REVIEW 3 major objections 3 minor 7 references
Remarks on natural differential operators with tensor fields
T0 review · 3 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read The paper claims all finite-order natural differential operators between tensor fields of the stated types are R-bilinear and of order one, and classifies the results in several cases.
desk verdict Useful bilinear classifications undermined by a false main theorem: Theorem 1.1's counting argument misses solutions to (1.2), and Section 4's completeness claim is invalid. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing identity is the homogeneity equation (1.2), $$\sum_{l=0}^{k} ((p+l-1)a_l+(s-r+l)b_l)=s-r+p,$$ which rescaling naturality imposes on the polynomial orders $a_l$ in $\partial^l\varphi$ and $b_l$ in $\partial^l\psi$. The proof asserts this equation has exactly two non-negative integer solutions, corresponding to the bilinear first-order monomials. The second mechanism is the method of an auxiliary linear symmetric connection $K$, together with the second-order reduction theorem: derivatives are replaced by covariant derivatives with respect to $K$, the operator is required to be independent of $K$, and the resulting homogeneous linear equations fix the coefficients of the absolute invariant tensors appearing in (1.3)–(1.5).
What would settle it
For $p=2$, $r=0$, $s=1$, define $\Phi(S,\psi)_{jkl}=S^i_{jk}\psi_i\psi_l$, one contraction of $S\otimes\psi\otimes\psi$. This is a natural differential operator of order 0, it is not R-bilinear, and it satisfies the homogeneity equation because $1\cdot(p-1)+2\cdot(s-r)=1+2=3=s-r+p$. Its existence directly contradicts the theorem's conclusion that every such operator is R-bilinear and of order 1.
Extended reading notes
Core claim
The paper's central claim is Theorem 1.1: for $p>1$ and $s>r$, every finite-order natural differential operator $\Phi$ mapping a $(1,p)$-tensor field $\varphi$ and an $(r,s)$-tensor field $\psi$ into an $(r,s+p)$-tensor field is R-bilinear and of order 1. The proof uses naturality under constant rescaling to force $\Phi$ to be a polynomial in the jet variables, then counts degrees in the homogeneity equation (1.2); the paper claims this count leaves exactly two monomials, $\varphi\cdot\partial\psi$ and $\partial\varphi\cdot\psi$. A second theorem removes the type restrictions when bilinearity is assumed. The concrete sections turn the resulting order-1 form into explicit parameter families of operators built from invariant tensors and contractions.
Load-bearing premise
The proof of Theorem 1.1 assumes the homogeneity equation (1.2) has only the two bilinear solutions because all its coefficients are positive, but when $s-r=p$ the value $b_0=2$ also solves the equation (and in Section 4 so does $a_0=1$, $b_0=2$), giving natural order-0 non-bilinear operators such as $S^i_{jk}\psi_i\psi_l$.
Editorial extensions
If this is right
- If Theorem 1.1 holds, for the covered tensor types there are no higher-order or genuinely nonlinear natural operators; every natural operator is a bilinear first-order expression built from one derivative of one input times the other input, contracted through invariant tensors.
- For R-bilinear natural operators, order 1 holds for every $p,r,s$, so classification reduces to solving finite linear systems for the coefficient tensors in (1.3)–(1.5).
- Up to a constant multiple, the only natural bilinear operator taking two vector fields to a vector field is the Lie bracket, so the classification reproduces the classical uniqueness of the bracket.
- For a vector field and a 1-form, every natural bilinear operator is a real linear combination of $d(\psi(X))$ and $i_X d\psi$; for a vector field and a $(0,2)$-tensor, the four generators are $L_X\psi$, $L_X\tilde{\psi}$, $d(X\lrcorner\psi)$, and $d(X\lrcorner\tilde{\psi})$.
- For a $(1,2)$-tensor field $S$ and a 1-form $\psi$, all natural operators would form a 19-parameter family, with the Yano–Ako operator expressible as $d(\psi\circ\mathrm{Alt}\,S)(X,Z,Y)+d\psi(S(X,Y),Z)$.
