Canonical Principal Parameters in Differential Geometry

• Canonical principal parameters are specialized local coordinates on surfaces that normalize invariant geometric functions and reduce complex PDE systems.
• They are employed to adapt principal curvature lines or isothermal coordinates in Euclidean ℝ⁴ and Minkowski space, ensuring local uniqueness up to finite ambiguities.
• The approach simplifies Bonnet-type theorems by condensing the local theory of surfaces into systems governed by a reduced number of invariant functions.

Canonical principal parameters are geometrically distinguished local coordinates on a surface in which the principal-line parametrization is supplemented by a normalization imposed by the surface invariants. In the recent literature, the exact phrase is used most explicitly for surfaces in $\mathbb R^4$ without minimal points and for marginally trapped surfaces in Minkowski $4$-space, where the normalization converts a larger Bonnet-type system into a reduced PDE description by four or three functions, respectively (Kassabov et al., 31 Jul 2025, Maksimović et al., 8 May 2026). Closely related theories use the shorter terms “canonical parameters,” “canonical coordinates,” or “canonical isothermal coordinates,” especially for minimal surfaces in $\mathbb R^4$, $\mathbb R^4_1$, and $\mathbb R^3_1$, where the same structural role is played by normalized coordinates adapted to curvature geometry or holomorphic data (Ganchev et al., 2016, Ganchev et al., 2016, Kassabov et al., 2023).

1. Terminology and geometric meaning

The phrase is not uniform across the literature. For general surfaces in $\mathbb R^4$ and for marginally trapped surfaces in $\mathbb R^4_1$, the term canonical principal parameters is explicit and denotes principal parameters satisfying an additional normalization condition expressed through invariantly defined one-variable functions $\varphi(u)$ and $\psi(v)$, with the canonical condition $\varphi(u)=1$ and $4$0 (Kassabov et al., 31 Jul 2025, Maksimović et al., 8 May 2026). In the minimal-surface literature in higher codimension, by contrast, the prevalent terms are canonical parameters or canonical isothermal coordinates, even when the construction is the direct analogue of the classical canonical principal parametrization from $4$1 (Ganchev et al., 2016, Ganchev et al., 2016).

The underlying idea is stable across these settings. One first chooses coordinates adapted to a distinguished geometric net—principal lines in the non-minimal codimension-two theories, or isothermal coordinates in the minimal theories—and then uses the remaining reparametrization freedom to normalize the metric coefficients or holomorphic invariants. This yields local uniqueness up to the expected finite ambiguities, and it reduces the local geometry to a smaller list of invariant functions satisfying a natural PDE system. A recurring source of confusion is that canonical principal parameters are not the same notion as a canonical principal direction: the former is a coordinate normalization, whereas the latter singles out one principal direction determined by an ambient vector field (Garnica et al., 2011, Kelleci et al., 2018, Kelleci et al., 2017).

2. General surfaces in $4$2

The most direct modern formulation appears for smooth surfaces

$4$3

that are free of minimal points, equivalently

$4$4

Here $4$5 is the mean curvature vector, and $4$6 are the invariants associated to the Weingarten-type map $4$7. A parametrization is by principal lines exactly when

$4$8

With principal parameters, one chooses the geometrically determined orthonormal frame

$4$9

and a unit normal $\mathbb R^4$0 such that $\mathbb R^4$1 is positively oriented. The corresponding invariant functions are

$\mathbb R^4$2

with

$\mathbb R^4$3

The canonical principal parameters are defined by a normalization of $\mathbb R^4$4 and $\mathbb R^4$5 extracted from $\mathbb R^4$6 and their first derivatives: principal parameters $\mathbb R^4$7 are canonical when the resulting one-variable normalization functions satisfy

$\mathbb R^4$8

The paper proves that each surface free of minimal points locally admits canonical principal parameters, and that any two such coordinate systems differ only by sign, translation, and interchange of the principal directions (Kassabov et al., 31 Jul 2025).

