Log-Expansions from Combinatorial Dyson-Schwinger Equations
Olaf Kr\"uger
TL;DR
This paper establishes a detailed link between combinatorial Dyson-Schwinger equations and log-expansions of Green's functions in quantum field theory, deriving key orders from the Callan-Symanzik equation and reproducing known results in QED.
Contribution
It provides a novel derivation of log-expansions for Green's functions directly from Dyson-Schwinger equations, connecting combinatorial and analytical approaches.
Findings
Derived leading-log and next-to-leading log orders from the Callan-Symanzik equation.
Reproduced known next-to-next-to-leading log results for QED photon propagator.
Demonstrated the connection between Dyson-Schwinger equations and log-expansions.
Abstract
We give a precise connection between combinatorial Dyson-Schwinger equations and log-expansions for Green's functions in quantum field theory. The latter are triangular power series in the coupling constant and a logarithmic energy scale --- a reordering of terms as is the corresponding log-expansion. In a first part of this paper, we derive the leading-log order and the next-to-leading log orders from the Callan-Symanzik equation. In particular, only depends on the -loop -function and anomalous dimensions. For the photon propagator Green's function in quantum electrodynamics (and in a toy model, where all Feynman graphs with vertex sub-divergences are neglected), our formulas reproduce the known expressions for the next-to-next-to-leading log approximation in the…
| Electron and Positron | Photon | Interaction vertex |
|---|---|---|
| \SetScale0.5 | \SetScale0.5 | \SetScale0.5 |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Log-Expansions from Combinatorial Dyson-Schwinger Equations
Olaf Krüger
Abstract
We give a precise connection between combinatorial Dyson-Schwinger equations and log-expansions for Green’s functions in quantum field theory. The latter are triangular power series in the coupling constant and a logarithmic energy scale — a reordering of terms as is the corresponding log-expansion. In a first part of this paper, we derive the leading-log order and the next-to*(j)*-leading log orders from the Callan-Symanzik equation. In particular, only depends on the -loop -function and anomalous dimensions. For the photon propagator Green’s function in quantum electrodynamics (and in a toy model, where all Feynman graphs with vertex sub-divergences are neglected), our formulas reproduce the known expressions for the next-to-next-to-leading log approximation in the literature. In a second part of this work, we review the connection between the Callan-Symanzik equation and Dyson-Schwinger equations, i.e. fixed-point relations for the Green’s functions. Combining the arguments, our work provides a derivation of the log-expansions for Green’s functions from the corresponding Dyson-Schwinger equations.
UWThPh 2019-18
1 Prologue: Green’s Functions in Quantum Field Theory
Quantum field theory (QFT) predicts probabilities for certain particle processes. For example, initialize an experiment, where a virtual photon decays into an electron-positron pair. One cannot predict if and when the photon decays, and if it does, where the electron and positron will go in the end. However, one can put two detectors and somewhere and predict the probability, that an electron enters and a positron enters . This situation is totally different from deterministic classical mechanics, which predicts the exact time evolution for given initial conditions.
Such probabilities can be found in the following way: Each QFT is defined by a Lagrangian, which is a functional of the described particle fields. It consists of a kinetic part (terms that are quadratic in the fields) and interaction terms (of cubic or higher order in the fields) — from all these, one can read off Feynman rules in a simple way. Now, a given particle process corresponds to a diagram (called Feynman graph) that translates via these rules to the so-called Feynman amplitude. Finally, the latter translates via the LSZ formula to a probability amplitude that squares to the actual probability 111This is similar to electrodynamics, where the field is an amplitude and only the intensity can be measured..
Consider for example the experiment above. The QFT that describes this process is quantum electrodynamics (QED). The corresponding Lagrangian is a functional of the photon field , the electron and its anti-particle (i.e. the positron). It contains one kinetic term for the photon and one for the electron and positron together; as well as one interaction term, i.e. 222As a convenient toy model, we consider QED in Landau gauge. Therefore, another term in the Lagrangian vanishes [1] and the theory contains only one coupling constant. We discuss more general cases in Section 4.. Here, the coupling constant is the electric charge of the electron and is a Dirac matrix. Each process involving these particles translates to a graph that consists of edges and vertices: A photon corresponds to a wiggly line and a straight arrow-line indicates an electron or positron (depending on the direction of the arrow). An interaction between these three particles is represented by a vertex. Table 1 shows all these basic ingredients of QED Feynman graphs. The virtual photon decay described above corresponds to the vertex graph in Table 1 and Feynman rules state that the respective Feynman amplitude is proportional to the electric charge . It also depends on other factors, e.g. on the relative positions of the detectors.
Quantum mechanics now tells us that the obtained result is not exact. Indeed, one has to consider all possible ways in which the final state (an electron in and a positron in ) is achieved. For example, both particles could actually interchange another photon before entering the detectors. The respective Feynman diagram is the 1-loop graph
[TABLE]
and one has to add the resulting Feynman amplitude to the previous one before computing the actual probability. Note that the latter amplitude is of order , because by Feynman rules, each vertex constitutes one factor of . Thus, the contribution to the overall probability is small compared to the initial one. All in all, one has to consider infinitely many Feynman graphs with more and more loops and vertices, whose contributions to the final amplitude (the quantum corrections) become smaller and smaller. At some point, one may truncate the resulting Feynman amplitude when it is accurate enough.
There are two problems in this calculation: First, each loop diagram translates via Feynman rules to a divergent integral. In order to extract the correct quantum corrections, one has to regularize the integrals, i.e. to keep track of the different kinds of divergences. Then, one applies a renormalization scheme, i.e. one introduces ‘Counter terms’ into the Lagrangian, which lead to additional Feynman graphs, whose amplitudes exactly cancel the divergences. In this way, the resulting Feynman amplitudes become finite 333Another way to think about regularization and renormalization is the following: The infinities are absorbed by the fields and coupling constants in the Lagrangian. In this way, they become unphysical, but the resulting probability amplitudes remain finite and are physically observable.. A second problem are infrared divergences that occur when the probability of a particle process is computed from the Feynman amplitude. We do not discuss these any further, because our work is focused on the computation of Feynman amplitudes.
