Lambda^{de}(x,y)\Big) v_2^{ec}(y) \stackrel{\mathrm{s.p.}}{=}\\\stackrel{\mathrm{s.p.}}{=} -\frac{L}{\pi} \int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(y)B_\mu^{bc}(y)v_2^{ac}(y)+\mathcal{O}(1). \end{gathered}

In this case, the operator \partial_{x_\nu}D_\nu^{ad}(x) in the left parts can be replaced by aither D_\nu^{ad}(x)\partial_{x_\nu} or -2A^{ad}(x).

Proof of Lemma 15

Consider the left hand side of the first relation. Taking into account the expansions near the diagonal (?) and (?), only the following three terms can have a singular contribution

4\int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(x) \Big(\partial_{y^\mu}G^{\Lambda,\mathbf{f}}(x-y)\Big) \Big(\partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x-y)\Big) v_2^{ab}(y),

4\int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(x) \Big(\partial_{y^\mu}\eta_1^{bc}(x,y)\Big) \Big(\partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x-y)\Big) v_2^{ac}(y).

The first two are convergent integrals when regularization is removed and do not contain a singular component, since the following relations are fulfilled

\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2x\, \Big(\partial_{x^\mu}G^{\Lambda,\mathbf{f}}(x)\Big) \Big(\partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x)\Big)=0,

\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2x\, \Big(\partial_{x^\mu}G^{\Lambda,\mathbf{f}}(x)\Big) \bigg(-\partial_{x_\nu}\partial_{x^\nu} \int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2z\, G^{\Lambda,\mathbf{f}}(x-z)\partial_{z^\rho}G^{\Lambda,\mathbf{f}}(z) -\partial_{x^\rho}G^{\Lambda,\mathbf{f}}(x)\bigg)\stackrel{\mathrm{s.p.}}{=}0.

Note that the latter equality is proved by integration by parts. Indeed, using spherical averaging and the fact that the support of the function \partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x) is located in the ball \mathrm{B}_{1/\Lambda}, we get

\begin{aligned} \eqref{33-r-6}&\stackrel{\mathrm{s.p.}}{=}\frac{\delta_{\mu\rho}}{2} \int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2x\, \Big(\partial_{x_\eta}\partial_{x^\eta}G^{\Lambda,\mathbf{f}}(x)\Big) \bigg(\partial_{x_\nu}\partial_{x^\nu} \int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2z\, G^{\Lambda,\mathbf{f}}(x-z)G^{\Lambda,\mathbf{f}}(z) +G^{\Lambda,\mathbf{f}}(x)\bigg)\\&\stackrel{\mathrm{s.p.}}{=} \frac{\delta_{\mu\rho}}{2} \int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2x\, \Big(\partial_{x_\eta}\partial_{x^\eta}G^{\Lambda,\mathbf{f}}(x)\Big) \bigg( \int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2z\, \Big(\partial_{z_\eta}\partial_{z^\eta}G^{\Lambda,\mathbf{f}}(z)\Big)G^{\Lambda,\mathbf{f}}(x-z) +\frac{\ln(\Lambda/\sigma)}{2\pi}\bigg)\\&\stackrel{\mathrm{s.p.}}{=} \frac{\delta_{\mu\rho}}{2}\frac{\ln(\Lambda/\sigma)}{2\pi} \int_{\mathrm{B}_{1}}\mathrm{d}^2x\, \Big(\partial_{x_\eta}\partial_{x^\eta}G^{1,\mathbf{f}}(x)\Big) \bigg( \int_{\mathrm{B}_{1}}\mathrm{d}^2z\, \Big(\partial_{z_\eta}\partial_{z^\eta}G^{1,\mathbf{f}}(z)\Big) +1\bigg)=0 . \end{aligned}

Thus, only the third part of (Proof) is of interest. Let us rewrite it in more explicit equivalent terms, taking into account factorization after shifting the variable x\to x+y

4\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2x\, \big(\partial_{x^\mu}\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2z\, G^{\Lambda,\mathbf{f}}(x-z)\partial_{z_\rho}G^{\Lambda,\mathbf{f}}(z)\bigg) \Big(\partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x)\Big) \bigg(\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(y)B_\rho^{bc}(y)v_2^{ac}(y)\bigg).