Reading between the lines
- The homogeneity count is not exhaustive: when $s-r=p$, $b_0=2$ also solves (1.2), so order-0 products such as $\psi\otimes\psi$ (with the appropriate contractions) are natural operators of the stated type, and in Section 4 the solution $a_0=1$, $b_0=2$ gives operators like $S^i_{jk}\psi_i\psi_l$ that are not R-bilinear.
- A corrected classification would therefore be a finite family of polynomial operators of order 0 and 1 rather than purely bilinear first-order operators; the auxiliary-connection computation would still work if the list of admissible monomials is enlarged.
- The same counting method can be applied mechanically to any pair of tensor types: enumerate all non-negative integer solutions of the homogeneity equation, then solve the $K$-independence linear system for each candidate monomial.
- Because Theorem 1.2 is separate from Theorem 1.1, the bilinear classifications in Sections 2 and 3 are not affected by the gap in the homogeneity count; only the claim that all natural operators are bilinear would need a corrected argument.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies natural differential operators that transform two tensor fields into a tensor field. Its main result, Theorem 1.1, claims that for p > 1 and s > r, every finite-order natural differential operator taking a (1,p)-tensor field and an (r,s)-tensor field into an (r,s+p)-tensor field is R-bilinear and of order 1. Theorem 1.2 claims that R-bilinearity alone forces order 1 for all p,r,s. Sections 2 and 3 classify bilinear natural operators in concrete cases (vector fields, 1-forms, (0,2)-tensors, (1,1)-tensors), and Section 4 uses Theorem 1.1 to give a 19-parameter classification of natural operators taking a (1,2)-tensor S and a 1-form ψ into a (0,3)-tensor.
Significance. If Theorem 1.1 were correct, it would be a strong rigidity statement: naturality plus the specified tensor types would force bilinearity and first order, making finite classifications feasible. The bilinear classifications in Sections 2 and 3 are of independent interest and appear to be obtained by a standard and potentially correct invariant-theoretic method. However, the central claim is falsified by simple order-0 counterexamples, and the completeness statement in Section 4, which depends directly on Theorem 1.1, is therefore invalid. The paper's main advertised contribution does not hold as stated.
major comments (3)
- [§1, proof of Theorem 1.1, equation (1.2)] The claim that positivity of all coefficients in equation (1.2) leaves only the two solutions a0=1,b1=1 and a1=1,b0=1 is arithmetically false. For p=2, r=0, s=1, equation (1.2) becomes Σ_{l=0}^k (l+1)(a_l+b_l)=3, and the solution with b0=3 and all other entries zero satisfies it. This corresponds to the natural order-0 operator Φ(S,ψ)=ψ⊗ψ⊗ψ, which sends a (1,2)-tensor S and a 1-form ψ to a (0,3)-tensor, is natural, and is not R-bilinear. More generally, whenever s−r divides p, the operator ψ^{⊗(1+p/(s−r))}, with appropriate contractions to obtain the target tensor type, is a natural order-0 non-bilinear operator. Hence Theorem 1.1 is false.
- [§4, Theorem 4.1] The proof of Theorem 4.1 begins by invoking Theorem 1.1 to assert that all natural operators transforming a (1,2)-tensor S and a 1-form ψ into (0,3)-tensors are R-bilinear and of order 1. Since Theorem 1.1 is false, this premise is invalid. The operator Φ(S,ψ)=ψ⊗ψ⊗ψ is a concrete natural operator of order 0 that is not in the 19-parameter family described by Lemmas 4.1–4.4. The claimed completeness of the classification in Theorem 4.1 therefore does not hold.
- [Abstract and Introduction] The abstract states that the paper gives 'the full classification of such operators' in several situations. Because Theorem 1.1, which is the basis for the unrestricted classification in Section 4, is false, this advertised claim is not supported. The bilinear classifications in Sections 2 and 3 may still be correct, but they do not justify the paper's stated scope.
minor comments (3)
- [§4.1, Lemma 4.3] The last displayed operator in Lemma 4.3 is missing a closing parenthesis: 'ψ(Z)d(C^1_1 S)(X,Y' should be 'ψ(Z)d(C^1_1 S)(X,Y)'.