After regularization and renormalization in a momentum subtraction scheme, each Feynman graph contributes a quantum correction to the initial Feynman amplitude. This correction depends on the coupling constants and the scalar products between the momenta of the particles that enter the process. Here, it is very convenient to factor out an energy scale and to define dimensionless scattering angles as . Now, renormalization requires some sort of boundary condition. In our case, we assume that is known for certain values of and . One usually calls renormalization scale and renormalization point 444In quantum chromodynamics (QCD), there are other kinds of boundary conditions. However, this does not alter the structure of that is discussed in the following.. Then, it turns out that is a polynomial in the coupling constants and the logarithm of the energy scale [2]. Hence, renormalized Feynman rules can be written as a linear map from the set of Feynman graphs to the polynomial algebra , where the dependence of on the scattering angles and is hidden in the coefficients of the polynomial. In the following, we will make this more explicit.
First, is an algebra homomorphism — the domain of definition is the algebra of all one-particle irreducible (1PI) Feynman loop graphs 5551PI Feynman graphs remain connected when one internal edge is removed. For example, \SetScale{0.5}\begin{picture}(40.0,18.0)(0.0,15.0)\Photon(0,36)(20,36){3.3}{2.5} \Vertex(20,36){1.7} \Photon(60,36)(80,36){3.3}{2.5} \Vertex(60,36){1.7} \LongArrowArcn(40,36)(20,35,125) \LongArrowArcn(40,36)(20,125,215) \LongArrowArcn(40,36)(20,215,305) \LongArrowArcn(40,36)(20,305,395) \Vertex(40,16){1.7} \Vertex(40,56){1.7} \Photon(40,16)(40,56){3.3}{4} \end{picture} is 1PI, while \SetScale{0.5}\begin{picture}(70.0,18.0)(0.0,15.0)\Photon(0,36)(20,36){3.3}{2.5} \Photon(60,36)(80,36){3.3}{2.5} \Photon(120,36)(140,36){3.3}{2.5} \Vertex(20,36){1.7} \Vertex(60,36){1.7} \LongArrowArcn(40,36)(20,80,260) \LongArrowArcn(40,36)(20,260,440) \Vertex(80,36){1.7} \Vertex(120,36){1.7} \LongArrowArcn(100,36)(20,35,125) \LongArrowArcn(100,36)(20,125,215) \LongArrowArcn(100,36)(20,215,305) \LongArrowArcn(100,36)(20,305,395) \Vertex(100,16){1.7} \Vertex(100,56){1.7} \Photon(100,16)(100,56){3.3}{4} \end{picture} is not. The latter is called one-particle reducible (1PR).
(of a given QFT) and their disjoint unions (The disjoint union is an associative product and the empty graph is the corresponding unit element). Furthermore, Feynman rules are normalized such that a 1PI Feynman graph contributes a change of relative to the corresponding zero-loop Feynman amplitude 666In this way, is a scalar, although the corresponding amplitude may be of a tensorial structure (e.g. it could be proportional to a Dirac matrix ).. This implies that 1PI graphs without loops would map to (the unit element in ). Note that these are not in but are identified with the empty graph . For example, the QED graphs in Table 1 would map to by renormalized Feynman rules 777There is no problem, that three different graphs are identified with the same element . All of them are mapped to , so why distinguish? and the 1-loop graph in Eq. (1.1) maps to a term that is proportional to (and not ). With this normalization, a one-particle reducible graph (1PR) corresponds to the disjoint union of its 1PI parts, e.g.
[TABLE]
Secondly, the algebra acquires a grading. In general, it is graded by the number of vertices of each kind in a graph. However, for QFT’s with only one interaction term, it is more convenient to grade the algebra by the number of loops,
[TABLE]
For example, , the graph in Eq. (1.1) belongs to and the disjoint union in Eq. (1.2) is an element in etc. In this case, there exists a redefinition of the coupling constant as , such that renormalized Feynman rules map each -loop graph to a term proportional to . We show this explicitly in Section 3.1. For a vertex with three outgoing edges (e.g. QED), one finds , whereas for a vertex with four outgoing edges (e.g. in theory), . For simplicity, we state our results using this redefinition. The generalization to QFT’s with more than one coupling constant is presented in Section 4.
All in all, renormalized Feynman rules are an algebra homomorphism
[TABLE]
and the quantum corrections of a certain particle process are found by applying to an infinite sum of graphs in (1PI graphs and disjoint unions thereof). This results in a series expansion in the coupling constant and the logarithm , where the dependence on the scattering angles and is hidden in the coefficients.
With this definition of renormalized Feynman rules, we will shortly explain the notion of Green’s functions, for which we need some more notations: The residue of a Feynman graph is its external leg structure. For example, the graph in Eq. (1.1) has residue r=\SetScale{0.5}\begin{picture}(15.0,0.0)(0.0,6.0)\Photon(0,18)(16,18){2}{2.5} \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture} and those in Eq. (1.2) have residue r=\SetScale{0.5}\begin{picture}(11.0,0.0)(0.0,6.0)\Photon(0,18)(20,18){2}{3.5}\end{picture}. Furthermore, its degree denotes the number of external edges — Feynman diagrams with are called ‘propagator graphs’ and those with ‘vertex graphs’. Finally, we define , which is for vertex- and for propagator graphs.
Now, in order to compute the exact Feynman amplitude of a certain particle process, one has to find the Green’s functions of the respective QFT. These encode all necessary information about the full quantum corrections of all particle processes under consideration. For example, consider another QED experiment: A source emits an electron, which is absorbed at a later time by a detector . The simplest way for the electron would be to go straight from to . Then, the quantum corrections to the corresponding Feynman amplitude are obtained by applying renormalized Feynman rules to the sum of all Feynman diagrams with residue r=\SetScale{0.5}\begin{picture}(11.0,0.0)(0.0,6.0)\ArrowLine(0,18)(20,18)\end{picture}. We denote this sum by an ellipse. It is given by
[TABLE]
where the circle denotes the sum of all 1PI loop-graphs. Thus, thanks to the geometric series,
[TABLE]
This works for all propagator graphs. Hence, let
[TABLE]
Then, the corresponding propagator Green’s function is defined as . It is the inverse of the quantum corrections to the respective 1-particle process (the particle goes from to ).
