Conformal Hessian Complex

• Conformal Hessian Complex is a differential complex in 3D that uses symmetric and traceless (conformal) tensors, replacing the full Hessian with its deviatoric form.
• It extends the classical Hessian complex by enforcing tensor constraints through BGG-based local bubble constructions and smoothness conditions for exact finite element realizations.
• The framework supports stable, structure-preserving discretizations in higher-order PDE systems, with applications in elasticity, fluid mechanics, and advanced geometric modeling.

In three dimensions, the conformal Hessian complex is a differential complex of conformal tensors, meaning tensors in $\mathbb S\cap\mathbb T$, i.e. simultaneously symmetric and traceless. It is the conformal analogue of the usual Hessian complex, and in its continuous form it is

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$

where ${\rm CH}$ is the kernel of $devhess$, the first operator is the deviatoric Hessian, and the tensor spaces carry nontrivial trace and smoothness conditions required for conformity (Huang, 2 Aug 2025). In the finite element literature, the complex is constructed by extending the Bernstein–Gelfand–Gelfand framework to conformal tensor fields and by combining discrete BGG arguments with local bubble complexes, geometric decomposition, and reduction operations. The resulting complexes are exact and are designed to support stable, structure-preserving discretizations for higher-order PDE systems (Huang, 2 Aug 2025).

1. Position within Hessian-complex theory

The conformal Hessian complex is best understood against the background of the ordinary three-dimensional Hessian complex. In the latter, the continuous sequence is

$$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$

with differential operators given by the Hessian, row-wise curl, and row-wise divergence (Arf et al., 2021). In that setting, $\mathbb S$ denotes symmetric matrix fields and $\mathbb T$ traceless matrix fields. The complex is stated to be closed and exact under the assumptions that $\Omega\subset\mathbb R^3$ is bounded, simply connected, Lipschitz, and has connected boundary (Arf et al., 2021).

The conformal Hessian complex modifies this pattern by imposing the conformal tensor constraint at the tensor levels. Instead of passing from symmetric tensors to traceless tensors through row-wise curl, it works entirely in $\mathbb S\cap\mathbb T$, and it replaces the full Hessian by its deviatoric part: $devhess\,v = dev(\nabla^2 v).$ This is why the scalar kernel is no longer ${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$0 alone. The kernel of ${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$1 is

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$2

so the conformal scalar nullspace contains affine functions together with the quadratic radial mode (Huang, 2 Aug 2025).

This contrast is structural rather than merely notational. In the ordinary Hessian complex, symmetry and tracelessness appear in different spaces. In the conformal Hessian complex, both tensor spaces consist of fields that are already symmetric and traceless. A plausible implication is that the conformal complex is tuned to models where trace-free structure is intrinsic rather than auxiliary.

2. Continuous sequence, operators, and conformity conditions

The continuous conformal Hessian complex in three dimensions is

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$3

and the paper states that the resulting conformal Hessian complex is exact (Huang, 2 Aug 2025). The operators are the Hessian

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$4

the deviatoric Hessian

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$5

the symmetric curl ${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$6, defined using the row-wise curl and symmetrization, and the double divergence

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$7

A technically important space is

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$8

Its conformity is characterized by continuity of the face traces

${\rm CH}\xrightarrow{\subset} H^2\xrightarrow{devhess} H(\symcurl; \mathbb{S}\cap\mathbb{T})\xrightarrow{\symcurl} H(divdiv; \mathbb{S}\cap\mathbb{T}) \xrightarrow{divdiv} L^2\xrightarrow{}0,$9

and the edge trace

${\rm CH}$0

A piecewise smooth tensor belongs to ${\rm CH}$1 if and only if these traces have matching jumps across interior faces and edges (Huang, 2 Aug 2025).

The finite element construction is organized by smoothness vectors

${\rm CH}$2

which encode vertex-, edge-, and face-level continuity orders. For the conformal Hessian complex, the crucial chain is

${\rm CH}$3

These smoothness relations determine how the scalar and tensor spaces are matched across simplices (Huang, 2 Aug 2025).