- [§3.1, Theorem 3.1] Theorem 3.1 is quoted from reference [3] without proof; it would be clearer to state explicitly that this is a recalled classification rather than a new result.
- [Throughout] The manuscript contains numerous OCR artifacts and typographical inconsistencies (e.g., misaligned subscripts, missing spaces, occasional garbled formulas). A careful copyedit is needed before any resubmission.
Circularity Check
No circularity: the paper's derivations are self-contained, and the known flaw in Theorem 1.1 is an arithmetic error, not a circular argument.
full rationale
The derivation chain is not circular. Theorem 1.1 is derived from equivariance and a homogeneity condition: the paper writes the weighted homogeneity equation (1.2) and infers bilinearity and order one. Whether or not that inference is mathematically correct, it is not assumed as an input; it is presented as a consequence of the homogeneous function theorem. The later classifications, including Theorem 4.1, are obtained by writing the general polynomial ansatz from (1.3)-(1.5), imposing independence of an auxiliary linear connection, and solving the resulting linear systems. These steps do not reintroduce the target classifications as assumptions. The only self-citations are to the author's book [3] for the standard auxiliary connection method and for the recalled classification in Theorem 3.1; these are presented as known external results, not as devices forcing the paper's main conclusions. The known defect in the proof of Theorem 1.1, namely the assertion that positivity of coefficients in (1.2) leaves only two non-negative solutions, is a real mathematical error and a correctness concern, not a circularity: for p=2, r=0, s=1 the monomial (a0,b0)=(0,3) solves (1.2), giving a natural non-bilinear zero-order operator omitted by Section 4. That shows the conclusion can fail, but it does not show that the conclusion was built into the assumptions.
Assumptions & free parameters
assumptions (4)
- standard math Homogeneous function theorem for equivariant mappings under positive scaling
- domain assumption Second reduction theorem for natural operators depending on a symmetric linear connection
- domain assumption Absolute invariant tensors are linear combinations of tensor products of the identity
- domain assumption Method of auxiliary linear symmetric connection K
Cite this review
Pith. "Pith review of Remarks on natural differential operators with tensor fields." pith.science (2026). https://pith.science/paper/PPJNK5OX
@misc{pith2026190804528,
author = {Pith},
title = {Pith review of: Remarks on natural differential operators with tensor fields},
year = {2026},
howpublished = {\url{https://pith.science/paper/PPJNK5OX}},
note = {Machine review of arXiv:1908.04528}
}
read the original abstract
We study natural differential operators transforming two tensor fields into a tensor field. First, it is proved that all bilinear operators are of order one, and then we give the full classification of such operators in several concrete situations.
Reference graph
Works this paper leans on
-
[1]
A. Fr ¨olicher, A. Nijenhuis : Theory of vector-valued differential forms, I, II, Nederl. Akad. Wetensch. Proc. Ser. A. 59 (1956) 338–350, 351–359
work page 1956
-
[2]
I. Kol ´aˇr, P. W. Michor, J. Slov ´ak: Natural Operations in Differential Geometry, Springer–Verlag 1993
work page 1993
- [3]
-
[4]
Salimov : On operators associated with tensor fields , J
A. Salimov : On operators associated with tensor fields , J. Geom. 99 (2010) 107145, DOI 10.1007/s00022-010-0059-6
-
[5]
J. A. Schouten : On the differential operators of first order in tensor calculu s, Rapport ZA 1953-012, Math. Centrum Amsterdam. (1953), 6pp
work page 1953
-
[6]
J. A. Schouten : Ricci-Calculus: An Introduction to Tensor Analysis and Its Geometrical Applications (2nd ed.), Springer-Verlag Berlin Heidelber g 1954
work page 1954
-
[7]
K. Yano, M. Ako : On certain operators associated with tensor fields , Kodai Math. Sem. Rep. 20 (1968) 414-436. Department of Mathematics and Statistics, Masaryk Univers ity Kotl´aˇrsk´a 2, 611 37 Brno, Czech Republic e-mail: janyska@math.muni.cz
work page 1968
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.