Transport of Structure

Define operations, relations, or other structure on one mathematical carrier through a chosen equivalence so that the equivalence becomes structure-preserving by construction.

Version

v1 · 2026-08-30 · History

Domain-specific #

2993

Origin domain

mathematics

Subdomain

structural mathematics

Aliases

Transport of structures, Transference of structure, Transported structure

Core Idea

Transport of Structure is the mathematical construction that equips one carrier with operations, relations, topology, geometry, or another specified structure by carrying an existing structure across a chosen invertible identification. Given a structured object \((A,S)\), an otherwise unstructured carrier \(B\), and a bijection or appropriate equivalence \(f:A\to B\), define the structure \(f_*S\) on \(B\) so that \(f:(A,S)\to(B,f_*S)\) becomes an isomorphism by construction. Bourbaki made this operation explicit in its treatment of species of structures, isomorphisms, and transportable relations; universal algebra expresses the same construction through operations and identities.^[1][2]

For an \(n\)-ary operation \(\omega_A:A^n\to A\), the transported operation is

\[ \omega_B(b_1,\ldots,b_n) = f\!\left(\omega_A(f^{-1}b_1,\ldots,f^{-1}b_n)\right). \]

For an \(n\)-ary relation \(R_A\), define

\[ R_B(b_1,\ldots,b_n) \quad\Longleftrightarrow\quad R_A(f^{-1}b_1,\ldots,f^{-1}b_n). \]

Constants move by \(c_B=f(c_A)\). Any equations or first-order relations built from the signature are preserved because the defining formulas make \(f\) preserve and reflect the structure.

The construction is canonical relative to the specified map and structural kind. The bare assertion that \(A\) and \(B\) have the same cardinality does not pick a preferred transported structure, because different bijections can induce different structures on \(B\). A genuinely canonical identification can make the result canonical in context; an arbitrary identification merely makes it well-defined after a choice. This dependence is part of the abstraction, not a defect to hide.

Structural Signature

A complete instance has eight roles:

• source carrier \(A\) — the underlying set, space, object, or type already bearing structure;
• source structure \(S\) — named operations, constants, relations, open sets, charts, measures, or other data with axioms;
• target carrier \(B\) — an object on which the corresponding structure has not yet been fixed for the purpose at hand;
• transport map \(f:A\to B\) — ordinarily a bijection, isomorphism at a weaker structural level, equivalence, or equality path;
• inverse or lifting mechanism — \(f^{-1}\), quasi-inverse data, or a universal property sufficient to move structural inputs back to \(A\);
• transport rule \(f_*S\) — definitions on \(B\) obtained by conjugating operations, transferring relations, or moving admissible local data;
• compatibility result — \(f\) becomes an isomorphism or prescribed homomorphism of the newly structured objects;
• choice and coherence record — the construction states whether the result is unique, canonical only relative to \(f\), independent of a choice, or coherent under composition.

The standard invariants are identity, composition, and recovery:

\[ (\operatorname{id}_A)_*S=S, \qquad (g\circ f)_*S=g_*(f_*S), \qquad (f^{-1})_*(f_*S)=S. \]

These are direct consequences of substitution for strict bijective transport. They show that transport is not arbitrary redescription: it acts coherently along a chain of identifications.

The recognition boundary is directional. An isomorphism between two already structured objects verifies sameness. Transport uses an identification at the carrier or weaker-structure level to define the missing or selected target structure. Before the definition, \(f\) need not be a morphism of the intended structured kind because that target structure does not yet exist.

What It Is Not

Transport of Structure is not merely Isomorphism. Isomorphism asks whether a map between two already structured objects preserves structure in both directions. Transport starts with structure on one side and creates the corresponding structure on the other, thereby arranging that the chosen map is an isomorphism. The existence test and the definition-making operation are related but not interchangeable.

It is not an arbitrary bijection. A bijection provides enough set-level reversibility to transport finitary operations and relations, but no structure is moved until explicit formulas are supplied. Conversely, when the map is not bijective, an operation cannot generally be transported without additional descent, extension, congruence, density, continuity, or universal-property hypotheses.

It is not simply pullback or pushforward. Many transports are expressed as pullbacks or pushforwards: an inner product can be pulled back along a linear isomorphism, and a topology can be pushed forward along a bijection. Generic pullback or pushforward, however, can exist along noninvertible maps, can be one-sided, and need not make the map an isomorphism. Transport of a complete structure emphasizes equivalence and recovery.

