Bundle Gerbe¶
A geometric gerbe presentation built from a surjective submersion, a line bundle on paired fibers, and associative multiplication over triples.
Core Idea¶
A bundle gerbe is a concrete geometric presentation over a manifold M. Choose a surjective submersion Y → M, form the pair fiber product Y[2] = Y ×_M Y, put a complex line bundle L on Y[2], and supply an isomorphism L(y₁,y₂) ⊗ L(y₂,y₃) → L(y₁,y₃) over each compatible triple in Y[3]. This multiplication must be associative on compatible quadruples. Murray's original formulation uses the equivalent principal C×-bundle language. The presentation is more than a line bundle or an element of cohomology: the submersion, pairwise line data and coherent product are all part of the object.[1][2]
Every bundle gerbe has a Dixmier–Douady class in H³(M, ℤ), which may be zero. A compatible connection and curving are optional differential-geometric additions. Their curvature three-form represents the image of the integral class in real cohomology, rather than recording every integral torsion detail. Raw isomorphism of bundle-gerbe presentations is finer than their classification by H³; the cohomology correspondence uses stable isomorphism in the cited theory.[1][3][2]
Structural Signature¶
Signature: base manifold + surjective submersion + line bundle on pair fibers + associative triple multiplication → bundle gerbe; the Dixmier–Douady class follows as an invariant.
- Base and surjective submersion. Y → M provides local presentation data and the pair, triple and quadruple fiber products. Without a common base, the pairwise construction is undefined.[1][2]
- Pair-product line bundle. L assigns a one-dimensional complex line to each pair (y₁,y₂) lying over the same point of M. A line bundle merely on M is not this required constituent.[1][2]
- Associative multiplication. Pairwise data compose over triples, and the two ways of composing over a quadruple agree. A line bundle on Y[2] without this coherence is not yet a bundle gerbe.[1][2]
- Dixmier–Douady invariant. The integral degree-three class records twisting; in a central-extension lifting construction it is the obstruction to a lift. A zero class is allowed and does not erase the defining data.[1]
- Optional connection and curving. When specified, these add differential information and a curvature three-form. They are not admission requirements for a bare bundle gerbe.[1][2]
What It Is Not¶
It is not a bare class in H³(M, ℤ). A class can record stable equivalence while omitting the chosen submersion, line bundle and multiplication. It is not simply a line bundle: the line bundle is an internal part on Y[2], while the whole object includes descent-style composition over triples. Nor is the curvature three-form identical to the full integral class; passage to real cohomology cannot retain torsion information.[1][3]
Murray distinguishes a bundle gerbe from a sheaf of groupoids: a bundle gerbe gives rise to an abstract gerbe, but the two are not literally the same type of object. The live Gerbe entry uses stack-theoretic local groupoids, so that relation alone does not establish strict subsumption of the concrete presentation by the live abstract node.[1]
Scope of Application¶
The definition applies in smooth geometry over a specified base manifold with a suitable submersion and complex line/C× bundle on its pair fiber product. Murray's lifting construction starts from a principal G-bundle and a central C× extension of G. Murray and Stevenson later construct a basic bundle gerbe on specified unitary-group bases from spectral data. Both satisfy the formal roles but serve different questions: obstruction to lifting versus an explicit basic gerbe and its differential form.[1][2]
Connections, curving, Wess–Zumino motivation and a particular three-form normalization belong to chosen constructions, not every bundle gerbe. The cited unitary-group paper does not warrant a claim about arbitrary compact Lie groups or every spacetime B-field.[2]
Clarity¶
Three levels should be kept separate. First is the presentation (Y, L, multiplication). Second is an invariant: its integral Dixmier–Douady class. Third is a differential refinement when connection and curving have been chosen. A class alone does not reconstruct a particular raw presentation; a three-form curvature supplies only real-image information. Murray explicitly notes that different bundle-gerbe presentations can yield isomorphic associated gerbes without being isomorphic as bundle gerbes. The later stable-isomorphism theorem explains the H³ classification at the appropriate equivalence level.[1][3]
This also clarifies the relation to a line bundle. L lives on pairs of points above M and supplies a necessary internal constituent, but no line bundle by itself has the bundle-gerbe multiplication over triples and quadruples.[2]
Manages Complexity¶
The formalism packages a degree-three twisting question into explicit pairwise line data and a finite coherence test. To recognize the object, examine Y[2] for the line bundle, Y[3] for its multiplication and Y[4] for associativity. In Murray's lifting example, these local ingredients turn a global extension-lift question into a Dixmier–Douady obstruction. In the unitary-group construction, the same structural checklist organizes spectral pair data and later curvature calculations.[1][2]
This reduction does not collapse distinct equivalence notions. Replacing a presentation by a stably isomorphic one can preserve its class even though raw bundle-gerbe isomorphism fails. If the question concerns a chosen connection or holonomy, the bare H³ class is insufficient; the extra differential data must be retained.[1][3]
Abstract Reasoning¶
