Characterization of the D'Alembertian by the Poincar\'e Invariance

Hiromichi Nakazato; Tohru Ozawa

REVIEW 1 major objections 5 minor 24 references

Characterization of the D'Alembertian by the Poincar\'e Invariance

T0 review · 1 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A characterization theorem: Poincaré-invariant linear partial differential operators in Minkowski space-time are exactly the polynomials in the d'Alembertian, and the only second-order one with dilation invariance is the free wave…

desk verdict A correct and clean but non-novel proof that Poincaré-invariant linear PDE operators are polynomials in the d'Alembertian. read the letter →

arxiv 2506.02513 v1 pith:H3GEOMS5 submitted 2025-06-03 math-ph math.MP

Hiromichi Nakazato , Tohru Ozawa This is my paper

classification math-phmath.MP MSC 35Q7535L5535E2035Q60

keywords D'AlembertianPoincaréinvarianceLorentzdilationlinearpartialdifferentialoperatorswaveoperatortransformation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proves the converse of a standard fact: any linear partial differential operator on Minkowski space-time that commutes with all Poincaré transformations (Lorentz boosts and rotations together with translations) must have constant coefficients and be a polynomial in the d'Alembertian □ = ∂_t² − Δ. As a consequence, the only second-order operators with Poincaré symmetry are α□ + β, and if the operator is also invariant under dilations in the weak sense that its zero set is preserved under scaling, the mass term β vanishes, forcing exactly the free wave equation α□u = 0. This matters because it shows the wave operator's special role in relativistic physics is not an accident: it is the unique linear second-order operator selected by the symmetry group of special relativity together with scale invariance.

What carries the argument

The central device is the symbol polynomial p(τ,ξ) = Σ a_{jα}(0) τ^j ξ^α, obtained by applying L to the exponential eigenfunctions $e^{{τt + ξ·x}}$. Invariance of L under translations forces constant coefficients, and Lorentz invariance forces p to be invariant under the Lorentz group. Writing p = Σ_{ℓ=0}^m p_ℓ with p_ℓ homogeneous of degree ℓ, the proof defines φ_ℓ(τ,ξ) = p_ℓ(τ,ξ) / |τ² − |ξ|²|^{ℓ/2} on the complement of the light cone. Because the Lorentz group acts transitively on each region of constant Minkowski norm and φ_ℓ is homogeneous of degree zero, φ_ℓ must be constant on each of the regions {τ²>|ξ|²} and {τ²<|ξ|²}; comparing the two constants then shows p_ℓ = b_ℓ (τ² − |ξ|²)^{ℓ/2} for even ℓ and p_ℓ = 0 for odd ℓ. This forces L to be a polynomial in the d'Alembertian, □ = ∂_t² − Δ.

What would settle it

A direct symbolic check would settle the key identity (3.39): for small n and ℓ, take the most general homogeneous Lorentz-invariant polynomial of degree ℓ in τ, ξ, and verify it equals a constant times (τ² − |ξ|²)^{ℓ/2} on the complement of the cone. A cleaner falsifier: construct a linear partial differential operator of the form (2.17) with nonconstant smooth coefficients that commutes with all translations and Lorentz transformations; if any such operator exists, Theorem 1 is false. A computational search in low order, testing whether the invariance conditions on the coefficients force them all to be constant, would settle it.

Watch

Extended reading notes

Core claim

The paper's main theorem states that for a linear partial differential operator L of order m written as L = Σ_{j+|α|≤m} a_{jα}(x) ∂_t^j ∂^α with continuous coefficients and nonvanishing top-order part, the following are equivalent: (1) L is Poincaré invariant in the sense that it commutes with pull-backs of all translations and all Lorentz transformations; (2) all coefficients are constants and L commutes with Lorentz transformations; (3) L = Σ_{j=0}^{[m/2]} b_j □^j for some constants b_j with b_{[m/2]} ≠ 0. The proof passes from the operator to its symbol polynomial p(τ,ξ) = Σ a_{jα}(0) τ^j ξ^α, which must be invariant under the Lorentz group, then decomposes p into homogeneous parts and uses the transitivity of the Lorentz group on the surfaces of constant τ² − |ξ|² to show each homogeneous part is a power of (τ² − |ξ|²). Corollary 2 adds that a second-order operator which is Poincaré- and dilation-invariant is exactly α□ with α ≠ 0.

