Bundle of Principal Parts¶

A finite-order vector bundle of local section data over a smooth algebraic variety.

Version: v1 · 2026-10-07 · History
Domain-specific #: 13820
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomain: Algebraic Geometry → Mathematics
Aliases: Principal Parts Bundle

Core Idea¶

For a smooth algebraic variety \(X\), a line bundle \(L\) and a nonnegative order \(n\), the bundle of principal parts \(P^n_X(L)\) collects the order-\(n\) local behavior of sections of \(L\) as its base point varies over \(X\). It is constructed from the \(n\)th infinitesimal neighborhood of the diagonal in \(X\times X\), not from a selected collection of global sections. Under the smoothness and line-bundle assumptions it is locally free: its geometric realization is a vector bundle over \(X\).^[1]^[2]

A chosen linear system can map into this bundle by taking finite jets of its sections. That extra map makes ramification or osculation questions possible, but neither the map nor a special rank-drop locus defines \(P^n_X(L)\). The bundle already exists when no such system is chosen.^[1]^[3]^[4]

Structural Signature¶

Signature: smooth base \(X\) + invertible input \(L\) + finite order \(n\) → \(n\)th thickening of the diagonal → locally free sheaf of truncated section data over \(X\) → compatible order truncations \(P^n_X(L)\to P^{n-1}_X(L)\).^[1]^[3]

Base and input. \(X\) supplies points and local coordinates; \(L\) specifies which sections are expanded. Changing \(L\) can change the bundle.^[1]
Order and construction. The ideal of the diagonal is taken to its \((n+1)\)st power. Pulling \(L\) to that thickened diagonal and pushing along a projection assembles finite-order data over all base points.^[1]
Local-freeness and rank. If \(X\) is smooth of dimension \(d\) and \(L\) is a line bundle, \(P^n_X(L)\) has rank \(\binom{d+n}{n}\). Its local Taylor coefficients are coordinate descriptions of an intrinsic bundle, rather than the definition by themselves.^[1]^[3]
Order tower. For \(n\geq 1\), the map to the previous order has kernel \(L\otimes\operatorname{Sym}^n\Omega^1_X\). At order zero, \(P^0_X(L)=L\).^[2]^[1]^[3]
Optional section map. A finite-dimensional space of sections may be mapped to their \(n\)-jets. Its rank is an application-specific question; no such space or rank-drop locus is required for the bundle to exist.^[1]^[4]

What It Is Not¶

A single Taylor polynomial at one point is a fiberwise piece of information, not the varying sheaf over \(X\). A singular-base principal-parts sheaf may still be definable, but the smooth-base local-freeness and rank conclusion cannot simply be carried over. Gatto and Ricolfi discuss locally free replacements in Gorenstein-curve settings precisely because this distinction matters.^[1]

The input line bundle \(L\) is not automatically the higher-order output when \(n>0\); it is the order-zero case. Nor is a ramification point, scroll inflection locus, projective bundle, or fixed-origin jet group the principal-parts bundle itself. Each requires additional data or a different kind of object.^[1]^[4]

Scope of Application¶

For a smooth projective curve \(C\), the bundle \(P^r_C(L)\) accepts the order-\(r\) data of sections of a linear system \((L,V)\) of dimension \(r+1\). A rank drop in the resulting section-to-jet map detects a ramification point; its determinant is the Wronskian. In the canonical linear system on a smooth plane quartic, the Weierstrass points are flexes. This curve use is one application of a bundle that can be constructed before \(V\) is specified.^[1]

For a smooth complex projective scroll \(X\to C\), the hyperplane line bundle \(L\) has a higher-dimensional principal-parts bundle \(P^k_X(L)\). The embedding's section space gives a map \(j^k:V\otimes\mathcal O_X\to P^k_X(L)\); the images of its fiber maps define osculating spaces. Lanteri, Mallavibarrena and Piene derive a scroll-specific upper bound on this map's rank and, under their numerical and generic-rank hypotheses, define an inflectional locus where it falls below that bound. The scroll bound is not the rank of the bundle itself.^[4]^[3]

Clarity¶

Begin with the construction question: what are \(X\), \(L\) and \(n\), and is the smooth/locally free conclusion justified? Only then ask what chosen sections are being evaluated and what geometric condition their image studies. This order keeps a canonical object separate from a contingent experiment on that object.^[1]^[3]