Given a proposed bundle gerbe, write down M and Y → M. Verify that the claimed L or principal C× bundle is over Y[2], not only over M. Exhibit the multiplication over Y[3] and check associativity on Y[4]. Only after those checks compute or invoke its Dixmier–Douady class. Ask whether a zero class signifies a trivializable presentation in the relevant sense, and use stable rather than raw isomorphism when applying the H³ classification theorem.[1][3]
For differential claims, separately identify a compatible connection and curving and state whether the resulting three-form represents the real image of an integral class. The form by itself cannot distinguish integral torsion classes.[1][2]
Knowledge Transfer¶
The object-level test transfers literally from central-extension lifting gerbes to Murray–Stevenson's basic unitary-group gerbe: both have a submersion, line bundle on pair fibers and associative multiplication. What changes is how those data are constructed and which invariant or question is emphasized. The first example detects a lift obstruction; the second supports explicit geometry of a basic class and a differential representative.[1][2]
Outside smooth gerbe geometry, pairwise data with coherent composition may recur as an analogy. Without the submersion and line-bundle construction, it is not literally this named object. The related abstract Gerbe entry concerns the associated stack-like structure; the live Line Bundle parent supplies the required internal rank-one constituent, not the entire gerbe.
Examples¶
Lifting bundle gerbe. Given a principal G-bundle Y → M and a central extension C× → Ĝ → G, Murray uses the extension to form pairwise C×-torsor data above Y[2]. Group multiplication induces the bundle-gerbe product. Its Dixmier–Douady class obstructs lifting Y to a principal Ĝ-bundle; the bundle gerbe is trivial in the relevant sense precisely when that lift exists.[1]
Mapped back: base/submersion → the principal G-bundle Y → M; pair-product line bundle → extension lifts of pairwise G differences; associative multiplication → the extension's group law; class → lift obstruction; connection/curving → not required for this bare example. The central extension is this construction's accent, not a universal source of all bundle gerbes.[1]
Basic bundle gerbe on a unitary group. Murray and Stevenson construct a bundle gerbe from spectral data over specified unitary-group bases, then add a connection and curving. They calculate a basic three-form representing the real Dixmier–Douady class in their setting.[2]
Mapped back: base/submersion → the chosen unitary group and its spectral-choice space; pair-product line bundle → the line data over paired spectral choices; associative multiplication → the product satisfying Definition 2.1; class → the basic integral twisting class whose real image is calculated; connection/curving → extra differential data chosen in the paper. This is an explicit unitary-group case, not a general claim about all WZW targets.[2]
Structural Tensions¶
The cited sources do not establish an intrinsic pair of opposing aims that every bundle gerbe must balance. Raw isomorphism and stable isomorphism are different equivalence questions, not competing forces within the object. The diagnostic is which comparison a problem asks for: a chosen geometric presentation or its stable H³ twisting class. Raw isomorphism retains presentation detail; stable isomorphism is the equivalence used for the cited classification. Choosing the wrong level gives the wrong answer to that problem, but neither level is a constitutive tradeoff of a bundle gerbe.[1][3]
Structural–Framed Character¶
The construction is strongly structural within geometry: the same submersion, pair-line and associative-product roles occur in lifting and unitary-group cases. It has little evaluative weight; a nonzero obstruction or class is a mathematical fact under assumptions, not praise or blame. Human mathematical practice chooses models and equivalence conventions, while the coherence identities are proved or checked once those choices are made. No institution confers bundle-gerbe status. Vocabulary such as “gerbe” and “bundle” travels across areas, but importing this exact named object requires the concrete pair-fiber-product presentation rather than a resemblance to layered information. The broader stack-theoretic Gerbe may be associated to this object, yet the live entry is not the same presentation. Its character: a domain-specific geometric object whose role pattern travels across cases in its field, with no demonstrated substrate-independent bundle-gerbe Prime.[1][2]
Structural Core vs. Domain Accent¶
The skeletal relation is a line bundle of pairwise comparison data with coherent composition over triples above a base. The domain-bound mechanism includes a smooth surjective submersion, complex line/C× fibers and associative multiplication; without those, the named object is lost. A central-extension group, spectral choices on a unitary group, optional connection and curving, and particular curvature normalization are accents of examples.[1][2]
The general idea of a line bundle is already represented by live Line Bundle, here an identity-bearing part through composition/part_of with the parent inside the child. A broader coherent local-to-global assembly pattern is a future Prime question, not a portable identity established by these two geometric examples. Unlike nongeometric carriers and their role mappings would be needed before asserting such a Prime or a direct edge from Bundle Gerbe. Live Local-to-Global Aggregation appears upstream of Line Bundle, but the part-of relation here does not automatically inherit that Prime for the whole bundle gerbe. The separate live Fiber Bundle relation missing upstream of Line Bundle is a catalog issue, not a reason to duplicate this entry's direct constituent edge.