Load-bearing premise

The proof depends on the Lorentz group connecting any two frequency vectors that lie on the same shell of constant τ² − |ξ|² outside the light cone; without that transitivity, invariant polynomials other than powers of τ² − |ξ|² could exist.

Editorial extensions

If this is right

Every Poincaré-invariant linear partial differential operator of finite order on R × Rⁿ has constant coefficients and is a polynomial in □, so the invariant operators form exactly the unital algebra generated by the d'Alembertian.
The only second-order Poincaré-invariant operators are the Klein–Gordon-type operators α□ + β with α ≠ 0; the only one also invariant under dilations is α□, i.e., the free wave operator.
The same equivalence holds at the level of Lorentz invariance alone once coefficients are constant: Lorentz-invariant constant-coefficient operators are exactly polynomials in □.
The Euclidean analogue, Theorem 2, characterizes rotation- and translation-invariant operators as polynomials in the Laplacian Δ, so the result transfers the classical Laplace characterization to Minkowski space-time.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper — the transitivity argument is purely algebraic, so the same conclusion should hold for any nondegenerate quadratic form of signature (1,n), including fields on curved Lorentzian vector spaces, provided the operator form and invariance definitions carry over.
Beyond the paper — the dilation-invariance condition is tested only on the zero set; a natural next experiment is whether the analogous 'zero-set' invariance under scaling selects the free wave equation among quasilinear or nonlinear wave operators, where integration by parts does not produce the λ² factor used in the proof.
Beyond the paper — the paper's final remarks point to Galilei invariance; the same exponential-symbol method, run with the Galilei group in place of the Lorentz group, would be the direct way to test whether the free Schrödinger operator is selected up to a mass parameter.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Plainly: the paper proves a classical fact and proves it well. The main theorem—translation and Lorentz invariance force a constant-coefficient operator whose symbol is a Lorentz-invariant polynomial, which then must be a polynomial in τ²−|ξ|²—is standard invariant theory. What the authors add is a fully self-contained, elementary derivation using exponentials rather than group representation theory, plus a neat kernel-based dilation argument that deletes the mass term. The proof is sound; I checked the steps and found no load-bearing gap. The rotation-invariance lemma is proved in the appendix, and the use of the transitivity of O(1,n) on the norm shells is correct.

The soft spot is novelty and citation practice. The converse direction, that invariant polynomials are generated by the Minkowski quadratic form, is a known result going back to classical invariant theory of the orthogonal group. The paper cites [10,11] for the Euclidean analogue but does not cite any prior source for the Lorentzian converse. That is a real omission, though minor. Also, the statement of Theorem 1 has a small edge-case wrinkle: for odd m, the derived symbol has degree at most m−1, so the assumption of a nonvanishing top-order part is inconsistent with invariance; the claim b_[m/2]≠0 is vacuous there. This doesn't affect Corollaries 1 and 2, which are the actual point. The dilation-invariance definition, based on kernel equivalence rather than commutation, is unusual but explicitly justified and works.

I'd send this to a referee. It is correct, readable, and potentially useful as a reference for the elementary proof. But an editor should be aware the central result is old, and a referee should be asked to verify the citation gap and the odd-m wording. If the authors add a sentence acknowledging the classical nature of the theorem, it's an acceptable note. For my own work, I'd cite a standard source rather than this paper, but I'd consider recommending it to a colleague looking for a self-contained proof.

Referee Report

1 major / 5 minor

Summary. The paper characterizes linear partial differential operators on Minkowski spacetime that commute with the Poincaré group. Theorem 1 states that a Poincaré-invariant operator of order m with nonvanishing top-order coefficients must be a polynomial in the d'Alembertian; Corollary 1 restricts second-order operators to α□+β, and Corollary 2 shows that adding dilation invariance (defined as invariance of the zero set) forces β=0. The proof passes to the symbol polynomial via exponential test functions, uses translation invariance to obtain constant coefficients, then rotation and Lorentz invariance to reduce the symbol to powers of τ²−|ξ|². An appendix proves the analogous Euclidean characterization of the Laplacian.