It is not just a change of coordinates. A coordinate change usually re-expresses an already fixed geometric object in another chart or basis. Transport can instead equip a new carrier with structure it did not previously have. A change of basis is an instance only when the new representation is explicitly defined by conjugating the structure through the basis isomorphism.

It is not a claim that the induced structure is naturally preferred. Given the cyclic group \(\mathbb Z/4\mathbb Z\) and an arbitrary bijection to a four-element set, the target becomes a cyclic group, but nothing about the unlabeled four-element set singles out that group law or that bijection. “Canonical by transport” always requires the phrase “relative to which identification?”

Finally, it is not theorem transfer alone. Once \(f\) is an isomorphism, propositions invariant under that structure transfer between source and target. Transport of Structure is the prior construction that makes this license available.

Scope of Application

The home domain is structural mathematics. Bourbaki's set-theoretic presentation treats structures through scale sets, transportable relations, species of structures, and isomorphisms.^[1] Universal algebra supplies the clearest formula: a signature lists operations and constants, equations specify a class of algebras, and conjugation through a bijection transports every operation while preserving the equations.^[2]

In linear algebra and analysis, linear isomorphisms transport inner products, norms, bilinear forms, orientations, and operators. If \(f:V\to W\) is a linear isomorphism and \(W\) has inner product \(\langle-,-\rangle_W\), then

\[ \langle v_1,v_2\rangle_V =\langle f(v_1),f(v_2)\rangle_W \]

defines the unique inner product on \(V\) making \(f\) an isometry.

In topology, a bijection \(f:X\to Y\) transports \(\tau_X\) to \(\tau_Y=\{f(U):U\in\tau_X\}\), making \(f\) a homeomorphism. In differential topology, a homeomorphism \(h:X\to M\) can transport a smooth atlas from \(M\) to \(X\): charts \((U,\varphi)\) become \((h^{-1}U,\varphi\circ h)\), and the transition maps remain the original smooth transitions.^[3]

In category theory, transport appears whenever algebraic structures are treated as objects over an underlying category. Mac Lane's algebraic theories and their algebras formalize operations as product-preserving functors.^[4] Holm proves broader ascent and descent results: under suitable product-preserving adjoint situations, algebraic structures can move along completion, Stone–Čech compactification, universal covering, and universal locally connected refinement maps, even when the underlying map is not simply a bijection.^[5] These are generalized transports with extra hypotheses, not evidence that arbitrary maps suffice.

In dependent type theory, an equality path \(p:x=_A y\) transports an inhabitant of a dependent family \(P(x)\) to one of \(P(y)\). Univalence and the structure identity principle connect equivalence with identity and explain why equivalent structured objects support coherent transport.^[6] This is a foundational generalization of the same role pattern, though its identity/path machinery should not be conflated with ordinary set-theoretic conjugation.

Clarity

Six questions make a transport claim precise:

1. What structure is being moved? Name the signature, relation family, topology, atlas, metric, measure, order, or categorical structure.
2. Which direction is transport occurring? Specify source, target, and whether notation is pullback \(f^*\) or pushforward \(f_*\).
3. What property of the map permits transport? Bijection, homeomorphism, linear isomorphism, equivalence, covering map, completion map, or an adjoint universal property are different hypotheses.
4. What are the target definitions? Write the transported operations, relations, charts, or universal characterization explicitly.
5. What compatibility is obtained? State whether the map becomes an isomorphism, homomorphism, continuous map, local diffeomorphism, or something weaker.
6. In what sense is the result canonical or unique? Relative uniqueness for a fixed map is not independence from all possible identifications.

The most common ambiguity is the word “same.” If \(A\) and \(B\) are merely equipotent, any selected bijection permits set-based transport, but the output may depend on the selection. If a problem supplies a distinguished map—such as a quotient universal map, completion embedding, or canonical equivalence—the construction may acquire a stronger naturality claim. The draft should never infer “canonical” from existence alone.

A second ambiguity concerns the level of structure. A homeomorphism can transport a smooth structure so that it becomes a diffeomorphism for the transported atlas, but it need not be a diffeomorphism relative to a pre-existing smooth structure on the target. The weaker identification and stronger transported structure must be named separately.

Manages Complexity

Transport of Structure replaces repeated axiom verification with a single structural argument. Defining a group law on a new set directly would require checking associativity, identity, and inverses. Defining it by conjugation through a bijection makes those laws follow from the source group because every target expression translates back to a source expression. The construction compresses “reprove every law” into “verify the transport map and formula.”