For interaction processes, the situation is similar, but there is no geometric series to compute. Therefore, one simply defines
[TABLE]
and calls with the vertex Green’s function of residue . Knowing these functions allows to compute the correct Feynman amplitude of any interaction process that is described by the theory. Let us illustrate this again at the example of QED. First, consider the photon decay described at the beginning. The full quantum corrections are given by the Green’s functions as follows:
[TABLE]
Here, the circle on the vertex denotes the sum of all 1PI graphs including the zero-loop graph, in contrast to the ‘propagator circles’. Indeed, the argument of on the lhs above denotes the sum of all Feynman graphs with residue r=\SetScale{0.5}\begin{picture}(15.0,0.0)(0.0,6.0)\Photon(0,18)(16,18){2}{2.5} \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture}. Secondly, consider a scattering between two electrons. The corresponding Feynman graphs have four external legs and in this case, the zero-loop graph
[TABLE]
is not 1PI. But again, the full quantum corrections to the corresponding Feynman amplitude are given by the Green’s functions:
[TABLE]
Note that the argument of is the sum of all Feynman graphs with residue \SetScale{0.5}\begin{picture}(15.0,0.0)(0.0,6.0)\ArrowLine(16,18)(2,32) \ArrowLine(2,4)(16,18) \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture}. As before, the circle on the vertex with four outgoing edges denotes the sum of all corresponding 1PI graphs. However, it does not contain the respective zero-loop Feynman graph, because it does not exist 888A comment on the notation: Circles denote the sum of all 1PI graphs with the respective residue. For propagators, the zero-loop graph is excluded, while for vertices, it is included (if it exists)..
Let us summarize: One can compute the Feynman amplitude of any particle process when the Green’s functions of the corresponding QFT are known. These are a power series in the coupling constant and the logarithm of the energy scale . Hence, . But renormalized Feynman rules tell us slightly more, namely that a -loop graph maps to a polynomial of degree smaller or equal than in . Therefore, can be written as a triangular power series,
[TABLE]
where is a degree- polynomial that encodes the contributions of all -loop graphs to 999The is a convention that is related to the different signs in the definitions of in Eqs. (1.5) and (1.6).. In particular, these polynomials vanish at the renormalization scale (for ), such that
[TABLE]
The Green’s functions are infinite power series and the best one can do is to approximate them perturbatively. One possibility is to truncate the sum in Eq. (1.9) at some loop-order. Since , this is a reasonable approach, at least for energies near the renormalization scale . However, the perturbative computation breaks down at energies far away from , where . Then, it is better to reorder the terms in Eq. (1.9) as follows:
[TABLE]
This is called the log-expansion for the Green’s function . is the leading log order (LLO), is the next-to-leading log order (NLLO) and in general, is the next-to(j)-leading log order (N*(j)*LLO). A truncation of the above sum, such that terms of order are neglected is called the next-to(n)-leading log (N(n)LL) approximation of . Perturbatively, this gives accurate results, as long as the functions are regular at , which may well be the case for energies even far away from the renormalization scale.
In this work, we present the following results:
- •
First, we derive the N*(j)*LL approximation of the Green’s functions in Eq. (1.10) with 101010These are the propagator Green’s functions and the vertex Green’s function that belongs to the interaction term in the Lagrangian. from the Callan-Symanzik equation. The latter is a simple first-order partial differential equation for that describes its dependence on the energy scale , see Eq. (2.7). In particular, only depends on the -loop -function and anomalous dimension, which are defined in Section 2.1 and only depend on the Feynman amplitudes of at most -loop graphs. In principle, this calculation must be long known in particle physics. However, this is barely exhibited in sufficient clarity.
- •
As an application, we compute the NNLL approximation for the propagator Green’s functions in a class of QFT toy models, in which Feynman graphs with vertex sub-divergences are neglected. An example is the photon propagator Green’s function \SetScale{0.3}H^{\begin{picture}(6.0,0.0)(0.0,0.0)\Photon(0,18)(20,18){2}{3.5}\end{picture}} in QED, thanks to the Ward-Takahashi identities (see e.g. [3, 4, 5]). For these toy models, NNLL approximations have already been derived using chord diagrams [6] or the shuffle Hopf algebra of words [7]. Our results (see Eq. (2.3)) agree with [6], but differ from [7]. The reason for the mismatch is a mistake in [7], where the shuffle products of commutators were computed using a wrong algorithm.
- •
In the second part of this paper, we review the connection between the Callan-Symanzik equation and Dyson-Schwinger equations (DSE’s) [8, 9], i.e. fixed-point relations for the infinite sums of Feynman graphs in Eqs. (1.5) and (1.6). This can be seen by extending to the Hopf algebra of Feynman graphs [10, 11] and writing the DSE’s using so-called insertion operators [12], see Eq. (3.17) 111111The leading-log approximation of the Green’s functions has been already derived using DSE’s and the Hopf-algebra of words [13]. Our results agree, see the first term in Eq. (2.2)..
- •
Finally, we comment on possible extensions of our results. For the Green’s functions in Eq. (1.10) with (such as \SetScale{0.3}H^{\begin{picture}(9.0,0.0)(0.0,0.0)\ArrowLine(16,18)(2,32) \ArrowLine(2,4)(16,18) \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture}} in QED), DSE’s are barely discussed in details in the literature. When these are found, our results can be generalized in a simple way to find the corresponding log-expansions. On the other hand, there are DSE’s for Green’s functions in QFT’s with more than one interaction term [11, 14, 15, 16] — we generalize our results to such cases in Section 4.
