Operational Calculus on Curved Differentials: Optimal N-Complex Bounds and Persistent Homology
Mauricio Angel · 2026
Read Operational Calculus on Curved Differentials: Optimal N-Complex Bounds and Persistent Homology by Mauricio Angel (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-14155.txt.
Introduction
Angel sets out a canonical normal form for powers of a curved differential within curved differential algebras. The operator calculus avoids heavy combinatorics. From it he gives sharp tests for when curvature constraints force N-complex structures, showing that a nilpotent curvature power still guarantees a strict complex structure of related order. He also treats curvature as a filtration control on a square-zero chain complex, gaining Lipschitz control of barcodes under curvature change, and illustrates the mechanism on a four-vertex flag complex.
Catalog record
| CREATOR | Mauricio Angel |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27853 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.14155
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Operational Calculus on Curved Differentials: Optimal N-Complex Bounds and Persistent Homology Mauricio Angel Abstract We establish a canonical normal form for the iterates of a curved differential in curved differential algebras (CDA), proving that d 2 m = ( a d c ) m d^{2m}=(ad_{c})^{m} and d 2 m + 1 = ( a d c ) m ∘ d d^{2m+1}=(ad_{c})^{m}\circ d . This operator calculus clarifies the underlying algebraic structure of CDAs and bypasses the need for complex combinatorics. Using this framework, we provide sharp criteria for curvature constraints to induce N-complex structures. We demonstrate that, while the nilpotency of the curvature element ( c n = 0 c^{n}=0 ) is insufficient to bound the nilpotency of d d to 2 n 2n , it fundamentally guarantees a strict ( 4 n − 2 ) (4n-2) -complex structure. On the applied side, we model curvature as a filtration controller on a genuine square–zero chain complex. This places us under the standard persistence stability framework and yields a Lipschitz control of barcodes with respect to degreewise curvature variation. A reproducible toy example on a four–vertex flag complex illustrates the mechanism. 1 Introduction Curved differential algebras (CDA) arise as a natural extension of differential graded algebras (DGA), incorporating the notion of curvature into the differential operator. Unlike traditional differentials that strictly satisfy the square-zero condition ( d^{2}=0 ), the curvature c in a CDA dictates that d^{2}=[c,-] . This enrichment of the algebraic structure provides a versatile framework to study phenomena in algebraic topology, deformation theory, and homological algebra where geometric or algebraic obstructions prevent a flat differential. The foundational theory of curved algebras has been significantly advanced by Keller [ 8 ] , Nicolas [ 11 ] , and Brzeziński [ 2 ] , who established their connections with derived categories and corings. Lazarev [ 9 ] and Căldăraru [ 3 ] further explored their characteristic classes and Hochschild homology, and Pauksztello [ 12 ] characterizes when a differential graded R-S-bimodule induces a full embedding of derived categories. These papers collectively suggest that curved differential graded algebras generalize other algebras by providing a framework for studying algebraic structures in a more general and flexible way. However, a persistent technical challenge in working with CDAs has been the algebraic complexity of the higher iterates of the differential, d^{k} . Naive expansions of d^{k} generate nested commutators and derivations that obscure the underlying homological properties. In this paper, we resolve this obstruction by establishing a canonical normal form for the iterates of a curved differential (Theorem 3). We prove that every iterate d^{k} factors uniquely as a power of the inner derivation ad_{c} followed by at most one copy of d . This structural result drastically simplifies computations and reveals how curvature constraints naturally induce N -complex structures (where d^{N}=0 ).A central contribution of this work is the precise identification of the bounds for such N -complexes. We demonstrate through a sharp counterexample (Section 4.3) that the mere nilpotency of the curvature element ( c^{n}=0 ) does not yield the naive bound d^{2n}=0 . Rather than the naive 2n -bound, our normal form combined with degreewise left/right nilpotency yields the universal (4n-2) -bound. The free example with n=2 shows that this exponent is sharp in that case; for general n , optimality remains open. Finally, to connect with persistent homology, we refrain from curved-to-dg lifts and instead let curvature drive the filtration on an honest chain complex. Classical stability then applies verbatim, delivering an interleaving/bottleneck bound controlled by the \ell^{\infty} shift of the filtration (Theorem 10; Corollary 11). By connecting CDAs to corings [ 2 ] , N -complexes [ 1 ] , and persistent homology, this work establishes a new research program at the intersection of homological algebra and topological data science. 