Significance. If the results are correct, the paper gives a fully self-contained, elementary proof of a classical fact: Lorentz-invariant (constant-coefficient) differential operators are polynomials in the d'Alembertian. The proof uses only invariant-polynomial arguments and no numerical or computational machinery, and the statement of Corollary 2 is a clean uniqueness result for the wave operator. The main reservation is that the characterization itself is not new, and the paper should acknowledge the prior literature. The proof is transparent and suitable for a pedagogical or reference treatment, provided the minor technical gap in Theorem 1 for odd m is fixed.

major comments (1)

[Theorem 1 (Section 2) and proof of (2)⇒(3) (Section 3)] The final step of the proof asserts b_{[m/2]} ≠ 0 "since the coefficients of the highest order m of L do not vanish identically." This is correct only for even m. If m is odd, the right-hand side of (3.40) is a polynomial in τ²−|ξ|² of degree at most 2[m/2]=m−1, so all coefficients of total order m in p vanish identically, contradicting the hypothesis ∑_{j+|α|=m}|a_{jα}|≠0. Consequently, as stated, Theorem 1 is not well-posed for odd m: statement (3) cannot hold for an operator of order m when m is odd. The theorem should be restricted to even m, or amended to state explicitly that no odd-order Poincaré-invariant operator with nonvanishing top-order part exists. This does not affect Corollaries 1 and 2, but it is a load-bearing point in the main theorem.

minor comments (5)

[Section 3, definition of φℓ and Cases (i)-(ii)] The text says "We prove that φℓ is a locally constant function on (R×R^n)\Γ" and then treats Cases (i) and (ii) separately. Since the complement of the light cone has several connected components, the argument actually proves that φℓ is constant on each component, with different values in the timelike and spacelike regions; the final single formula (3.39) relies on the relation (3.36). Please clarify this componentwise wording. In addition, the spacelike points with τ=0 are not covered by Case (ii) (which assumes τ>0) and should be obtained by continuity.
[Section 2, Eq. (3.31)] The definition of φℓ for odd ℓ involves a half-integer power in the denominator; although it is proved just before that pℓ=0 for odd ℓ, this is notationally awkward and should be stated more explicitly, for example by defining φℓ only for even ℓ.
[Remark 1 and Introduction] The converse characterization—that Poincaré-invariant differential operators are polynomials in the d'Alembertian—is a classical result in the theory of invariant differential operators, and the paper should cite prior work on this point instead of presenting it as new. The current references [2-7] only document the forward direction.
[Throughout] There are several typographical errors, including "D'Alemertian" for "D'Alembertian" in Section 1 and "hols" for "holds" in the proof of Proposition 1 in Section 4. These should be corrected.
[Section 5] The final remarks section is very short and does not discuss the relation of the results to the existing literature or the interpretation of the weak dilation-invariance notion; a brief discussion would improve the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity identified: the characterization of the d'Alembertian is derived from the invariance assumptions by a self-contained proof.

full rationale

The derivation is self-contained. The reduction of L to its symbol p(ξ) via exponentials (eqs. (2.19)-(2.22) and (3.1)-(3.4)) is a standard algebraic identity proved in the text. Translation invariance forces constant coefficients by comparing p(x-y,ξ) with p(x,ξ) (eqs. (3.5)-(3.8)), not by assumption. Rotation invariance is reduced to O(n)-invariance of the polynomials p_j, and the needed proposition on rotation-invariant homogeneous polynomials is stated and proved in the Appendix (Proposition 1, eq. (4.1)). Lorentz invariance is then used to show each homogeneous part p_ℓ is constant on the norm shells, using the transitivity of O(1,n) on the regions of constant τ²-|ξ|² (Cases (i) and (ii), eqs. (3.34)-(3.39)); this is a geometric fact applied directly, not an imported conclusion. Time reflection eliminates odd terms (eqs. (3.27)-(3.30)), and continuity extends the identity to the light cone. The result L = Σ b_j □^j follows algebraically from the symbol identity (3.40). No parameter is fitted, no target result is assumed in the hypothesis, and the external citations (Peetre's theorem in Remark 3, the standard invariance of □ in Remark 1, and the Euclidean characterization [10,11] reproved in Section 4) are not load-bearing: each is either proved in the paper or is a standard external fact. The self-citation [12] appears only in a closing remark and plays no role in the proof. Hence there is no circular step.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No fitted parameters and no invented physical entities. The central claim depends on classical polynomial invariant facts and the locality/finite-order assumption for differential operators; these are stated or proved in the paper.

