On the Hopf algebra structure of perturbative quantum field theories

TL;DR

The paper casts renormalization in perturbative quantum field theories as a natural Hopf algebraic structure, where Feynman diagrams are organized into forests of subdivergences and renormalization operations coincide with the Hopf antipode. By constructing a PW-based Hopf algebra with a renormalization map $R$, it shows how the forest formula yields finite, renormalized results via $m[(S\otimes id)\Delta[X]]$, and identifies primitive diagrams as those without subdivergences. The approach is illustrated with toy models and extended to real quantum field theories, clarifying how internal and overlapping divergences factorize into nested/disjoint components. The work also discusses limits of strict Hopf structure for general $R$, the potential for quasi-Hopf generalizations, and hints at connections to knots and number theory through the algebraic organization of divergences.

Abstract

We show that the process of renormalization encapsules a Hopf algebra structure in a natural manner. This sheds light on the recently proposed connection between knots and renormalization theory.

Figures (5)

• Figure 1: A Feynman graph $\Gamma$ with subdivergences. There are various subgraphs $\gamma_i$. Ultimately, we can consider $\Gamma$ as an expression in various Feynman graphs without subdivergences, which are nested into each other and in this way built up the given graph. The dashed boxes indicate five forests, each containing (sub)-divergences. The box indicated by the number 5 contains the whole graph $\Gamma$, while boxes $1,\ldots,4$ contain subdivergences $\gamma_1,\ldots,\gamma_4$. Note that box number 4 contains the graph $\gamma_4$ which is itself built up from graphs $\gamma_1,\gamma_2$ and $\gamma_0$.
• Figure 2: We redraw the Feynman graph of the previous figure. This Feynman graph belongs to the class spanned by the word $(((\gamma_1)(\gamma_2)\gamma_0)(\gamma_3)\gamma_0)$, by the procedure explained in the text. The bracket configuration is obtained from the forest configuration above by ignoring the horizontal lines in the dashed boxes. Note that the forests $4,5$ both contain the graph $\gamma_0$ as the outermost letter on the rhs. Accordingly, if we shrink all subdivergences in these forests to a point, it is the graph $\gamma_0$ which remains.
• Figure 3: Some Feynman graphs belonging to the classes $(x)$, $(x)x)$, $((x)(x)x)$ and $(((x)x)x)$. We indicate the order in $\hbar$.
• Figure 4: Two Feynman graphs. $\Gamma_1$ belongs to the class $(((\gamma_1)\gamma_1)\gamma_1)$, it is directly factorizable as we can nullify the external momentum $q$. $\Gamma_2$ splits into a contribution in this class, plus a contribution $(X\gamma_1)$, where $X=([\gamma_{1,k-l}-\gamma_1]\gamma_1)$.
• Figure 5: Using Eq.(\ref{['sd']}) back and forwards, we obtain an easy expression for the overall divergence of an overall linearly divergent two-point function. This gurantees all necessary factorization properties and disentangles the overlapping divergences in accordance with the forest formula.