The paper is organized as follows: In the next Section, we derive the log-expansions of the Green’s functions with for QFT’s with only one interaction term in the Lagrangian. We also give the example of the propagator Green’s functions in toy models that neglect Feynman graphs with vertex sub-divergences. The results are obtained from the Callan-Symanzik equation (2.7). In Section 3, we rederive that relation from Dyson-Schwinger equations for the infinite sums of Feynman graphs in Eqs. (1.5) and (1.6). There, we also review some known aspects about the combinatorics of Feynman graphs (we extend to a Hopf algebra and rederive the essential properties of renormalized Feynman rules ). In Section 4, we generalize our results to QFT’s with more than one interaction term in the Lagrangian and conclude in Section 5.
2 Results
In this section, we present the derivation of the log expansion in Eq. (1.10) with from the Callan-Symanzik equation (2.7), see Section 2.2. Therefore, we first introduce the anomalous dimension and the -function in Section 2.1, as they appear in the final results. Finally, Section 2.3 gives a simple example of the log-expansion for propagator Green’s functions in QFT toy models, where all Feynman graphs with vertex sub-divergences are neglected.
2.1 The Anomalous Dimensions and -Function
For each vertex residue , define
[TABLE]
Here, means that the edge is incident to the vertex . The square root must be expanded in a Taylor series, similar to the geometric series of propagator Feynman graphs. For example, in QED:
[TABLE]
The square roots are graphically represented by half-ellipses. Since renormalized Feynman rules are an algebra homomorphism, one can easily evaluate these infinite sums of Feynman graphs:
[TABLE]
In physics literature, is often referred to as the amputated Green’s function for the vertex . In more mathematical literature, these functions are called the invariant charges.
Let us shortly motivate the above definition by showing how the amputated Green’s functions appear in physics computations. For example, consider any particle interaction process that is described by the QFT. Let be the residue of the respective Feynman graphs. Then, one can easily show that the corresponding Feynman amplitude (including all quantum corrections) is given by times a polynomial of the amputated Green’s functions. For example, the righthandsides of Eqs. (1.7) and (1.8) read
[TABLE]
Secondly, it is common in particle physics to redefine the coupling constants of the theory as . This makes them dependent on the energy scale and has the advantage that a perturbative expansion for the Green’s functions in terms of and the logarithmic energy scale may behave much better. This is an alternative way to approximate the Green’s functions accurately without using the log-expansion. The are called the running coupling parameters. In our case, there is only one coupling constant and we redefined it such that counts the number of loops in a Feynman graph. Hence, we must also redefine the respective invariant charge as 121212Here, is the vertex residue that corresponds to the interaction term in the Lagrangian.
[TABLE]
Then, the running coupling parameter is . At the renormalization scale (), one has , because .
Now, the anomalous dimension is defined as the negative of the -linear part of the Green’s function ,
[TABLE]
Similarly, the -function (in our case, there is only one) is the -linear part of the invariant charge,
[TABLE]
Indeed, it follows from Eq. (2.2) that it is just a linear combination of the various anomalous dimensions. For example, in QED:
[TABLE]
Two remarks are in order: First, the notion of an anomalous dimension for vertex type residues is quite unusual in particle physics. Indeed, it is redundant, since can be expressed by the -function and the various propagator anomalous dimensions via Eq. (2.5). In our case however, the definition in Eq. (2.4) including vertex type residues is very convenient because the formulas for the log-expansions of the Green’s functions generalize to all kinds of residues. Secondly, the minus sign in Eq. (2.4) is chosen such that it cancels the sign in Eq. (1.5) for propagator type residues . In this case, the anomalous dimension is given by the -linear part of the Feynman amplitude that corresponds to the sum of all respective 1PI propagator graphs.
The functions and can be computed perturbatively. Neglecting terms of order in Eqs. (2.4) and (2.5) gives the so-called -loop anomalous dimensions and -loop -function. These approximations can be obtained by only evaluating Feynman graphs with at most loops. It turns out that the N*(n)*LL approximation of the Green’s functions only depend on the -loop anomalous dimensions and -function. We show this explicitly in the next section.
2.2 Log-Expansions
We start with the following formula for the Green’s functions:
[TABLE]
A detailed derivation from combinatorial Dyson-Schwinger equations is collected in Section 3 (for a short proof, see Section 3.4). Taking a linear combination of the above relations for different residues implies an ordinary differential equation for the invariant charge, i.e.
[TABLE]
Together with the initial conditions
[TABLE]
Eqs. (2.7) and (2.8) determine the full -dependence of the Green’s functions 131313Note that this only requires to know the anomalous dimensions and -function of the QFT, i.e. to say the -linear parts of the Green’s functions..
In particle physics, the above relations are usually written in terms of the (-dependent) running coupling parameter . Then, Eq. (2.8) corresponds to the renormalization group equation (RGE) 141414The usual definition of the -function in the literature slightly differs from ours:
E.g., the QED 1-loop -function is . The notations in this paper are consistent, such that our 1-loop -function is linear in the (loop-counting) coupling parameter. and Eq. (2.7) represents the Callan-Symanzik equation for the Green’s function with :
[TABLE]
The derivation of the log-expansions for the Green’s functions requires another differential equation for the invariant charge . Therefore, one integrates Eq. (2.8) using separation of variables. With the above definition of the running coupling parameter, this leads to
[TABLE]
Here, the integration constant is well chosen such that for . Since we cannot solve this for , it is more convenient to take the derivative with respect to . This results in
[TABLE]
With these considerations, we can now give the main equations that determine the log-expansions for the Green’s functions. One therefore defines a log-expansion for the invariant charge 151515Accordingly, the log-expansion for the running coupling is given by .,
[TABLE]
and a change of variables in Eqs. (2.7 - 2.10) finally results in
[TABLE]
Now, one uses Eqs. (2.13) and (2.14) in order to find the N*(n)LL approximation of the invariant charge 161616This is equivalent to solving the RGE and implies a log-expansion for the running coupling parameter , see e.g. [17].. The obtained result can then be used in Eq. (2.12) to deduce the N(n)*LL approximation of the Green’s functions .