The same paper places the conformal Hessian complex beside the conformal elasticity complex, which involves the linearized Cotton–York operator

${\rm CH}$4

That operator is not part of the conformal Hessian complex itself, but its appearance shows that the conformal Hessian complex sits inside a broader operator algebra of conformal tensor complexes (Huang, 2 Aug 2025).

3. BGG origin and local bubble construction

The conformal Hessian complex is obtained by applying the BGG construction to a diagram built from the de Rham complex and the div-div complex, then projecting to the conformal tensor sector. In the paper’s formulation, it arises from the last two rows of a BGG diagram combining the polynomial or smooth de Rham complex, the div-div complex, and the operator algebra relating these to ${\rm CH}$5, ${\rm CH}$6, and the trace and deviatoric projections (Huang, 2 Aug 2025).

A major methodological point is that the construction is not carried out only at the global finite element level. Instead, BGG is applied to local bubble complexes. On a tetrahedron ${\rm CH}$7, scalar bubble spaces are

${\rm CH}$8

and tensor-valued bubble spaces are defined in operator-adapted forms such as

${\rm CH}$9

These spaces encode vanishing trace conditions tailored to the relevant operators (Huang, 2 Aug 2025).

The key device is the geometric decomposition of bubble spaces into contributions from tetrahedron interiors, faces, edges, and vertices. For $devhess$0, one such decomposition is

$devhess$1

This isolates the interior conformal part and the extra face-supported modes that occur when the face smoothness is low (Huang, 2 Aug 2025).

The construction also uses a reduction operation by quotienting out polynomial kernel spaces. In the conformal Hessian setting this includes quotienting by $devhess$2 for $devhess$3, as well as quotienting by $devhess$4, $devhess$5, or $devhess$6 in adjacent conformal or polynomial complexes. The stated purpose of this reduction is to remove nullspaces in operator kernels and ensure that dimension counts match exactness requirements (Huang, 2 Aug 2025).

The resulting exact bubble conformal Hessian complex is

$devhess$7

The paper proves the two onto-properties

$devhess$8

and

$devhess$9

which are the local exactness statements at the tensor and scalar ends (Huang, 2 Aug 2025).

A crucial intermediate ingredient is a family of two-dimensional face bubble complexes. In particular, on each triangular face $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$0, the paper constructs the conformal face div-div complex

$$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$1

where

$$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$2

These face complexes control the traces through which the global $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$3 conformity is enforced (Huang, 2 Aug 2025).

4. Finite element realizations and exactness with varying smoothness

The global finite element conformal Hessian complex is

$$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$4

The scalar spaces $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$5 and $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$6 are smooth scalar finite element spaces defined by vertex, edge, face, and volume degrees of freedom. The tensor spaces are built using degrees of freedom that encode operator-relevant traces: for $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$7, traces $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$8, $$P_1(\Omega)\xrightarrow{\ \iota_{P_1}\ } H^2(\Omega)\xrightarrow{\ \nabla^2\ } H(\Omega,\mathrm{curl},\mathbb{S})\xrightarrow{\ \nabla\times\ } H(\Omega,\mathrm{div},\mathbb{T})\xrightarrow{\ \nabla\cdot\ } L^2(\Omega)\to 0,$$9, face quantities involving $\mathbb S$0, $\mathbb S$1, and edge data; for $\mathbb S$2, traces $\mathbb S$3, $\mathbb S$4, together with edge and volume moments (Huang, 2 Aug 2025).

The paper emphasizes two families. The first is a minimal smoothness version with

$\mathbb S$5

yielding the concrete chain

$\mathbb S$6

The second is a more flexible less-smooth family produced by the paper’s tilde operation: $\mathbb S$7 which gives exact complexes while allowing inequalities rather than equalities among the smoothness vectors (Huang, 2 Aug 2025).

Unisolvence is proved by matching dimension counts with the dimensions of the polynomial tensor spaces. Global exactness is then established by combining local bubble exactness, onto-properties for $\mathbb S$8 and $\mathbb S$9, and a global dimension count based on the Euler formula

$\mathbb T$0

In the paper’s summary, local bubble exactness together with correct degree-of-freedom counting yields the global finite element complex (Huang, 2 Aug 2025).