The two ranks in a scroll example are easily confused. The bundle rank is \(\binom{d+k}{k}\) for \(d=\dim X\). The image rank of \(j^k\) can be lower; the scroll geometry places the special \(kd+1\) bound under the paper's setting. A rank-drop locus concerns that map, not a failure of the principal-parts bundle to be locally free.^[3]^[4]

Manages Complexity¶

The diagonal construction makes many local Taylor expansions compatible across coordinate changes. Instead of treating each derivative list separately, it gives one sheaf that varies over \(X\), with a controlled truncation at every order. The exact sequence identifies the new order-\(n\) layer as a symmetric cotangent contribution tensored with \(L\).^[1]^[3]

Separating construction from evaluation also makes case assumptions visible. Smoothness supports the stated vector-bundle properties. Embedding data supplies a specific \(V\) for osculation; a curve's linear system supplies another. A theorem about a selected map's rank cannot be promoted into a theorem about all principal-parts bundles.^[1]^[4]

Abstract Reasoning¶

Take \(n=0\). The diagonal neighborhood retains only the value of a local section, so \(P^0_X(L)=L\). Increasing to \(n=1\) retains first-order variation as well. The quotient back to order zero forgets that first-order part, and its kernel is \(L\otimes\Omega^1_X\). Higher orders repeat this operation with symmetric powers of the cotangent sheaf.^[2]^[1]

Now hold \(X\), \(L\) and \(n\) fixed but choose different spaces \(V\) of global sections. The resulting evaluation maps may have different ranks or degeneracy loci, while the underlying \(P^n_X(L)\) is unchanged. That counterfactual is the quickest test of whether a proposed feature defines the bundle or only an application of it.^[1]^[4]

Knowledge Transfer¶

In a new algebraic-geometric setting, specify the base, line bundle and order; verify smoothness or explicitly name a replacement theory for singularities; then compute the bundle rank and truncation sequence. If a paper studies flexes, osculating spaces or inflection, separately identify the selected section space and the map into the bundle. This prevents a case-specific rank bound from being mistaken for a universal structural role.^[1]^[3]^[4]

The algebraic vector-bundle realization also gives a safe ontology step. A finite locally free sheaf has an associated total space over \(X\) with local affine-space trivializations. That realizes the live Fiber Bundle parent signature; the sheaf as a module should not be described as though it were literally the total space without this construction.^[5]

Examples¶

Smooth plane quartic. A smooth plane quartic is a genus-three curve whose canonical line bundle is the restriction of the plane hyperplane bundle. Its three-dimensional canonical section space maps into the order-two jet/principal-parts bundle, which has rank three on a curve. A determinant rank drop yields the Wronskian ramification locus; in this case the Weierstrass points coincide with flexes. The choice of canonical sections and the flex interpretation are additional to the bundle's definition.^[1]

Smooth projective scroll. On a smooth scroll embedded in projective space, the input is its hyperplane line bundle. The principal-parts bundle has the ordinary smooth-base rank \(\binom{d+k}{k}\), while the selected embedding sections have a scroll-constrained fiberwise image. Under the source paper's hypotheses, points where that image rank drops below \(kd+1\) form its \(k\)th inflectional locus. A different scroll or section choice changes the map's special behavior, not the definition of the carrier.^[4]^[3]

Structural Tensions¶

No conflict between local and global data is required to define the bundle. The local Taylor description and global sheaf construction are compatible views of the same object. The source papers do expose an application-specific diagnostic: a selected section space may fail to fill the available jet target. On scrolls, geometry constrains the image even though the principal-parts bundle keeps its full rank. This is a property of the chosen map and setting, not an intrinsic all-instance opposition in the bundle.^[1]^[4]

A singular base creates a boundary rather than an all-instance tension. The principal-parts sheaf may fail to be locally free; locally free substitutes require further hypotheses and constructions. The present identity keeps smoothness explicit rather than silently importing singular-curve replacement results.^[1]

Structural–Framed Character¶

Vocabulary travel: “jets” and “principal parts” may denote equivalent smooth-context carriers, while related jet groups and singular replacements have different identities. Evaluative weight: whether a chosen linear system detects ramification or an embedding has the desired osculation is a question about that application, not the base bundle's existence. Institutional origin: this is a formal algebraic-geometric construction documented by the Stacks Project and the cited research papers.^[2]^[1]

Human-practice dependence: the construction is mathematical; practitioners choose \(X\), \(L\), \(n\) and optional sections to solve a problem. Import versus recognition: recognizing a principal-parts bundle requires the order-indexed sheaf construction and smooth local-freeness for this entry; calling any list of derivatives or any rank-drop set a “jet bundle” imports the name without the object. Its character: a portable order-indexed, locally compatible finite-jet carrier over a base, realized here as an algebraic vector bundle; applications depend on additional maps.^[1]^[3]