assumptions (4)

standard math Rotation-invariant polynomials in R^n are polynomials in |x|^2, with odd homogeneous parts vanishing.
Invoked in Section 3 around (3.21) and proved in the Appendix, Proposition 1, using the O(n) action and homogeneity.
standard math O(1,n) acts transitively on vectors with a fixed nonzero value of τ²−|ξ|², and each such vector can be normalized to e0 or e1.
Used in Section 3 Cases (i) and (ii) to conclude that φℓ is locally constant on the complement of the light cone.
domain assumption Local linear operators with the local property have finite order and the form (2.17) with smooth coefficients.
Remark 3 states this restriction and cites Peetre's theorem [9]; the theorem does not apply to nonlocal operators.
domain assumption Dilation invariance is defined through equivalence of kernels, not through commutation with the dilation operator.
Corollary 2 relies on this kernel-equivalence definition; Remark 2 explains why the commutation form is not used.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Characterization of the D'Alembertian by the Poincar\'e Invariance." pith.science (2026). https://pith.science/paper/H3GEOMS5

@misc{pith2026250602513,
  author       = {Pith},
  title        = {Pith review of: Characterization of the D'Alembertian by the Poincar\'e Invariance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H3GEOMS5}},
  note         = {Machine review of arXiv:2506.02513}
}

abstract

Many physical models are described by partial differential equations and the most important mathematical structure of the equations is governed by the corresponding linear partial differential operators. Those linear partial differential operators are sometimes determined by the symmetry under the group of motion. In this paper, the d'Alembertian is shown to be characterized as the only linear partial differential operator of the second order that is invariant under the Poincar\'e group and dilations in the Minkowski space-time $\mathbb R\times\mathbb R^n$. The method of proof depends on the analysis of the invariance of the corresponding polynomial in space-time under the time reflections and space rotations.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages

[1]

Characterization of the D'Alembertian by the Poincar\'e Invariance

INTRODUCTION Many physical models are described by partial differential equations and the most important mathematical structure of the equations is governed by the corresponding linear partial differential operators. Those linear partial differential operators are sometimes determined by the symmetry under the group of motion. Symmetry is a decisive persp...

work page Pith review arXiv 2025
[2]

To this end, we start introducing basic notation

MAIN RESUL TS In this section, we state our main results. To this end, we start introducing basic notation. We denote by x = (t, x) = (x0, x1, . . . , xn) a point in space-time R × Rn. We also use the notation x = t x =   x0 x1 ... xn   as a column-vector representation to which (1 +n) × (1 +n)-matrices apply. Let ( e0, e1, . . . ,en) be the sta...

work page
[3]

Proof of Theorem 1 For ξ = (τ, ξ) ∈ R × Rn, we introduce eξ ∈ C∞(R × Rn; R) by eξ(x) = eξ(t, x) = exp(τ t+ ξ · x) for x = (t, x) ∈ R × Rn

PROOF OF THE MAIN RESUL TS In this section, we prove Theorem 1 and Corollary 2, since Corollary 1 follows directly from Theorem 1 with m = 2. Proof of Theorem 1 For ξ = (τ, ξ) ∈ R × Rn, we introduce eξ ∈ C∞(R × Rn; R) by eξ(x) = eξ(t, x) = exp(τ t+ ξ · x) for x = (t, x) ∈ R × Rn. (3.1) Then we see that eξ(0) = 1 and for any ( j, α) ∈ Z≥0 × Zn ≥0, ∂j t ∂αe...

work page
[4]

We designate a point in the Euclid space Rn as x = (x1,

APPENDIX: CHARACTERIZA TION OF THE LAPLACIAN BY THE EUCLIDEAN MOTION GROUP In this section, we review the characterization of the Laplacian by the Euclidean motion group [10, 11]. We designate a point in the Euclid space Rn as x = (x1, . . . , xn) =   x1 · · · xn   = xjej, where (ej; j = 1, . . . , n) is the standard basis of Rn given by e1 =  ...