2 Conventions and Setting Throughout we fix a field k of characteristic 0 . All graded components A^{i} of our (associative, unital) \mathbb{Z} -graded algebra A^{\bullet}=\bigoplus_{i\in\mathbb{Z}}A^{i} are finite-dimensional k -vector spaces. We consider a curved dg-algebra (A^{\bullet},m,d,c) in the sense that |d|=1,\qquad|c|=2,\qquad d^{2}=\operatorname{ad}_{c}:=[c,-],\qquad d(c)=0, (1) where [x,y]=xy-(-1)^{|x||y|}yx is the graded commutator and m denotes the product. We use the standard Koszul sign rule. Inner derivations and degreewise norms. For each i we fix an operator norm \|\cdot\|_{i} on A^{i} and define \|\operatorname{ad}_{c}\|_{i\to i+2}:=\sup_{0\neq a\in A^{i}}\frac{\|[c,a]\|_{i+2}}{\|a\|_{i}}. Unless stated otherwise, norms on \bigoplus_{i}A^{i} are taken to be the \ell^{\infty} -sum of the degreewise norms. When we refer to “boundedness of \operatorname{ad}_{c} ” we mean \sup_{i}\|\operatorname{ad}_{c}\|_{i\to i+2}<\infty . Graded commutator identities. For a homogeneous x we write \operatorname{ad}_{x}=[x,-] . We repeatedly use d(xy)=d(x)\,y+(-1)^{|x|}x\,d(y),\qquad[d,\operatorname{ad}_{x}]=\operatorname{ad}_{d(x)}. (2) In particular, from ( 1 ) we obtain [d,\operatorname{ad}_{c}]=\operatorname{ad}_{d(c)}=0 , so d and \operatorname{ad}_{c} super-commute as graded derivations of degrees 1 and 2 respectively. Terminology. Following Positselski, curved differential graded algebras are often abbreviated as CDG-algebras (see [ 13 , 14 ] ); our notation d^{2}=\operatorname{ad}_{c} fits this convention, with c playing the role of the curvature element. 3 Iterates of a Curved Differential We recall from § 2 that a curved dg-algebra is a quadruple (A^{\bullet},m,d,c) over a field k of characteristic 0 with |d|=1 , |c|=2 , d^{2}=\operatorname{ad}_{c}:=[c,-],\qquad d(c)=0, (3) and the graded commutator [x,y]=xy-(-1)^{|x||y|}yx . We also use the graded Leibniz rule and the identity [d,\operatorname{ad}_{x}]=\operatorname{ad}_{d(x)} for homogeneous x . 3.1 Structural commutation and a reduction principle The following observation underlies all computations with iterates of d . Lemma 1 (Super-commutation of d and \operatorname{ad}_{c} ) . In any curved dg-algebra ( ( 3 ) ) one has [d,\operatorname{ad}_{c}]=\operatorname{ad}_{d(c)}=0. In particular, d and \operatorname{ad}_{c} commute as endomorphisms of the graded vector space A^{\bullet} . Proof. The standard identity [d,\operatorname{ad}_{x}]=\operatorname{ad}_{d(x)} gives [d,\operatorname{ad}_{c}]=\operatorname{ad}_{d(c)} , which vanishes by d(c)=0 . ∎ The quadratic relation d^{2}=\operatorname{ad}_{c} together with Lemma 1 implies that every word in the letter d can be reduced to a canonical normal form. For later use we isolate the following consequence. Lemma 2 (Commutation with powers of \operatorname{ad}_{c} ) . For all m\geq 0 one has d\,(\operatorname{ad}_{c})^{m}=(\operatorname{ad}_{c})^{m}d . Proof. By Lemma 1 the operators d and \operatorname{ad}_{c} commute. Hence so do d and (\operatorname{ad}_{c})^{m} . ∎ 3.2 Normal form for d^{k} We can now state and prove the main structural result of this section. Theorem 3 (Normal form for the iterates of d ) . Let (A^{\bullet},m,d,c) satisfy ( 3 ). Then, as endomorphisms of A^{\bullet} , d^{2m}=(\operatorname{ad}_{c})^{m},\qquad d^{2m+1}=(\operatorname{ad}_{c})^{m}\circ d\qquad\text{for all }m\geq 0. (4) Equivalently, every iterate d^{k} factors uniquely as a power of the inner derivation \operatorname{ad}_{c} followed by at most one copy of d . Proof. We proceed by induction on m\geq 0 . For m=0 the statement reads d^{0}=\mathrm{id} and d^{1}=d , which is tautological. For m=1 we have d^{2}=\operatorname{ad}_{c} by ( 3 ) and d^{3}=d\circ d^{2}=d\circ\operatorname{ad}_{c}=(\operatorname{ad}_{c})\circ d by Lemma 2 . Assume the claim holds for a fixed m\geq 1 . Then d^{2(m+1)}=d^{2}\circ d^{2m}=\operatorname{ad}_{c}\circ(\operatorname{ad}_{c})^{m}=(\operatorname{ad}_{c})^{m+1}, and d^{2(m+1)+1}=d\circ d^{2(m+1)}=d\circ(\operatorname{ad}_{c})^{m+1}=(\operatorname{ad}_{c})^{m+1}\circ d, using Lemma 2 . The induction closes. ∎ Remark 1 (No ad hoc binomial sums are needed) . The proof only uses the universal relations d^{2}=\operatorname{ad}_{c} and [d,\operatorname{ad}_{c}]=0 . Thus, unlike heuristic expansions that distribute d across products and insert formal powers of c , the normal form ( 4 ) is canonical and independent of additional commutativity