Structural Core vs. Domain Accent¶

The core is the smooth base, invertible input, finite order, infinitesimal-diagonal construction, locally free carrier and compatible truncation tower. Curve ramification and scroll inflection are unlike accents that use selected section maps. Removing those maps leaves the bundle. Removing the order-indexed diagonal formation rule or applying the smooth rank formula to an unqualified singular base loses the admitted identity.^[1]^[3]^[4]

The object remains domain-specific to algebraic geometry. A future Prime question is whether the order-indexed local-data carrier and compatible truncation tower recur with the same roles outside algebraic geometry. These sources establish only the algebraic-geometric construction, so no Prime promotion is claimed.^[1]

This entry is a kind of Fiber Bundle.

Fiber Bundle — broader in every case. Each bundle of principal parts is a kind of fiber bundle: finite local freeness gives an associated algebraic vector-bundle total space over \(X\), locally a product with compatible linear transitions. The bundle of principal parts adds the line-bundle input, order and thickened-diagonal rule.^[1]^[5]
Line Bundle — input/order-zero case. \(L\) is the input and \(P^0_X(L)=L\); the higher-order bundles generally have greater rank and are not all line bundles.^[1]
Projective Bundle and Associated Bundle — distinct constructions. Projectivizing fibers or specifying a principal-bundle quotient adds data absent from \(P^n_X(L)\).^[1]
Holomorphic Vector Bundle — narrower category. A complex analytic description may apply in suitable cases; the algebraic construction is not limited to that category.^[1]
Jet Group — different jet object. Jets of invertible maps at a fixed origin form a group, unlike the varying bundle of section data.^[1]

Relationships to Other Abstractions¶

Current abstraction Bundle of Principal Parts Domain-specific

Parents (1) — more general patterns this builds on

Bundle of Principal Parts is a kind of Fiber Bundle Domain-specific

On a smooth base, finite local freeness makes the principal-parts sheaf an algebraic vector bundle, a Fiber Bundle subtype.

Hierarchy path (1) — routes to 1 parentless root

Bundle of Principal Parts → Fiber Bundle → Mathematical structure → Set and Membership

Neighborhood in Abstraction Space¶

Bundle of Principal Parts sits in a sparse region of the domain-specific corpus (79^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2551 abstractions)

Nearest neighbors

Ideal sheaf — 0.83
Étale morphism — 0.82
Ringed Space — 0.82
Zariski Tangent Space — 0.82
Picard Group — 0.82

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

The principal-parts bundle is not an evaluation map, a Wronskian, a ramification or inflection locus, or a promise that every arbitrary principal-parts sheaf over a singular scheme is locally free. State the base, input, order and smoothness before transferring a rank or exact-sequence claim.^[1]^[4]

References¶

[1] Letterio Gatto and Andrea T. Ricolfi, “Jet bundles on Gorenstein curves and applications”, 2019 author-authored survey, §1.3 Proposition 1.3, §1.4 Lemma 1.8, and §§3.1–3.3. Gives the diagonal construction, smooth locally free sequence and classical curve examples. Its historical curve results are surveyed rather than newly proved here. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v ↩w ↩x ↩y ↩z ↩27 ↩28 ↩29 ↩30

[2] The Stacks Project, “Finite order differential operators,” §17.29, especially Lemma 17.29.3, Definition 17.29.4 and Lemma 17.29.6. Defines modules of principal parts, their order-surjections and the first-order exact sequence. The higher-order smooth local-freeness formula is supported by the other sources. registry ↩a ↩b ↩c ↩d

[3] Raquel Mallavibarrena and Ragni Piene, “On fundamental forms and osculating bundles”, 2024, §2. Under its stated smooth embedded-setting hypotheses, gives the binomial rank, exact sequence and truncated Taylor description. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l

[4] Antonio Lanteri, Raquel Mallavibarrena and Ragni Piene, “Inflectional loci of scrolls”, 2007, §1. Original scroll study defining the section-to-principal-parts map, osculating spaces and the inflection locus under its numerical and generic-rank conditions. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l

[5] The Stacks Project, “Vector bundles,” §27.6, Definition 27.6.1. Associates a relative-Spec vector-bundle total space to a quasi-coherent sheaf; finite local freeness supplies local affine-space trivializations. registry ↩a ↩b