work page
[5]

For related subjects, we refer the reader to the references [13–19]

FINAL REMARKS A similar approach to the linear partial differential operators from the point of view of symmetry is possible for the quantum mechanics [12]. For related subjects, we refer the reader to the references [13–19]. ACKNOWLEDGMENTS We are grateful to the referee for useful comments. H.N. is partially supported by JSPS KAKENHI Grant No. 23K03268....

work page
[6]

R. W. Sharpe, Differential Geometry, Cartan ’s Generalization of Klein ’s Erlangen Program , Graduate Texts in Mathe- matics 166, Springer 1997

work page 1997
[7]

De Faria, and W

E. De Faria, and W. De Melo, Mathematical Aspects of Quantum Field Theory , Cambridge Studies in Advanced Mathe- matics 127, Cambridge University Press, 2010

work page 2010
[8]

Scheck, Classical Field Theory, On Electrodynamics, Non-Abelian Gauge Theories and Gravitation , 2nd

F. Scheck, Classical Field Theory, On Electrodynamics, Non-Abelian Gauge Theories and Gravitation , 2nd. Edition, Springer 2018

work page 2018

Show all 24 references

[9]

Zeidler, Quantum Field Theory I: Basics in Mathematics and Physics (Springer, Heidelberg, 2006)

E. Zeidler, Quantum Field Theory I: Basics in Mathematics and Physics (Springer, Heidelberg, 2006)

2006
[10]

Zeidler, Quantum Field Theory II: Quantum Electrodynamics (Springer, Heidelberg, 2009)

E. Zeidler, Quantum Field Theory II: Quantum Electrodynamics (Springer, Heidelberg, 2009)

2009
[11]

Zeidler, Quantum Field Theory III: Gauge Theory (Springer, Heidelberg, 2011)

E. Zeidler, Quantum Field Theory III: Gauge Theory (Springer, Heidelberg, 2011)

2011
[12]

J. D. Jackson, Classical Electrodynamics, 3rd. Edition, Wiley 1999

1999
[13]

H¨ ormander,Linear Partial Differential Operators , Die Grundlehren der Mathematischen Wissenschaften 116, Springer 1963

L. H¨ ormander,Linear Partial Differential Operators , Die Grundlehren der Mathematischen Wissenschaften 116, Springer 1963

1963
[14]

Peetre, R´ ectification ` a l’article≪Une caract´ erisation abstraite des op´ rateurs diff´ erentiels≫, Math

J. Peetre, R´ ectification ` a l’article≪Une caract´ erisation abstraite des op´ rateurs diff´ erentiels≫, Math. Scand. 8 (1960), 116–120

1960
[15]

Nomura, Spherical Harmonics and Group Representations , Nippon Hyoron Publ., 2018 (in Japanese)

T. Nomura, Spherical Harmonics and Group Representations , Nippon Hyoron Publ., 2018 (in Japanese)

2018
[16]

Shimakura, Elliptic Partial Differential Operators , Kinokuniya Publ., 1978 (in Japanese)

N. Shimakura, Elliptic Partial Differential Operators , Kinokuniya Publ., 1978 (in Japanese)

1978
[17]

Nakazato, T

H. Nakazato, T. Ozawa, Characterization of the time-dependent free Schr¨ odinger operator by the Galilei invariance, to appear in the proceedings ICIAM 2023, Springer

2023
[18]

Bargmann, On Unitary Ray Representations of Continuous Groups, Ann

V. Bargmann, On Unitary Ray Representations of Continuous Groups, Ann. of Math. 59 (1954), 1–46

1954
[19]

A. O. Barut, Conformal Group →Schr¨ odinger Group→Dynamical Group – The Maximal Kinematical Group of the Massive Schr¨ odinger Particle, Helv. Phys. Acta46 (1973), 493–503

1973
[20]

Hamermesh, Galilei Invariance and the Schr¨ odinger Equation, Ann

M. Hamermesh, Galilei Invariance and the Schr¨ odinger Equation, Ann. Phys. 9 (1960), 518–521. 15

1960
[21]