The first step is simply achieved by a Taylor series expansion of Eq. (2.14) in . The [math]th order reads
[TABLE]
which gives the LLO of the invariant charge. Now, consider all terms of order in Eq. (2.14) with and note that they do not contain any of the functions with . Hence, collecting all terms including the function , one finds
[TABLE]
where the rhs only consists of the functions . This is a nice recursive formula for the various log orders of the invariant charge, except for . Two remarks are in order: First, the -linear terms in Eq. (2.14) vanish completely, because an -derivative was taken in order to obtain Eq. (2.10). However, can be found from the -linear part of Eq. (2.13), which reads
[TABLE]
The solution of this ordinary first-order differential equation with the respective initial condition is given by
[TABLE]
Secondly, an explicit form of the recursive relation for is not required, because Eq. (2.14) with cleared fractions can be given directly to a computer algebra program in order to obtain the various log orders. For example, the order in Eq. (2.14) reads
[TABLE]
when Eqs. (2.15) and (2.16) are used.
The second step (the derivation of the log-expansion for the Green’s functions ) is a bit more involved, because one has to integrate with respect to . Indeed, from Eq. (2.12),
[TABLE]
In order to obtain the N*(n)*LL approximation of , one needs to expand the exponent up to order . As promised earlier, this requires knowledge of the -loop anomalous dimension and -function. The computation is quite technical but can always be done with the help of a computer algebra program. Here, we contentedly give the NNLL approximation, i.e.
[TABLE]
2.3 Examples: Truncated Propagator Green’s Functions and QED
Let us give a short example of the above formula (2.2). In QED, the Ward-Takahashi identities state that the Green’s functions for the fermion propagator \SetScale{0.3}G^{\begin{picture}(6.0,0.0)(0.0,0.0)\ArrowLine(0,18)(20,18)\end{picture}} and the 3-point vertex \SetScale{0.3}G^{\begin{picture}(9.0,0.0)(0.0,0.0)\Photon(0,18)(16,18){2}{2.5} \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture}} coincide. Therefore, the invariant charge in Eq. (2.3) is given by the inverse of the photon propagator Green’s function. Furthermore, the -function equals the respective anomalous dimension, see Eq. (2.6):
[TABLE]
The above relation is a special case of a broader class of QFT toy models that describe only one particle (edge-type in Feynman graphs), where the invariant charge and -function are given by
[TABLE]
for integers . Examples are in QED, as well as () for a truncated Yukawa theory (-theory), where all vertex divergences are neglected. In all these cases, there is only one propagator Green’s function with log-expansion and it is given by 171717Note that this is consistent with Eqs. (2.13, 2.18 and 2.20).. The NNLL approximation can be obtained in two ways: Either by inserting the relation between the -function and the anomalous dimension into Eq. (2.2); or by using the explicit expression for the log-expansion of the invariant charge in Eqs. (2.11, 2.15 - 2.17). In both cases, we obtain 181818In [6], equivalent expressions have been derived using Chord diagrams. These formulas contain the symbols , which are related to the anomalous dimension as follows:
[TABLE]
3 Mathematical Background
This section collects the essential properties of Feynman graphs and renormalized Feynman rules (presented in a mathematical framework) that lead to the Callan-Symanzik equation (2.7). The derivation of this relation is given in Section 3.4 and requires two main ingredients: The first one is the well known exponential formula for , see Eq. (3.8). We motivate this property of renormalized Feynman rules and give a short introduction to the Hopf algebra of Feynman graphs in Section 3.2. The second ingredient is an (also well known) expression for the co-product of the infinite sums in Eqs. (1.5) and (1.6), see Eq. (3.10). We review in Section 3.3, how this relation is obtained from fixed-point equations for , i.e. so-called combinatorial Dyson-Schwinger equations (DSE’s). In the literature, the latter are written down in full detail only for those sums with 191919Hence, or is the vertex that corresponds to the interaction term in the Lagrangian., which is the reason why we restrict our results to that case. For presentation purposes, we consider QFT’s with only one interaction term in the Lagrangian (we explain the redefinition of the coupling constant in Section 3.1). However, DSE’s also exist in theories with more than one coupling parameter — the generalization to that case is given in Section 4.
3.1 The Grading of
For QFT’s with only one interaction term in the Lagrangian, one redefines the coupling constant as , which implies that a -loop graph maps to a term proportional to under renormalized Feynman rules. Let us show this now.
First, Euler’s formula states that for a graph with loops, vertices and internal edges, the following identity holds:
[TABLE]
Secondly, let be the degree of the vertex 202020i.e. the number of edges which are incident to that corresponds to the interaction term in the Lagrangian. For example, in QED and in -theory. Then, a Feynman graph with residue has external edges, hence
[TABLE]
Together with Euler’s formula and the definition of , one finally finds
[TABLE]
One concludes that with the correct normalization of renormalized Feynman rules, a graph with loops and vertices maps to a term that is proportional to .
3.2 Feynman Rules from a Hopf Algebraic Point of View
This section represents a minimal introduction to the Hopf algebra of Feynman graphs [10, 11]. As a motivation, consider a typical observation during the renormalization of Feynman amplitudes:
[TABLE]
Here, and only within this section, we write renormalized Feynman rules with an upper index . It indicates the energy scale at which Feynman graphs are evaluated. The goal of this section is to generalize the above observation (to the exponential formula). This can be achieved in a very elegant way, once we introduced the Hopf algebra of Feynman graphs.
Let us start with the algebra , i.e. the vector space equipped with an associative product (the disjoint union) and a unit element (the empty graph). In a more mathematical notation,
[TABLE]
where id denotes the identity map on .
