Continuous Submanifold Fields

• Continuous Submanifold Fields are collections of smoothly varying geometric data—such as tangent subspaces, tensor fields, and implicit representations—designed to rigorously encode both intrinsic and extrinsic submanifold structures.
• They employ approaches like soldered tensor fields, tractor calculus, and flow-generated constructions to capture compatibility conditions and curvature constraints essential for integrability.
• These frameworks support deformation analysis and admissibility conditions in variational settings, linking theoretical formulations to practical applications in differential and geometric analysis.

Searching arXiv for recent and foundational papers directly relevant to the phrase and its nearest rigorous formulations. “Continuous Submanifold Fields” is not a standard term in the cited literature. The nearest rigorous usages occur in several adjacent frameworks: smoothly varying tangent subspace fields and foliations, tensor fields soldered to normalized submanifolds, natural fields canonically attached to immersions, implicit field representations whose regular zero loci are submanifolds, and deformation or variational settings in which admissible configurations form a constrained submanifold of a larger space (Chen, 2013, Vaisman, 2010, Chern et al., 15 Jul 2025). In these settings, the ambient objects are typically $C^\infty$, so the operative notion is usually smooth rather than merely continuous. A common theme is that a submanifold is encoded not only as an embedded subset or immersion, but also through a field whose tangential compatibility, normal behavior, or induced evolution records intrinsic and extrinsic geometry.

1. Terminological scope and geometric background

In the survey literature, the closest precise notions to “continuous submanifold fields” are smoothly varying tangent subspace fields, foliations by submanifolds, one-parameter families of immersions, parallel or equidistant hypersurface families, curvature surface foliations, and deformation or variation theory (Chen, 2013). The default regularity is smooth: manifolds are assumed connected, of class $C^\infty$, and without boundary unless stated otherwise. This is significant because most constructions discussed under this umbrella require differential operators, curvature tensors, or Lie derivatives.

The basic geometric data are those of a Riemannian submanifold $f:(M,g)\to (\tilde M,\tilde g)$: the tangent bundle, the normal bundle, the induced metric, the second fundamental form $h$, the shape operator $A_\xi$, and the mean curvature vector $H$. They are linked by the Gauss and Weingarten formulas,

$$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$

and by the Gauss, Codazzi, and Ricci equations. These are the compatibility equations that distinguish an actual submanifold geometry from an arbitrary field of tangent and normal data.

A decisive distinction in this literature is that between a distribution and an integrated family of submanifolds. A field of tangent subspaces $x\mapsto E_x\subset T_x\tilde M$ is only a distribution; it becomes a genuine family of submanifolds when it is integrable. This distinction recurs in the form of soldering conditions, orbit theorems, admissibility PDEs, or incidence and curvature constraints. In that sense, the phrase “continuous submanifold fields” is best interpreted as a family of rigorous mechanisms for attaching smoothly varying geometric data to submanifolds and for deciding when those data integrate to actual embedded or immersed geometry.

2. Soldered tensor fields and normalized submanifolds

A particularly direct formalization is Vaisman’s notion of a tensor field soldered to a normalized submanifold (Vaisman, 2010). If $N^n\subset M^m$ is a submanifold with a chosen normal bundle $\nu N$, the splitting

$C^\infty$0

defines a normalized submanifold $C^\infty$1. In adapted coordinates $C^\infty$2, the normal directions are represented by $C^\infty$3 and the tangent directions by $C^\infty$4 along $C^\infty$5.

For a tensor field $C^\infty$6, soldering means that for every vector field $C^\infty$7 normal to $C^\infty$8,

$C^\infty$9

for all $f:(M,g)\to (\tilde M,\tilde g)$0, all tangent vectors $f:(M,g)\to (\tilde M,\tilde g)$1, and all covectors $f:(M,g)\to (\tilde M,\tilde g)$2. Geometrically, this says that the tensor field does not change, in any tangentially visible way, when moved in normal directions.

The theory separates algebraic adaptation from full soldering. Algebraic adaptation is the value-level condition that along $f:(M,g)\to (\tilde M,\tilde g)$3, the tensor maps tangent vectors and tangent covectors to tangent-compatible outputs. In adapted coordinates, soldering is equivalent to the component conditions

$f:(M,g)\to (\tilde M,\tilde g)$4

together with

$f:(M,g)\to (\tilde M,\tilde g)$5

The first pair expresses algebraic adaptation; the last equation expresses vanishing first-order variation in normal directions.

Once $f:(M,g)\to (\tilde M,\tilde g)$6 is algebraically adapted, the remaining defect is measured by the soldering obstruction

$f:(M,g)\to (\tilde M,\tilde g)$7

Then $f:(M,g)\to (\tilde M,\tilde g)$8 is soldered if and only if $f:(M,g)\to (\tilde M,\tilde g)$9. This gives a general obstruction theory for ambient tensor fields restricted to submanifolds.