It also separates carrier questions from structure questions. A set can carry many group laws, topologies, orders, or smooth structures. Transport forces the analyst to record which layer is fixed, which layer is being defined, and which identification ties them together. That discipline is particularly important when one underlying space admits inequivalent stronger structures.

The composition law supports modular construction. If structure moves from \(A\) to \(B\) and then \(B\) to \(C\), there is no need to expand every definition twice; direct transport along \(g\circ f\) gives the same result. In formal mathematics and software libraries, this coherence prevents duplicated instances and incompatible definitions.

Finally, the abstraction localizes failure. If transport fails, ask whether the map lacks an inverse, whether operations fail to descend through identified fibers, whether extension from a dense subobject lacks continuity, whether charts fail compatibility, or whether different choices fail coherence. The failure is attached to a missing structural role instead of being dismissed as “not natural.”

Abstract Reasoning

The core reasoning move is conjugation. To define or compute a target operation, move the inputs back to the source, apply the known operation, and move the result forward. This immediately predicts preservation of equations. For example, associativity on \(B\) follows by expanding both \((b_1*_B b_2)*_B b_3\) and \(b_1*_B(b_2*_B b_3)\) through \(f^{-1}\), using associativity on \(A\), and applying \(f\).

Transport licenses uniqueness reasoning. For a fixed bijection \(f\) and fixed source algebra, there is exactly one target operation of the given arity that makes \(f\) a homomorphism: the conjugation formula is forced. Similar uniqueness holds for a transported topology making \(f\) a homeomorphism and for an inner product making a linear isomorphism an isometry.

It licenses choice-sensitivity analysis. If \(f,g:A\to B\) are different bijections, compare \(f_*S\) and \(g_*S\) via the permutation \(g\circ f^{-1}:B\to B\). The two target structures are isomorphic, but they need not be equal as structures on the same carrier. They coincide precisely when the change between the identifications respects the relevant source structure. Thus “unique up to isomorphism” is weaker than “the same transported structure.”

It also licenses obstruction reasoning for noninvertible maps. A surjection \(q:A\to B\) can carry a group law down only when its fibers form a congruence—equivalently, in groups, when they arise from cosets of a normal subgroup. Otherwise \(q(a_1a_2)\) depends on representatives. An injection can restrict operations only when the image is closed under them. Generalized transports therefore require structural hypotheses that replace an inverse.

The final move is invariance: once the target is transported, every statement formulated purely in the transported structural language has the same truth value on source and target. Claims involving extrinsic embeddings, computational cost, labels, or an unrelated pre-existing structure need not transfer.

Knowledge Transfer

The construct transfers literally across mathematical subfields. Groups, rings, modules, orders, topologies, uniform structures, inner products, atlases, group actions, algebraic theories, and dependent families all use the same source–identification–definition–compatibility pattern. The surface formulas differ, but the operation remains transport rather than analogy because the target structure is defined through an exact map and preservation is proved.

Holm's treatment shows that the idea extends beyond bijective relabeling. Algebraic structures can ascend or descend along maps supplied by universal constructions when adjunction and finite-product preservation guarantee compatible operations.^[5] The broader lesson is not “every map transports everything,” but that an inverse can be replaced by a sufficiently strong universal mechanism.

Outside formal mathematics, similar practices—copying a schema through a reversible encoding or giving an API wrapper operations by delegation—may instantiate Isomorphism, Mapping, or Equivalence-Preserving Rewriting. They ordinarily lack a formally specified species of mathematical structure and proof of transported axioms. Calling them Transport of Structure is useful only when those formal roles are present; otherwise it is analogy.

That limitation fixes the classification. The construction is broad within mathematics but not an independent cross-substrate prime. Its portable skeleton is already owned by Isomorphism and Transformation; its autonomous residual is the mathematically exact act of defining target structure through an equivalence.

Examples

Transporting a group law. Let \((A,\cdot,e)\) be a group and \(f:A\to B\) a bijection. Define

\[ b_1 *_B b_2=f(f^{-1}(b_1)\cdot f^{-1}(b_2)), \quad e_B=f(e), \quad b^{-1_B}=f((f^{-1}b)^{-1}). \]

Then \((B,*_B,e_B)\) is a group and \(f\) is a group isomorphism. Every group axiom transfers by substitution. A different bijection may produce a different written operation on the same set \(B\).

Inner product. Given a linear isomorphism \(f:V\to W\) and an inner product on \(W\), define \(\langle v_1,v_2\rangle_V=\langle f(v_1),f(v_2)\rangle_W\). Positivity, conjugate symmetry, and linearity follow immediately, and \(f\) becomes an isometry. If \(f\) is not injective, the pullback form is degenerate on its kernel; invertibility is therefore essential to full transport.