Next, integrating by parts, using spherical averaging, and again taking into account the size of the function support \partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x), we obtain

2\int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2x\, \bigg(\int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2z\, G^{\Lambda,\mathbf{f}}(x-z)\partial_{z_\rho}\partial_{z^\rho}G^{\Lambda,\mathbf{f}}(z)\bigg) \Big(\partial_{x_\nu}\partial_{x^\nu}G^{\Lambda,\mathbf{f}}(x)\Big) \bigg(\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(y)B_\mu^{bc}(y)v_2^{ac}(y)\bigg).

Then, by scaling x\to x/\Lambda and z\to z/\Lambda and using the representation from (?), we arrive at the formula

\frac{\ln(\Lambda/\sigma)}{\pi} \bigg(\int_{\mathrm{B}_{1}}\mathrm{d}^2z\,\partial_{z_\rho}\partial_{z^\rho}G^{1,\mathbf{f}}(z)\bigg)^2 \bigg(\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(y)B_\mu^{bc}(y)v_2^{ac}(y)\bigg),

hence the stated statement. Similarly, the validity of the second relation (15) can be shown. The equality (15) is the sum of the previous two. To conclude the proof, we add that in the left hand sides the operator \partial_{x_\nu}D_\nu^{ad}(x) can be replaced by -2A^{ad}(x) from (?) or by D_\nu^{ad}(x)\partial_{x_\nu} by adding a term proportional to -\partial_{x^\nu}B_\nu^{ad}(x). Such addition does not lead to additional singularities, since the auxiliary integral is convergent.

Lemma

Lemma 16

FQ2X

Let v_1^{ab}(\cdot) and v_2^{ab}(\cdot) be matrix-valued doubly continuously differentiable functions on \mathbb{R}^2, decreasing well enough at infinity (such that the integrals below converge). Additionally, we require that the second partial derivatives have the Hölder exponent \alpha>0, see [He-1]. Then, taking into account the explicit form of regularization (?), the following asymptotic expansions are valid for large values of the regularizing parameter \Lambda

\begin{gathered} \int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(x) \Big(A^{cg}(y)G_\Lambda^{bg}(x,y)\Big) \Big(A^{ad}(x)G_\Lambda^{de}(x,y)\Big) v_2^{ec}(y) \stackrel{\mathrm{s.p.}}{=}\\\stackrel{\mathrm{s.p.}}{=}\Lambda^2 \bigg(\int_{\mathbb{R}^2}\mathrm{d}^2x\Big(A_0(x)G^{1,\mathbf{f}}(x)\Big)^2\bigg) \int_{\mathbb{R}^2}\mathrm{d}^2y\, v_1^{ab}(y) v_2^{ab}(y)+\mathcal{O}(1). \end{gathered}

Proof of Lemma 16

Let us introduce a convenient representation(It follows from formula (?). It is possible to check by substitution.) for the Green’s function of the form

G_\Lambda^{de}(x,y)=2G^{\Lambda,\mathbf{f}}(x-y)\delta^{de}- \int_{\mathbb{R}^2}\mathrm{d}^2z\,\rho_1(x-z)V^{df}(z)G_\Lambda^{fe}(z,y)+ \int_{\mathbb{R}^2}\mathrm{d}^2z\,\frac{\ln(|x-z|\sigma)}{2\pi}V^{df}(z)G_\Lambda^{fe}(z,y),

where

\rho_1(x)=G^{\Lambda,\mathbf{f}}(x)+\frac{\ln(|x|\sigma)}{2\pi}.