In a first step, a co-product and a co-unit are defined, such that extends to a bi-algebra. The important tool for our purposes is the co-product. It is a linear map , which decomposes a graph into its sub-graphs in a certain way. The explicit definition of its action on 1PI Feynman graphs is given by
[TABLE]
where the sum extends over (disjoint unions of) divergent sub-graphs of . Furthermore, corresponds to but with the sub-graphs shrinked to a point. For example,
[TABLE]
On a disjoint union of Feynman graphs, is defined such that it is compatible with the product (a requirement for bi-algebras),
[TABLE]
Here, Sweedler’s notation for the co-product is used, e.g. . The co-unit maps an element to the coefficient of in . Indeed, it is easy to show that the co-product is co-associative and that is the co-unit element for ,
[TABLE]
Hence, forms a bi-algebra.
Secondly, one defines an antipode to further extend to a Hopf algebra. It is the linear map , which is recursively given by
[TABLE]
All in all, forms a Hopf algebra, called the Hopf algebra of Feynman graphs.
With the above defined structures, the vector space of algebra homomorphisms extends to a group , which is called the convolution- or character-group [18]. For example, renormalized Feynman rules are a character, . The conjunction in is the convolution product : Let . Then,
[TABLE]
is an algebra homomorphism, hence . The convolution is associative because the co-product is co-associative and the product in (multiplication) is associative. The unit element in is the linear map and the inverse of a character is given by .
The observation in Eq. (3.2) can now be generalized in a beautiful way: Renormalized Feynman rules satisfy the property 212121This is long known but never really written down in its full beauty. See e.g. [2]. Note that Eq. (3.2) is just a special case of Eq. (3.7).
[TABLE]
It follows that the vector space of maps , equipped with the convolution product, forms a group. This is the so-called renormalization group and it is isomorphic to the additive group. Hence, there exists a linear map , such that
[TABLE]
This is the exponential formula. Here, we dropped the upper index in again, in order to adapt the notation to the rest of this paper. We can give a meaning to the character : It maps a (disjoint union of) Feynman graph(s) to the -linear part of the corresponding Feynman amplitude,
[TABLE]
In particular, the infinite sums in Eqs. (1.5) and (1.6) are mapped to the negative of the anomalous dimension (see Eq. (2.4)),
[TABLE]
From now on, we do not need the co-unit and antipode of the Hopf algebra anymore. These structures were only required for a proper definition of the convolution group . The proof of Eq. (2.7) only uses the co-product and the exponential formula in Eq. (3.8).
3.3 Graph Insertion and Dyson-Schwinger Equations
In this section, we motivate the following formula for the co-product 222222This formula is well known in the literature, see e.g. [19, 20, 21, 22], or in a slightly different context, also [23, 24]. of the infinite sums in Eqs. (1.5) and (1.6) 232323These map to the Green’s functions under renormalized Feynman rules.:
[TABLE]
Here, is the combinatorial invariant charge (), see Eq. (2.3). For example in QED:
[TABLE]
Furthermore, denotes the projection onto , hence for each .
Let us start by introducing the notion of graph insertion. Therefore, consider for example the QED Feynman graphs
[TABLE]
The first one is obtained, when the sub-graph \SetScale{0.3}\begin{picture}(24.0,0.0)(0.0,7.5)\Photon(0,36)(20,36){3.3}{2.5} \Vertex(20,36){1.7} \Photon(60,36)(80,36){3.3}{2.5} \Vertex(60,36){1.7} \LongArrowArcn(40,36)(20,80,260) \LongArrowArcn(40,36)(20,260,440) \end{picture} is inserted into \SetScale{0.3}\begin{picture}(24.0,0.0)(0.0,7.5)\Photon(0,36)(32,36){3.3}{3.5} \Vertex(32,36){1.7} \Line(80,72)(32,36) \ArrowLine(64,12)(32,36) \Line(80,0)(32,36) \ArrowLine(32,36)(64,60) \Photon(64,12)(64,60){-3.3}{4.5} \Vertex(64,60){1.7}\Vertex(64,12){1.7} \end{picture}. However, the second one does not contain any sub-graph and cannot be constructed via graph insertion. Note that the co-product of a 1PI Feynman graph without sub-graphs is given by , see Eq. (3.3). Hence, by the exponential formula (3.8) and all graphs with that property are called primitive graphs. On the other hand, each non-primitive 1PI Feynman graph can be obtained via graph insertions of primitive ones.
In the following, we make this more explicit. Therefore, let be a primitive graph. Each vertex of represents an insertion place for a 1PI vertex graph. Furthermore, each internal edge of type in is an insertion place for 1PI propagator graphs with residue . Note that more than one propagator graph can be inserted into an edge, whereas only one vertex graph can be inserted into a vertex of . Now, for each primitive , define an insertion operator as a linear map , such that a disjoint union of 1PI Feynman graphs maps to
[TABLE]
Here, the sum extends over all graphs that can be obtained by insertion of into the various insertion places of . The factor is assigned such that the insertion operator commutes with the co-product in the following way 242424The are Hochschild 1-co-cycles.:
[TABLE]
The following examples should make Eq. (3.12) clear — for a more rigorous definition, see [12]:
[TABLE]
Note that by Eq. (3.3), these examples imply Eq. (3.13).
Some remarks are in order: First, inserting zero-loop graphs into does not change it. This is consistent with our realization of zero-loop graphs as the unit element . For example, and . Secondly, inserting more than vertex graphs into a primitive with vertices does not give a well defined object in . For example, there is no element in that corresponds to
[TABLE]
Finally, there are non-primitive graphs in that are not in the image of some insertion operator. For example, \SetScale{0.3}\begin{picture}(24.0,0.0)(0.0,7.5)\Photon(0,36)(32,36){3.3}{3.5} \Vertex(32,36){1.7} \ArrowLine(32,36)(80,72)\ArrowLine(80,0)(32,36) \Photon(64,12)(64,60){-3.3}{4.5} \Vertex(64,60){1.7}\Vertex(64,12){1.7} \Photon(48,24)(48,48){-3.3}{2.5} \Vertex(48,24){1.7}\Vertex(48,48){1.7} \end{picture} contains a sub-graph, but it is not in the image of B_{+}^{\SetScale{0.15}\begin{picture}(12.0,0.0)(0.0,0.0)\Photon(0,36)(32,36){3.3}{3.5} \Vertex(32,36){1.7} \Line(80,72)(32,36) \ArrowLine(64,12)(32,36) \Line(80,0)(32,36) \ArrowLine(32,36)(64,60) \Photon(64,12)(64,60){-3.3}{4.5} \Vertex(64,60){1.7}\Vertex(64,12){1.7} \end{picture}}. However, the infinite sums in Eqs. (1.5) and (1.6) consist of such linear combinations of 1PI Feynman graphs that are in the image of some insertion operator . This will be essential in the following.