5. Structure-preserving discretization and numerical role

The conformal Hessian complex belongs to the broader program of structure-preserving discretization. For the ordinary Hessian complex, the spline-based construction on affine images of the unit cube follows the FEEC philosophy: build a finite-dimensional subcomplex, construct uniformly bounded commuting projections, and obtain stable, convergent mixed methods for the associated Hodge-Laplacians (Arf et al., 2021). In that setting, the discrete sequence mirrors the continuous operator chain exactly, and the theoretical payoff is that stability and convergence come essentially from preserving the complex structure (Arf et al., 2021).

The conformal Hessian complex extends this structural logic to symmetric and traceless tensors. The paper states that exact finite element complexes give commuting projections, stable mixed formulations, correct kernel representation, avoidance of spurious modes, and robust discrete analogues of integration-by-parts identities (Huang, 2 Aug 2025). These are precisely the features usually sought in FEEC-compatible discretizations of operator complexes.

Methodologically, the conformal paper identifies a practical advantage in using local bubble BGG constructions rather than global BGG constructions. The stated consequences are simpler derivation, feasible dimension counting, explicit element families with variable smoothness, and new, less smooth elements than previously known ones, including a new less-smooth conformal elasticity family (Huang, 2 Aug 2025). This concerns the conformal elasticity complex as well as the conformal Hessian complex, but the same construction machinery underlies both.

The applications highlighted for the conformal complexes are general relativity, Cosserat elasticity, fluid mechanics, and liquid crystal or $\mathbb T$1-tensor models, with the $\mathbb T$2-tensor noted as a prototypical conformal tensor because it is symmetric and traceless (Huang, 2 Aug 2025). In the nonconformal Hessian-complex literature, related structure-preserving discretizations are applied to Hodge-Laplacians and to the linearized Einstein-Bianchi system, where FEEC-compatible spatial discretization yields convergence of the semidiscrete solution (Arf et al., 2021). This suggests a common numerical motivation: preserving operator algebra and tensor constraints at the discrete level so that stability does not have to be recovered a posteriori.

6. Terminological scope and nearby uses of “conformal Hessian”

The phrase “conformal Hessian” also appears in distinct PDE and complex-geometric settings, and these should not be conflated with the conformal Hessian complex. In fully nonlinear elliptic PDE, a conformal Hessian equation is written in terms of the conformal Hessian

$\mathbb T$3

and the equation has the form

$\mathbb T$4

In that sense, the paper on singular solutions proves that for every $\mathbb T$5 there exists a viscosity solution in $\mathbb T$6 of the form

$\mathbb T$7

with

$\mathbb T$8

showing that uniformly elliptic conformal Hessian equations can admit singular solutions with optimal interior regularity (Nadirashvili et al., 2014).

In complex geometry, conformal structure enters complex Hessian theory differently. For the Dirichlet problem

$\mathbb T$9

on compact Hermitian manifolds with boundary, the locally conformal Kähler condition means that locally there exists a smooth function $\Omega\subset\mathbb R^3$0 such that

$\Omega\subset\mathbb R^3$1

is Kähler. The stated role of this assumption is that it makes the Bedford–Taylor type inductive construction of wedge products for bounded potentials work in the bounded setting; for general Hermitian metrics, the corresponding wedge product is not automatically available in that approach (Gu et al., 2016).

There is also a different geometric use of “conformal” in Hessian geometry. A selfsimilar Hessian manifold $\Omega\subset\mathbb R^3$2 with complete homothetic vector field $\Omega\subset\mathbb R^3$3 induces a globally conformally Kähler structure on $\Omega\subset\mathbb R^3$4 after the rescaling

$\Omega\subset\mathbb R^3$5

and under the paper’s homogeneity hypotheses this structure is homogeneous (Osipov, 2020). This is again separate from the conformal Hessian complex of finite element and BGG theory.

Accordingly, “conformal Hessian complex” in the strict sense refers to the exact tensor complex

$\Omega\subset\mathbb R^3$6

together with its bubble, face, and finite element realizations (Huang, 2 Aug 2025). The nearby phrases “conformal Hessian equation,” “locally conformal Kähler complex Hessian equation,” and “conformally Kähler Hessian manifold” designate different, though related, conformal structures in nonlinear PDE and differential geometry.