Note that the support of \rho_1(x) is located in the ball \mathrm{B}_{1/\Lambda}, and the function itself has a continuous first derivative in \mathbb{R}^2\setminus\{0\}. Next, we apply the operator A^{ad}(x)=A_0(x)\delta^{ad}/2+V^{ad}(x)/2 to the new representation, then we get

A^{ad}(x)G_\Lambda^{de}(x,y)=A_0(x)G^{\Lambda,\mathbf{f}}(x-y)\delta^{ae}+\rho_2^{ae}(x,y).

where

\rho_2^{ae}(x,y)=-\frac{1}{2} \int_{\mathbb{R}^2}\mathrm{d}^2z\,\Big(A_0(x)\rho_1(x-z)\Big)V^{af}(z)G_\Lambda^{fe}(z,y).

Let us show that the combinations

\int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(x) \Big(\rho_2^{cb}(y,x)\Big) \Big(\rho_2^{ae}(x,y)\Big) v_2^{ec}(y),

\int_{\mathbb{R}^2}\mathrm{d}^2x\int_{\mathbb{R}^2}\mathrm{d}^2y\,v_1^{ab}(x) \Big(A_0(x)G^{\Lambda,\mathbf{f}}(x-y)\Big) \Big(\rho_2^{ae}(x,y)\Big)v_2^{eb}(y),

do not contain singular contributions. To prove this fact, it is easiest to use the perturbative decomposition again and consider the most singular contribution. In the first case, there is only one. After shifting the variable and factoring, the interesting part of the integral can be represented as

\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2x \bigg(\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2z\,A_0(x)\rho_1(x-z)\partial_{z^\mu}G^{\Lambda,\mathbf{f}}(z)\bigg) \bigg(\int_{\mathrm{B}_{1/\sigma}}\mathrm{d}^2y\,A_0(x)\rho_1(x-y)\partial_{y^\nu}G^{\Lambda,\mathbf{f}}(y)\bigg).

After integration by parts, as well as using spherical symmetry and constraints on the support, we obtain

\frac{\delta_{\mu\nu}}{2}\int_{\mathrm{B}_{2/\Lambda}}\mathrm{d}^2x \bigg(\int_{\mathrm{B}_{3/\Lambda}}\mathrm{d}^2z\,A_0(x)\rho_1(x-z)G^{\Lambda,\mathbf{f}}(z)\bigg) \bigg(\int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2y\,A_0(x)\rho_1(x-y)A(y)G^{\Lambda,\mathbf{f}}(y)\bigg).

It is clear that the possible singular contribution is obtained after scaling the variables by substitution G^{\Lambda,\mathbf{f}}(z)\to\ln(\Lambda/\sigma)/(2\pi) and is equal to

\frac{\delta_{\mu\nu}\ln(\Lambda/\sigma)}{4\pi}\int_{\mathrm{B}_{2/\Lambda}}\mathrm{d}^2x \bigg(\int_{\mathrm{B}_{3/\Lambda}}\mathrm{d}^2z\,A_0(x)\rho_1(x-z)\bigg) \bigg(\int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2y\,A_0(x)\rho_1(x-y)A(y)G^{\Lambda,\mathbf{f}}(y)\bigg).

The last integral is zero, since

\int_{\mathrm{B}_{1/\Lambda}}\mathrm{d}^2z\,A_0(z)\rho_1(z)=0.

Thus, (Proof) does not contain singular contributions. Similar arguments are valid for the integral from (Proof). Therefore, the main part of the asymptotic decomposition of the left side of (16) can be obtained by replacing A^{ad}(x)G_\Lambda^{de}(x,y) and A^{cg}(y)G_\Lambda^{bg}(x,y) by A_0(x)G^{\Lambda,\mathbf{f}}(x-y)\delta^{ae} and A_0(x)G^{\Lambda,\mathbf{f}}(x-y)\delta^{cg}. This is what the stated statement follows from.

Source and licence

The text on this page is from arXiv:2507.05923v1 by P. V. Akacevich, A. V. Ivanov and I. V. Korenev, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2507/2507.05923.jsonl.