The closed formula for the co-product in Eq. (3.10) can be found from fixed-point relations (DSE’s) for the infinite sums . Let us motivate this again at the QED example, where the sums of all 1PI Feynman graphs of a certain residue satisfy [25]
[TABLE]
Here, we recall the notation of the introduction: a circle represents the sum of all respective 1PI graphs and an ellipse denotes the geometric series of propagator Feynman graphs, as in Eq. (1.4) 252525Remember that the zero-loop propagator graph is contained in the ellipse, but not in the circle on an edge — it belongs to the geometric series, but not to the sum of 1PI graphs.. On the rhs, these infinite sums are inserted into the one-loop primitives
[TABLE]
where only one vertex of the propagator primitives is dressed with an infinite sum in order to avoid double counting of graphs. The dots in Eqs. (3.14) represent all graph insertions into higher-loop primitives. The QED example is somewhat special, because the propagator DSE can be formulated for the quenched limit. Then, the first two fixed-point relations in Eqs. (3.14) are actually complete.
Eqs. (3.14) translate to DSE’s for the infinite sums in Eqs. (1.5) and (1.6). These can be written in a very elegant way using the insertion operators:
[TABLE]
Note that there is no double counting of Feynman graphs, although the structure differs from the one in Eqs. (3.14). Indeed, the (negative) power of \SetScale{0.3}X^{\begin{picture}(9.0,0.0)(0.0,0.0)\Photon(0,18)(16,18){2}{2.5} \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture}} () in the argument of an insertion operator corresponds to the number of vertices (type- edges) in . Hence, in contrast to Eqs. (3.14), each insertion place is dressed by an infinite sum. Here, the insertion operators take care that there is no double counting and the deeper reason for that is the commutation relation in Eq. (3.13) [12].
With these considerations, the above relations generalize in a simple way. Therefore, denote the sum of all insertion operators for -loop primitives with residue by
[TABLE]
Then, the general DSE’s [25, 9, 24, 8] for the infinite sums in Eqs. (1.5) and (1.6) with read
[TABLE]
Indeed, the insertion places of each -loop primitive graph are fully dressed by infinite sums 262626Let us show this quickly: Consider a -loop primitive graph . The argument of the corresponding insertion operator is , where by Eqs. (2.1) and (2.3),
The power of in the argument of is for and , if . This matches the number of vertices in , given by Eq. (3.1). Secondly, the power of in the argument of for some edge-type residue is equal to
This equals the number of type- internal edges in , which completes the proof. and the insertion operators ensure that there is no double counting. Note that Eqs. (3.15) coincide with this general form, but only includes the terms.
To close this section, we show that the co-product formula in Eq. (3.10) and the commutation relation in Eq. (3.13) are consistent with the DSE in Eq. (3.17). With slightly more effort, one can turn this into a proof of Eq. (3.10) by induction. For details, see e.g. [19, 20, 21, 22]. First, let with integers be an arbitrary monomial in . Then, Eq. (3.10) implies that
[TABLE]
Now, we compute the co-product of the rhs in Eq. (3.17). Therefore, we commute with the insertion operators via Eq. (3.13) and use the above relation for the monomials:
[TABLE]
Finally, we rearrange the sums as using an index shift, which results in
[TABLE]
Note that the terms on the rhs of the sign correspond to (see Eq. (3.17)). Hence, we are left with
[TABLE]
3.4 Rederivation of the Callan-Symanzik Equation
With the considerations of the previous sections, the proof of Eq. (2.7) is simple. It requires the exponential formula (3.8) and the co-product relation (3.10). Together with the definitions of the convolution product in Eq. (3.6) and the function in Eq. (3.9), we find
[TABLE]
4 Multiple Coupling Constants
In this section, we rederive the Callan-Symanzik equation from Dyson-Schwinger equations in QFT’s with multiple interaction terms in the Lagrangian. Therefore, we assume that each such term comes with a separate coupling constant and gives rise to a certain vertex in Feynman graphs. We denote the different vertices by and collect the coupling parameters to a vector . The positive integer is given by the number of interaction terms. In the following, we summarize the main changes to the one-coupling case.
First of all, it is inappropriate to grade the Hopf algebra of Feynman graphs with respect to the loop number in this setting. Instead, is multi-graded as [11, 16, 26, 27]
[TABLE]
Here, denotes a multi-index (a vector of integers) and the sub-spaces contain the following Feynman graphs:
[TABLE]
Note that this grading respects the product and co-product of .
Secondly, the Green’s functions depend on all coupling constants and the logarithmic energy scale . In particular, as before, where the are given by the infinite sums in Eqs. (1.5) and (1.6). Hence, the corresponding anomalous dimensions are power series in all coupling parameters,
[TABLE]
Furthermore, there is one amputated Green’s function (invariant charge) for each vertex , see Eq. (2.1). These define the running coupling parameters as before. In order to simplify the notation, we write
[TABLE]
Accordingly, there is one -function for each vertex , with power series expansion
[TABLE]
Finally, the Dyson-Schwinger equation for the infinite sums in Eqs. (1.5) and (1.6) is given by
[TABLE]
where
[TABLE]
is the generalization of Eq. (3.16) 272727This relation is barely written down properly in the literature, see e.g. [16, 28]. Note that in Eq. (4.1), each insertion place of a primitive is fully dressed with an infinite sum and there is no double-counting of graphs because the insertion operators are Hochschild 1-co-cycles, i.e. they satisfy the commutation relation in Eq. (3.13). In full analogy to the end of Section 3.3, the DSE in Eq. (4.1) implies the co-product formula
[TABLE]
where \big{|}_{\mathbf{k}} is the projection onto . This is the generalization of Eq. (3.10) .