Their main analytical role is reductive. Without canonization, the local theory uses the eight functions

$\mathbb R^4$9

subject to the Gauss–Codazzi–Ricci system. In canonical principal parameters, the surface is determined up to a motion by the four functions

$\mathbb R^4_1$0

together with a reduced PDE system. The reconstruction proceeds through a hyperbolic first-order system for $\mathbb R^4_1$1 and $\mathbb R^4_1$2, after which $\mathbb R^4_1$3 are recovered from explicit formulas in terms of $\mathbb R^4_1$4 and the reconstructed metric coefficients. In this sense, canonical principal parameters are the coordinate framework in which the full local theory of non-minimal surfaces in $\mathbb R^4_1$5 becomes a four-function Bonnet theorem (Kassabov et al., 31 Jul 2025).

3. Marginally trapped surfaces in Minkowski $\mathbb R^4_1$6-space

For spacelike surfaces in $\mathbb R^4_1$7, a surface is marginally trapped when

$\mathbb R^4_1$8

so the mean curvature vector is everywhere lightlike. In the framework developed for such surfaces, one uses principal parameters and a geometrically adapted frame

$\mathbb R^4_1$9

where $\mathbb R^3_1$0, $\mathbb R^3_1$1, $\mathbb R^3_1$2, and $\mathbb R^3_1$3 is a null normal satisfying

$\mathbb R^3_1$4

The seven geometric functions are

$\mathbb R^3_1$5

with

$\mathbb R^3_1$6

The theory isolates the general type condition

$\mathbb R^3_1$7

equivalently

$\mathbb R^3_1$8

Within this class, canonical principal parameters are again defined by the normalization

$\mathbb R^3_1$9

where $\mathbb R^4$0 arise from a first-order system for $\mathbb R^4$1 and $\mathbb R^4$2 involving four functions $\mathbb R^4$3 determined by $\mathbb R^4$4 and their derivatives (Maksimović et al., 8 May 2026).

The gain is sharper than in the Euclidean codimension-two case. The earlier Bonnet-type description requires seven functions satisfying a differential system, whereas canonical principal parameters reduce the local data to three smooth functions

$\mathbb R^4$5

with $\mathbb R^4$6, satisfying a natural PDE system. The paper proves local existence of canonical principal parameters for every marginally trapped surface of general type and local uniqueness up to sign, translation, and interchange of the coordinates. It also proves that every such surface is determined up to a motion in $\mathbb R^4$7 by the triple $\mathbb R^4$8 together with the canonical reconstruction system. Canonical principal parameters therefore play exactly the same structural role as in the Euclidean theory: they absorb coordinate freedom along principal lines and isolate the essential invariants of the geometry (Maksimović et al., 8 May 2026).

4. Minimal-surface analogues and the principal/isothermal divide

Minimal-surface theory in codimension two shows that the phrase canonical principal parameters is not universal even when the geometry is parallel. For minimal surfaces of general type in Euclidean $\mathbb R^4$9-space, the standard term is canonical parameters, not canonical principal parameters. The coordinates are isothermal,

$\mathbb R^4_1$0

and are adapted to the principal axes of the curvature ellipse rather than to principal curvature directions in the $\mathbb R^4_1$1 sense. Their defining normalization is

$\mathbb R^4_1$2

equivalently

$\mathbb R^4_1$3

The same paper explicitly recalls that in $\mathbb R^4_1$4 the classical canonical parameters for minimal surfaces are “principal and isothermal,” with

$\mathbb R^4_1$5

and natural equation

$\mathbb R^4_1$6

The higher-codimension Euclidean theory is therefore presented as the direct analogue of canonical principal parametrization, but with the curvature ellipse replacing the single shape operator (Ganchev et al., 2016).

For minimal space-like surfaces in $\mathbb R^4_1$7, the literature again uses canonical coordinates or canonical isothermal coordinates, not the exact phrase “canonical principal parameters.” The normalization is

$\mathbb R^4_1$8

and the coordinates are divided into first and second type. The paper emphasizes that these coordinates are analogous to canonical principal parameters in other minimal-surface settings because they are unique up to finite ambiguity, normalize the fundamental invariants, reduce the Weierstrass data to a canonical two-function form, and lead to curvature formulas and Bonnet-type classification data (Ganchev et al., 2016).