Topology. Given a topology \(\tau_X\) on \(X\) and bijection \(f:X\to Y\), the family \(f_*\tau_X=\{f(U):U\in\tau_X\}\) is a topology on \(Y\). Unions and finite intersections are preserved by images under a bijection, and \(f\) is a homeomorphism. The result is unique among topologies on \(Y\) with that property.

Smooth atlas. If \(M\) is a smooth manifold and \(h:X\to M\) a homeomorphism, transport every chart \((U,\varphi)\) to \((h^{-1}U,\varphi\circ h)\). Overlaps have the same coordinate transition functions as the source atlas, so smooth compatibility follows. The construction makes \(h\) a diffeomorphism relative to the transported smooth structure; it says nothing about compatibility with some other smooth structure already placed on \(X\).^[3]

Completion and covering spaces. Holm formalizes situations in which algebraic operations extend uniquely to a metric completion or descend to a universal cover so the canonical map becomes a homomorphism.^[5] These examples exhibit generalized transport: continuity, adjunction, and product preservation do the work otherwise supplied by an inverse bijection.

Dependent transport. For a type family \(P:A\to\mathcal U\) and path \(p:x=_A y\), dependent type theory provides a function \(\operatorname{transport}^P(p):P(x)\to P(y)\), reducing to the identity when \(p\) is reflexivity.^[6] This is not the same syntax as Bourbaki transport, but it preserves the exact structural pattern of moving inhabitants or structures along an identification with coherence under path composition.

Structural Tensions

Canonical result versus chosen identification. A fixed \(f\) forces the transported definitions, but a bare target carrier may admit many choices of \(f\). Diagnostic: is the map part of the data, and is it itself canonical or arbitrary?

Existing structure versus newly transported structure. A target may already have a topology, group law, or atlas. The transported structure can conflict with it. Diagnostic: compare the structures before declaring the original map a homeomorphism, isomorphism, or diffeomorphism.

Strict equality versus isomorphism. Two choices of transport produce isomorphic target objects but need not produce identical operations on the same underlying set. Diagnostic: does the application require literal definitional equality, canonical isomorphism, or merely existence of some isomorphism?

Invertible transport versus generalized ascent/descent. Conjugation along a bijection is automatic; extension along completion or descent along quotient/cover needs extra hypotheses. Diagnostic: what replaces \(f^{-1}\), and what theorem proves existence and uniqueness?

Internal structure versus extrinsic presentation. Transport preserves the named internal structure, not necessarily how the target sits in an ambient space. A transported smooth atlas on a topological cone may not agree with the smoothness suggested by its particular embedding. Diagnostic: which features belong to the transported language and which are extrinsic?

Convenient definition versus naturality. Transport can make proofs easy while producing a definition dependent on a non-natural choice. Diagnostic: does the construction commute with the relevant maps and symmetries, or only work after fixing coordinates?

Structural–Framed Character

Transport of Structure is mixed-structural within its domain. Its core formulas are formal, exact, and evaluatively neutral. Once the structural kind and identification are specified, no institutional or interpretive judgment decides the result. Identity, composition, recovery, and axiom preservation are structural facts.

Its framing is mathematical and load-bearing. “Carrier,” “signature,” “operation,” “relation,” “bijection,” “isomorphism,” “atlas,” “algebraic theory,” and “dependent family” are not optional accents; they specify what may be transported and what counts as preservation. Outside mathematics, looser copying and migration practices do not automatically satisfy this signature.

The naturality question gives the abstraction a small framed component even inside mathematics. Calling a construction “canonical” often records that it avoids arbitrary choices or behaves naturally, and that assessment depends on the surrounding category and admissible maps. The underlying transport formula is structural; the importance assigned to one identification over another comes from the problem's framing.

Structural Core vs. Domain Accent

The structural core is: move data across a reversible correspondence by conjugating all relevant operations and relations, so the correspondence becomes structure-preserving and the defining laws survive. Identity, composition, and inverse recovery capture its reusable skeleton.

The domain accent specifies mathematical structures, formal axioms, carrier maps, isomorphisms, universal properties, and coherence. Those commitments distinguish exact transport from ordinary copying, translation, or analogy. If they are removed, the residual is already represented by Isomorphism, Mapping, Transformation, or Equivalence-Preserving Rewriting.

This candidate therefore passes domain-specific autonomy. It has a stable Bourbaki formulation, exact formulas, characteristic uniqueness and choice questions, use across algebra/topology/category theory/type theory, and generalized ascent/descent results. It fails the prime bar because its literal full mechanism remains inside formal mathematics and its portable skeleton has catalog coverage.