In the Riemannian case, soldering recovers a classical extrinsic notion exactly. If $h$0 is the ambient metric and $h$1 the second fundamental form, then

$h$2

so the metric is soldered if and only if $h$3, equivalently if and only if the submanifold is totally geodesic. The paper also proves that on an almost Kähler manifold $h$4, the almost complex structure $h$5 is soldered to a submanifold if and only if the latter is $h$6-invariant and totally geodesic. In this framework, a “field attached continuously to a submanifold” is an ambient tensor whose tangential restriction is compatible with the tangent-normal splitting and whose first normal jet has no tangentially visible defect.

3. Canonical tangent fields and tractor-geometric fields

A second line of work studies natural fields intrinsically induced by an immersion. For an isometric immersion $h$7, the position vector splits as

$h$8

and the tangential component $h$9 is called the canonical vector field of the Euclidean submanifold (Chen et al., 2017). Because the ambient position vector is concurrent,

$A_\xi$0

one obtains the identities

$A_\xi$1

and consequently

$A_\xi$2

The main characterization is exact: $A_\xi$3 is conformal if and only if the submanifold is umbilical with respect to the normal component $A_\xi$4, equivalently

$A_\xi$5

In codimension one this forces the hypersurface to lie in a sphere centered at the origin or in a hyperplane not containing the origin; in codimension two the paper gives a dichotomy between hyperplane or hypersphere containment and a conformally flat local “locus of $A_\xi$6-spheres” description.

A conformal analogue of this field-theoretic viewpoint is developed through tractor calculus for embedded submanifolds of conformal manifolds (Curry et al., 2023). There the central bundle-valued field is the tractor second fundamental form $A_\xi$7, defined through the conformal tractor Gauss formula

$A_\xi$8

A submanifold is called distinguished when

$A_\xi$9

This condition is equivalent to parallelity of the tractor normal form and of the normal tractor projector along the submanifold. It interpolates between two classical extremes: for curves it is exactly the conformal circle condition, and for hypersurfaces it is exactly total umbilicity. The paper further proves that distinguished submanifolds are precisely the weakly conformally circular submanifolds, meaning that ambient conformal circles remain in the submanifold. Stronger notions of conformal circularity are characterized by adding conditions on the Fialkow tensor $H$0.

These two theories differ in language—ordinary vector fields versus tractor fields—but they share a common structure. Each isolates a canonical field attached to the immersion and expresses geometric rigidity by vanishing of a differential obstruction: the conformal Lie derivative condition for $H$1, and the vanishing tractor second fundamental form for conformal submanifolds.

4. Implicit, current-valued, and flow-generated representations

A more literal realization of “submanifold fields” appears in implicit representations. For codimension $H$2, a smooth complex-valued field $H$3 with nonempty regular zero set,

$H$4

represents a smooth codimension-$H$5 submanifold as $H$6 (Chern et al., 15 Jul 2025). The orientation is recovered from phase winding through

$H$7

and away from the zero set the phase

$H$8

defines an $H$9-family of codimension-$$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$0 hypersurfaces $$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$1 sharing the codimension-$$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$2 zero set as common boundary. The associated circle differential current satisfies

$$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$3

and the Liouville $$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$4-form $$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$5 on the implicit field space obeys

$$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$6

where $$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$7 is the Marsden–Weinstein symplectic form on the shape space of codimension-$$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$8 submanifolds. After quotienting by vertical motions in $$\tilde{\nabla}_X Y=\nabla_XY+h(X,Y),\qquad \tilde{\nabla}_X \xi=-A_\xi X + D_X\xi,$$9, one obtains a principal prequantum bundle with structure group

$x\mapsto E_x\subset T_x\tilde M$0

A different current-theoretic representation associates to every smooth embedded submanifold $x\mapsto E_x\subset T_x\tilde M$1 a positive symmetric supercurrent

$x\mapsto E_x\subset T_x\tilde M$2

where $x\mapsto E_x\subset T_x\tilde M$3 is formed from orthonormal conormals (Berndtsson, 2018). This supercurrent encodes tangent and normal geometry, and

$x\mapsto E_x\subset T_x\tilde M$4

recovers the induced Riemannian volume measure. Its differential is governed by the second fundamental form through

$x\mapsto E_x\subset T_x\tilde M$5

and minimality becomes the linear current condition

$x\mapsto E_x\subset T_x\tilde M$6

The paper extends this to positive symmetric supercurrents $x\mapsto E_x\subset T_x\tilde M$7 of bidimension $x\mapsto E_x\subset T_x\tilde M$8, declaring such a $x\mapsto E_x\subset T_x\tilde M$9 minimal when $N^n\subset M^m$0.