Levy-Leblond, Galilei Group and Nonrelativistic Quantum Mechanics, J

J.-M. Levy-Leblond, Galilei Group and Nonrelativistic Quantum Mechanics, J. Math. Phys. 4 (1963), 776–788

1963
[22]

Roncadelli, A Simplified Approach to the Schr¨ odinger Equation, Nuovo Cimento 50A (1979), 509–514

M. Roncadelli, A Simplified Approach to the Schr¨ odinger Equation, Nuovo Cimento 50A (1979), 509–514

1979
[23]

Z. E. Musielak and J. L. Fry, Physical theories in Galilean space-time and the origin of Schr¨ odinger-like equations, Ann. Phys. 324 (2009), 296–308

2009
[24]

Z. E. Musielak and J. L. Fry, General Dynamical Equations for Free Particles and Their Galilean Invariance, Int. J. Theor. Phys. 48 (2009), 1194–1202

2009

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Characterization of the D'Alembertian by the Poincar\'e Invariance

INTRODUCTION Many physical models are described by partial differential equations and the most important mathematical structure of the equations is governed by the corresponding linear partial differential operators. Those linear partial differential operators are sometimes determined by the symmetry under the group of motion. Symmetry is a decisive persp...

work page Pith review arXiv 2025

[2] [2]

To this end, we start introducing basic notation

MAIN RESUL TS In this section, we state our main results. To this end, we start introducing basic notation. We denote by x = (t, x) = (x0, x1, . . . , xn) a point in space-time R × Rn. We also use the notation x = t x =   x0 x1 ... xn   as a column-vector representation to which (1 +n) × (1 +n)-matrices apply. Let ( e0, e1, . . . ,en) be the sta...

work page

[3] [3]

Proof of Theorem 1 For ξ = (τ, ξ) ∈ R × Rn, we introduce eξ ∈ C∞(R × Rn; R) by eξ(x) = eξ(t, x) = exp(τ t+ ξ · x) for x = (t, x) ∈ R × Rn

PROOF OF THE MAIN RESUL TS In this section, we prove Theorem 1 and Corollary 2, since Corollary 1 follows directly from Theorem 1 with m = 2. Proof of Theorem 1 For ξ = (τ, ξ) ∈ R × Rn, we introduce eξ ∈ C∞(R × Rn; R) by eξ(x) = eξ(t, x) = exp(τ t+ ξ · x) for x = (t, x) ∈ R × Rn. (3.1) Then we see that eξ(0) = 1 and for any ( j, α) ∈ Z≥0 × Zn ≥0, ∂j t ∂αe...

work page

[4] [4]

We designate a point in the Euclid space Rn as x = (x1,

APPENDIX: CHARACTERIZA TION OF THE LAPLACIAN BY THE EUCLIDEAN MOTION GROUP In this section, we review the characterization of the Laplacian by the Euclidean motion group [10, 11]. We designate a point in the Euclid space Rn as x = (x1, . . . , xn) =   x1 · · · xn   = xjej, where (ej; j = 1, . . . , n) is the standard basis of Rn given by e1 =  ...

work page

[5] [5]

For related subjects, we refer the reader to the references [13–19]

FINAL REMARKS A similar approach to the linear partial differential operators from the point of view of symmetry is possible for the quantum mechanics [12]. For related subjects, we refer the reader to the references [13–19]. ACKNOWLEDGMENTS We are grateful to the referee for useful comments. H.N. is partially supported by JSPS KAKENHI Grant No. 23K03268....

work page

[6] [6]

R. W. Sharpe, Differential Geometry, Cartan ’s Generalization of Klein ’s Erlangen Program , Graduate Texts in Mathe- matics 166, Springer 1997

work page 1997

[7] [7]

De Faria, and W

E. De Faria, and W. De Melo, Mathematical Aspects of Quantum Field Theory , Cambridge Studies in Advanced Mathe- matics 127, Cambridge University Press, 2010

work page 2010

[8] [8]

Scheck, Classical Field Theory, On Electrodynamics, Non-Abelian Gauge Theories and Gravitation , 2nd

F. Scheck, Classical Field Theory, On Electrodynamics, Non-Abelian Gauge Theories and Gravitation , 2nd. Edition, Springer 2018

work page 2018