Instantiates / Related Primes¶
This entry is part of Line Bundle.
Every bundle gerbe contains a Line Bundle as a part: the line bundle lies over Y[2], and the bundle gerbe adds Y → M and associative multiplication. Gerbe is a related abstract stack-theoretic object: Murray constructs an associated gerbe from a bundle gerbe, but that does not make the concrete presentation literally the stack. Fiber Bundle is a broader description of the internal line bundle, not a separate broader abstraction for the same constituent.[1][2]
Relationships to Other Abstractions¶
Current abstraction Bundle Gerbe Domain-specific
Parents (1) — more general patterns this builds on
-
Bundle Gerbe is part of Line Bundle Domain-specific
A complex line bundle on the pair fiber product is an identity-bearing internal constituent of every bundle gerbe.The bundle-gerbe definition contains a complex line bundle over Y×_M Y, equivalently a principal C× bundle in Murray's presentation. Live Line Bundle names this rank-one internal constituent; the bundle gerbe adds the surjective submersion and associative multiplication. The bundle-gerbe object is not itself a line bundle, so composition/part_of has parent_in_child direction. The live Fiber Bundle entry is a broader description of that same constituent, not a second independent direct component.
Hierarchy path (1) — routes to 1 parentless root
- Bundle Gerbe → Line Bundle → Local-to-Global Aggregation
Neighborhood in Abstraction Space¶
Bundle Gerbe sits in a sparse region of the domain-specific corpus (84th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Vector Bundles & Classifying Constructions (10 abstractions)
Nearest neighbors
- Associated bundle — 0.82
- Holomorphic vector bundle — 0.82
- Bundle metric — 0.82
- Flat Vector Bundle — 0.82
- Dirac Structure — 0.81
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
A nonzero H³ class is not required: trivial bundle gerbes still have the defining data. Bare cohomology is not the presentation. A line bundle on M is not the required line bundle on Y[2], and a line bundle on Y[2] without associative multiplication is incomplete. Connection and curving are optional, and their three-form records the real image, not all integral torsion. The H³ correspondence uses stable isomorphism rather than raw bundle-gerbe isomorphism.[1][3][2]
References¶
[1] Michael K. Murray, Bundle Gerbes, original author preprint (25 July 1994), arXiv dg-ga/9407015, 19-page PDF. Especially Introduction PDF p. 3, §3 definition PDF pp. 7–8, §4 central-extension lifting PDF pp. 8–9, §5 Dixmier–Douady class, and §§6–8 connection/curvature. The later 1996 journal publication belongs to the same work lineage, not a second independent source. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v ↩w
[2] Michael K. Murray and Danny Stevenson, The Basic Bundle Gerbe on Unitary Groups, original author preprint (2008), arXiv 0804.3464. Especially Definition 2.1 PDF pp. 4–5, §3 construction, §4 connection, and §5 curving/curvature/class calculation. The group and normalization scope are those of the paper; connection and curving are additional data. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r
[3] Michael K. Murray and Daniel Stevenson, Bundle Gerbes, Stable Isomorphism and Local Theory, original author preprint (1999), arXiv math/9908135, §2 locally split maps and §3 Proposition 3.2 and Theorem 3.3, PDF pp. 4–5. Original title uses a colon after “Gerbes”; the comma in the link label preserves the full binder title. The H³ bijection is for stable-isomorphism classes. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g