The Callan-Symanzik equation can now be derived in the same way as in Section 3.4. Therefore, the co-product formula in Eq. (3.10) is replaced by Eq. (4.2) and the action of on the infinite sums in Eq. (3.9) must be generalized to
[TABLE]
Then, using the exponential formula in Eq. (3.8) and the definition of the convolution product in Eq. (3.6) results in
[TABLE]
which can be written in the same form as in the one-coupling case, see Eq. (2.7):
[TABLE]
In particular, taking certain linear combinations of these relations yields a system of ordinary differential equations, which is the analogue to Eq. (2.8),
[TABLE]
Together with the initial conditions and , Eqs. (4.3) and (4.4) fully determine the Green’s functions, when only the anomalous dimensions and -functions are known.
Two remarks are in order. First, in terms of the running coupling parameters, Eq. (4.4) represents the usual system of renormalization group equations 282828Again, our definition of the -functions differs from the one in the literature, . and Eq. (4.3) is the usual form of the Callan-Symanzik equation for the Green’s functions with :
[TABLE]
Secondly, the log-expansions for the Green’s functions can be derived as in Section 2.2 via a change of variables (for one of the coupling parameters ). In particular, let . Then, the solution to Eq. (4.3) is given by
[TABLE]
This formula only requires the log-expansion for the running coupling parameters , which can be found log order by log order from the RGE. We hope to investigate this case further and give examples in future work.
5 Conclusions
In this paper, we established a precise connection between Dyson-Schwinger equations for the Green’s functions in a given QFT and their corresponding log-expansions. We mainly discussed QFT’s with only one interaction term in the Lagrangian (Section 2.2), but also expanded the results to the more general case (Section 4).
The formulas for the log-expansions are given in Eqs. (2.18) and (4.5) — the next-to*(n)-leading log approximation is obtained by an analytical integration that requires the solution of the renormalization group equation up to the next-to(n)*-leading log order. In particular, it only depends on the -loop -function and anomalous dimension. An explicit expression for the next-to-next-to-leading log approximation is given in Eq. (2.2), when only one coupling constant is involved. In particular, for the photon propagator Green’s function in quantum electrodynamics (and in a toy model, where all Feynman graphs with vertex sub-divergences are neglected) our formulas lead to the known expressions in the literature (Compare Eq. (2.3) with [6]).
Our results may be used for an accurate perturbative calculation of the Green’s functions, for energy scales far away from the renormalization scale . However, the log-expansions do only converge, if the coefficients do well behave perturbatively. The latter must not be true — so far, we have no control about the dependence of the Green’s functions on the scattering angles and . In order to understand this better, one needs to generalize the exponential formula in Eq. (3.8) to include also the angle dependence. Together with the co-product formulas in Eqs. (3.10) and (4.2), this would lead to a generalization of the Callan-Symanzik equation, from which the correct perturbative approach may be obtained. However, this is a highly non-trivial problem, as it requires to understand the behavior of the Green’s functions orthogonal to the renormalization group flow .
Our work gives the derivation of the log-expansions for the Green’s functions from the corresponding Dyson-Schwinger equations. However, if the latter are not known, we cannot comment on the corresponding log-expansions. For example in QED, a detailed Dyson-Schwinger equation for the 4-point function \SetScale{0.3}G^{\begin{picture}(9.0,0.0)(0.0,0.0)\ArrowLine(16,18)(2,32) \ArrowLine(2,4)(16,18) \Vertex(16,18){1.7} \ArrowLine(16,18)(30,32) \ArrowLine(30,4)(16,18) \end{picture}} in terms of the insertion operators is not explicitly written down in the literature. We plan to do that in future work — with our analysis, this would lead to the respective Callan-Symanzik equation and then, to the corresponding log-expansion.
Finally, we would like to make the case with multiple coupling parameters more explicit. Therefore, we need to solve the system of renormalization group equations in Eq. (4.4) analytically, such that Eq. (4.5) can be integrated log-order by log-order.
Acknowledgments
First of all, I want to thank Dirk Kreimer and Karen Yeats for very illuminating discussions and for their patience in answering so many questions. Furthermore, I would like to thank Simon Plätzer and Angelika Widl for giving me insights to the particle physicist point of view. Finally, I acknowledge useful discussions with Henry Kissler, David Prinz, Stefan Fredenhagen, Michael Borinsky and John Gracey.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] H. Kißler, On the gauge dependence of Quantum Electrodynamics , Po S LL 2018 (2018) 032 , [ 1808.00291 ]. · doi ↗
- 2[2] F. Brown and D. Kreimer, Angles, Scales and Parametric Renormalization , Lett. Math. Phys. 103 (2013) 933–1007 , [ 1112.1180 ]. · doi ↗
- 3[3] J. C. Ward, An Identity in Quantum Electrodynamics , Phys. Rev. 78 (1950) 182 . · doi ↗
- 4[4] H. S. Green, A Pre-renormalized quantum electrodynamics , Proc. Phys. Soc. A 66 (1953) 873–880 . · doi ↗
- 5[5] Y. Takahashi, On the generalized Ward identity , Nuovo Cim. 6 (1957) 371 . · doi ↗
- 6[6] J. Courtiel and K. Yeats, Next-to k leading log expansions by chord diagrams , 1906.05139 .
- 7[7] O. Krüger and D. Kreimer, Filtrations in Dyson–Schwinger equations: Next-to j -leading log expansions systematically , Annals Phys. 360 (2015) 293–340 , [ 1412.1657 ]. · doi ↗
- 8[8] D. Kreimer and K. Yeats, An Etude in non-linear Dyson-Schwinger Equations , Nucl. Phys. Proc. Suppl. 160 (2006) 116–121 , [ hep-th/0605096 ]. · doi ↗