A third representation replaces the immersion map by an even Clifford-valued spin field $N^n\subset M^m$1 satisfying $N^n\subset M^m$2 and

$N^n\subset M^m$3

If $N^n\subset M^m$4 satisfies the Killing spin field equation

$N^n\subset M^m$5

the deformed frame obeys the Cartan moving frame equations, and the integrability condition reproduces the Gauss, Codazzi, and Ricci equations (Zou, 2024). In this formalism, the submanifold is encoded by a continuous field of frame rotations whose differential constraint forces local immersibility.

A final representation is flow-generated. Given smooth vector fields $N^n\subset M^m$6, a base point $N^n\subset M^m$7, and encoder functions $N^n\subset M^m$8, the reconstruction map

$N^n\subset M^m$9

has image contained in the orbit of the generated flows (Hanson et al., 2022). By Sussmann’s orbit theorem, this orbit is a connected immersed submanifold. Here the “field” is the finite family of ambient vector fields, while latent coordinates are interpreted as flow times.

Taken together, these constructions show that a submanifold can be encoded by a complex scalar field, a supercurrent, a Clifford-valued frame field, or a finite system of vector fields with flow parameters. The resulting notions are not equivalent, but each gives a rigorous field-based mechanism for representing submanifold geometry.

5. Deformation, admissibility, and neighborhood models

The deformation-theoretic side of the subject asks which infinitesimal fields along a submanifold integrate to actual nearby submanifolds. In graded manifolds, fixing both the dimension and the degree of an immersed submanifold imposes a first-order PDE on variational vector fields (Citti et al., 2019). If $\nu N$0 has fixed degree $\nu N$1, admissible variations are those preserving degree, and the corresponding infinitesimal admissibility condition is

$\nu N$2

for every simple $\nu N$3-vector $\nu N$4 with $\nu N$5. In local matrix form this becomes

$\nu N$6

The paper proves that the normal component $\nu N$7 carries the essential information, introduces the rank condition of strong regularity,

$\nu N$8

and shows that under strong regularity every compactly supported admissible vector field is locally integrable to an actual degree-preserving variation. It also exhibits isolated surfaces in the Engel group for which no nontrivial admissible normal deformation exists. When admissible variations do exist, the associated fixed-degree Euler–Lagrange operator can be of third order.

A complementary neighborhood theory is furnished by Euler-like vector fields and deformation to the normal cone (Sadegh et al., 2016). For an embedding $\nu N$9, a vector field $C^\infty$00 on $C^\infty$01 is Euler-like if for every smooth function $C^\infty$02 vanishing on $C^\infty$03 to order $C^\infty$04,

$C^\infty$05

where $C^\infty$06 vanishes on $C^\infty$07 to order $C^\infty$08 or higher. The paper explains, via the deformation space interpolating between $C^\infty$09 and the normal bundle $C^\infty$10, that there is a bijection between germs of tubular neighborhood embeddings and germs of Euler-like vector fields. In the classical case the zero fiber of the deformation space is $C^\infty$11; in the filtered case it becomes a bundle of nilpotent homogeneous spaces $C^\infty$12 built from osculating groups. The deformation space is therefore a smooth family connecting the ambient geometry to its infinitesimal normal model.

These two theories address different problems—admissible variation versus neighborhood equivalence—but they converge on the same point: not every field along a submanifold is geometrically realizable. Realizability is controlled by explicit differential equations, rank conditions, or grading constraints, and when these are satisfied the resulting family of nearby submanifolds is smooth in a precise sense.

6. Constraint submanifolds in variational field spaces

A broader usage suggested by recent work arises when physically admissible continuous field ansätze form a regular submanifold of an ambient parameter space rather than a geometric submanifold of an ambient manifold (Tiwana et al., 25 Nov 2025). For two-field relativistic continuous matrix product states, generic parameters $C^\infty$13 produce an infinite kinetic-energy density, and finite energy requires the regularity condition

$C^\infty$14

The admissible set

$C^\infty$15

is treated as a Riemannian submanifold of the full ansatz manifold. Its tangent space is the kernel of the linearized commutator map,

$C^\infty$16

and optimization is carried out intrinsically by projecting the gradient onto this tangent space and retracting back to the constraint surface. The projection is obtained from a linear matrix equation $C^\infty$17, the projection step is solved iteratively, and the descent is implemented with a constraint-preserving retraction and a Riemannian gradient method.

This is not submanifold theory in the classical embedded-manifold sense. It is, however, a precise example of a continuous field ansatz whose physically meaningful configurations lie on a regular submanifold defined by exact algebraic constraints. A plausible implication is that “continuous submanifold fields” can also designate constrained families of continuous fields whose admissibility is encoded geometrically, even when the ambient space is a variational manifold rather than a geometric manifold.

Across the cited literature, the phrase therefore names not a single established theory but a cluster of rigorous constructions. In each of them, a submanifold is controlled by fields—tensorial, vectorial, tractor-valued, complex-valued, current-valued, Clifford-valued, or variational—that encode compatibility with tangent and normal structure, determine admissible deformations, or reconstruct the submanifold from differential data.