Instantiates / Related Primes

Transport of Structure presupposes Isomorphism as its target compatibility: it defines target structure so a chosen carrier equivalence becomes a structure-preserving invertible map. prime:isomorphism is the proposed sole DAG parent.

It is related to Transformation, because conjugation transforms structural data, and to Invariance, because equations and relations invariant under isomorphism survive. Those edges would be redundant: Isomorphism already has ancestry through Bijectivity and Invariance in the live DAG.

It is also related to Equivalence-Preserving Rewriting and Mapping, but transport does more than rewrite an expression or pair elements: it constructs a complete target structure with a preservation theorem. Functor is relevant only for categorical generalizations; simple algebraic transport requires no functor between categories.

Relationships to Other Abstractions

Current abstraction Transport of Structure Domain-specific

Parents (1) — more general patterns this builds on

• Transport of Structure presupposes Isomorphism Prime

  Transport of Structure presupposes Isomorphism as its target compatibility: it defines target structure so a chosen carrier equivalence becomes a structure-preserving invertible map.

Hierarchy paths (4) — routes to 2 parentless roots

• Transport of Structure → Isomorphism → Bijectivity → Function (Mapping)

Show alternative paths (3)

Neighborhood in Abstraction Space

Transport of Structure sits in a sparse region of the domain-specific corpus (80^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

• Category theory — 0.82
• Mathematical structure — 0.82
• Kernel — 0.82
• Lifting theory — 0.82
• Local diffeomorphism — 0.81

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With

• Isomorphism: compares already structured objects; transport defines one side to obtain an isomorphism.
• Structure preservation: a property of a map after structures are fixed, not the operation that defines a structure.
• Pullback / pushforward: directional constructions that can exist along noninvertible maps and may not yield equivalence.
• Change of basis or coordinates: usually re-expression of a fixed object; it is transport only when definitions are induced through the coordinate isomorphism.
• Conjugation: the core formula for algebraic operations, not the whole cross-structure concept.
• Initial or final topology: structures induced by arbitrary maps; along a bijection these coincide with transport, but the general constructions are broader and one-sided.
• Descent: asks when local or source data glue or factor through a map; extra effectiveness and coherence conditions are required.
• Extension by continuity: can transport operations to a completion when density and continuity yield existence and uniqueness, but it is not automatic set-level transport.
• Structure Identity Principle: a univalent-foundational principle connecting equivalence and identity of structures; it generalizes and explains coherent transport but is not a synonym for the classical operation.
• Theorem transfer: a consequence of the resulting isomorphism, not the definition of the target structure itself.

References

[1] Nicolas Bourbaki, Theory of Sets, Elements of Mathematics (Springer, English reprint 2004), Chapter IV, §1, especially “Transportable Relations,” “Species of Structures,” and “Isomorphisms and Transport of Structures.” Publisher record: https://doi.org/10.1007/978-3-642-59309-3 registry ↩a ↩b

[2] Stanley N. Burris and H. P. Sankappanavar, A Course in Universal Algebra (Springer, 1981; Millennium Edition, 2012), especially Chapter II on algebras, identities, and varieties. Author-hosted edition: https://www.math.uwaterloo.ca/~snburris/htdocs/ualg.html registry ↩a ↩b

[3] John M. Lee, Introduction to Smooth Manifolds, 2^nd ed., Graduate Texts in Mathematics 218 (Springer, 2013), Chapters 1–2 on smooth structures, atlases, and smooth maps. https://doi.org/10.1007/978-1-4419-9982-5 registry ↩a ↩b

[4] Saunders Mac Lane, Categories for the Working Mathematician, 2^nd ed., Graduate Texts in Mathematics 5 (Springer, 1998), Chapter V, §6 on algebraic theories. https://doi.org/10.1007/978-1-4757-4721-8 registry ↩

[5] Henrik Holm, “A Note on Transport of Algebraic Structures,” Theory and Applications of Categories 30, no. 34 (2015): 1121–1131. https://www.tac.mta.ca/tac/volumes/30/34/30-34.pdf; preprint https://arxiv.org/abs/1504.07366 registry ↩a ↩b ↩c

[6] The Univalent Foundations Program, Homotopy Type Theory: Univalent Foundations of Mathematics (Institute for Advanced Study, 2013), especially §§2.3 and 9.8 on transport and the structure identity principle. https://homotopytypetheory.org/book/ registry ↩a ↩b