A different Lorentzian pattern appears for minimal timelike surfaces in $\mathbb R^4_1$9. There the paper uses the term canonical parameters, but explicitly states that when the Gauss curvature is negative, these parameters are principal. In the $\varphi(u)$0 branch one has

$\varphi(u)$1

so the coordinate lines are principal. In the $\varphi(u)$2 branch, however,

$\varphi(u)$3

and the coordinates are asymptotic rather than principal. This makes the dependence on signature and curvature sign explicit: canonical parameters need not be principal, but when they are, they provide the Lorentzian counterpart of canonical principal parametrization (Kassabov et al., 2023).

5. Canonical principal direction and adapted principal coordinates

A separate but neighboring literature studies canonical principal direction (CPD) rather than canonical principal parameters. For a hypersurface in a Riemannian manifold with ambient vector field $\varphi(u)$4, the condition is that the tangential projection $\varphi(u)$5 be a principal direction. For submanifolds of higher codimension in Euclidean space, the stronger definition requires the tangential component $\varphi(u)$6 of a fixed ambient direction $\varphi(u)$7 to be a principal direction of all shape operators (Garnica et al., 2011, Kelleci et al., 2018).

Although this is not a normalization theory for principal parameters, it produces principal-coordinate systems of the type often associated with canonical principal parametrization. For CPD surfaces in $\varphi(u)$8, one can choose a local orthonormal frame $\varphi(u)$9 with

$\psi(v)$0

where $\psi(v)$1 is the distinguished tangent direction. The paper proves local existence of coordinates $\psi(v)$2 such that

$\psi(v)$3

and both relevant shape operators are diagonal in the same frame (Kelleci et al., 2018). For hypersurfaces with canonical principal direction relative to a closed conformal vector field, the preferred principal direction

$\psi(v)$4

has geodesic integral curves, and local models arise as graphs of transnormal functions (Garnica et al., 2011).

The Minkowski $\psi(v)$5-space classification of CPD surfaces shows the same adapted-coordinate mechanism in Lorentzian signature. Relative to a space-like or light-like constant direction, the surface can be written in coordinates aligned with the canonical principal direction, with diagonal first fundamental form and diagonal shape operator in the diagonalizable cases. These coordinates are principal coordinates adapted to a distinguished direction, but they are not the same object as canonical principal parameters in the Bonnet-type normalization sense (Kelleci et al., 2017).

6. Extensions, misconceptions, and terminological drift

Two misconceptions recur. The first is that canonical principal parameters must always be attached to principal curvature lines. The codimension-two minimal theories show otherwise: in $\psi(v)$6 and $\psi(v)$7, the canonical coordinates are fundamentally isothermal and are organized by the curvature ellipse or by holomorphic data, even though they occupy the same structural position as canonical principal parameters (Ganchev et al., 2016, Ganchev et al., 2016). The second is that every use of “canonical” and “principal” in parametrization theory refers to curvature-line coordinates. Outside surface theory, “canonical parameterization” often means a parameter determined by intrinsic invariants of an unparameterized object.

For simple plane curves, one paper treats a parameterization as canonical when it is determined from geometric invariants such as arc-length and curvature. Its examples are the normalized arc-length parameter, the parameterization proportional to cumulative absolute curvature,

$\psi(v)$8

and the mixed curvarc-length parameter

$\psi(v)$9

together with the family

$\varphi(u)=1$0

There “principal” is not used in the differential-geometric sense of principal directions on surfaces (Tumpach, 2023). A later paper on closed plane curves makes this distinction even sharper by defining a canonical parameterization as a smooth section of the principal fiber bundle

$\varphi(u)=1$1

so that “principal” refers to a principal bundle rather than to principal curvature geometry (Ciuclea et al., 30 Sep 2025).

The modern differential-geometric use of canonical principal parameters is therefore best understood as one member of a broader family of canonical coordinate constructions. In the codimension-two surface theories where the phrase is explicit, it means principal coordinates normalized by invariants so that the local geometry is governed by a reduced PDE system and a Bonnet-type existence-and-uniqueness theorem. In related minimal and Lorentzian theories, the same role may instead be played by canonical isothermal coordinates or by canonical parameters that are principal only in specific curvature regimes.