Barycentric Projection Techniques

• Barycentric projection is a method that maps barycentric coordinates from a simplex onto points in complex geometric spaces, generalizing affine interpolation.
• In Riemannian geometry, it computes the Karcher mean by minimizing an energy functional, ensuring smooth, nearly isometric embeddings under curvature constraints.
• For polytopes and positive geometries, canonical forms and Wachspress coordinates underpin robust finite element analysis and reliable coordinate chart construction.

A barycentric projection is a map sending a set of barycentric coordinates—typically weights or points in a standard simplex—onto a point in a geometric object, such as a polytope or a Riemannian manifold, so as to generalize affine interpolation to nonlinear and higher-dimensional settings. The classical Euclidean case uses convex combinations, but significant advances now enable barycentric projections in curved and combinatorially complex geometries, including Riemannian manifolds and arbitrary "positive geometries." Foundational constructions rely on Karcher means for manifolds, adjoint hypersurfaces for polytopes, and canonical forms in projective and algebraic contexts. Barycentric projection thus appears as a core mathematical tool for coordinate charts, interpolation, and finite element analysis in both theoretical and applied geometry.

1. Barycentric Projection in Riemannian Geometry

The barycentric projection on a Riemannian manifold $(M, g)$ generalizes the Euclidean convex combination by defining the barycentric image of a point $w = (w_i) \in \Delta^n$, where $\Delta^n$ is the standard $n$-simplex, as the minimizer $x^* \in M$ of the energy functional

$E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$

where $p_0, \dots, p_n$ are points in $M$ lying within a strongly convex geodesic ball of sufficiently small radius $h$ (specifically $h < \frac{1}{2}\mathrm{inj\,rad}(M)$) (Deylen et al., 2016). This minimizer, called the Karcher mean or Riemannian barycenter, depends smoothly on the weights when the points are close and in general position. The barycentric projection map $w = (w_i) \in \Delta^n$0 defined by $w = (w_i) \in \Delta^n$1 is smooth, injective, and provides a coordinate chart (“Karcher simplex”) under these conditions.

Optimality at the minimizer is characterized by

$w = (w_i) \in \Delta^n$2

or, using $w = (w_i) \in \Delta^n$3,

$w = (w_i) \in \Delta^n$4

which is solved by iterative methods such as damped Newton steps. The numerical solution converges rapidly due to strong convexity for small $w = (w_i) \in \Delta^n$5.

2. Algebraic and Polyhedral Barycentric Projection

In the context of convex polytopes and more general positive geometries, barycentric projections rely fundamentally on canonical rational forms and adjoint hypersurfaces. Given a $w = (w_i) \in \Delta^n$6-dimensional polytope $w = (w_i) \in \Delta^n$7 defined by a simple hyperplane arrangement $w = (w_i) \in \Delta^n$8, the Wachspress coordinates furnish a canonical set of barycentric coordinates: $w = (w_i) \in \Delta^n$9 where $\Delta^n$0 is the product of linear forms defining faces not containing vertex $\Delta^n$1, and $\Delta^n$2 is the “adjoint” polynomial of minimal degree vanishing on all linear spans of non-faces of $\Delta^n$3 (Kohn et al., 2019). The collective map $\Delta^n$4 is called the Wachspress map, which is birational onto its image (the Wachspress variety $\Delta^n$5), and admits a canonical inverse—the barycentric projection—given by a linear projection from the span of the adjoint in $\Delta^n$6.

Generalized barycentric projections can be defined for positive geometries $\Delta^n$7, with barycentric weight for evaluation point $\Delta^n$8 at vertex $\Delta^n$9 given by

$n$0

where $n$1 and $n$2 are canonical rational functions with simple poles along the facets of $n$3 and $n$4 (Vaitkus, 2021). For polytopes, this recovers Wachspress coordinates; more generally, the same framework underpins “area coordinates,” mean value coordinates, and those for complex positive geometries such as Grassmannians.

3. Differential and Distortion Estimates

For Riemannian barycentric projections, distortion of the pullback metric is quantified explicitly. For a $n$5-full Karcher simplex, the mass-matrix $n$6—a positive-definite operator closely approximating the identity—governs the linearization: $n$7 with $n$8 and

$n$9

Metric distortion satisfies

$x^* \in M$0

and second derivative bounds are linear in $x^* \in M$1: $x^* \in M$2 These estimates guarantee that when $x^* \in M$3 is small, the barycentric projection provides a nearly isometric embedding of the Euclidean simplex into the Riemannian manifold with distortion vanishing quadratically in $x^* \in M$4 (Deylen et al., 2016).

4. Implementation and Computational Methods

Barycentric projection on manifolds is implemented as convex optimization: given $x^* \in M$5, solve $x^* \in M$6 for $x^* \in M$7 in $x^* \in M$8 using Newton's method: $x^* \in M$9 with convergence ensured by Hessian lower bounds and small initial geodesic spread. Conversely, inversion—computing the weights $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$0 from point $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$1—reduces to a dual quadratic optimization subject to simplex constraints.

For Wachspress and generalized coordinates, practical algorithms require only evaluation of linear and adjoint polynomials. For a polytope:

• Precompute the adjoint polynomial and facet-vertex incidences.
• For each query, evaluate canonical forms $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$2, $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$3, and return weights via normalization. The cost scales with the number of facets and can be optimized via determinant/minor computation (Vaitkus, 2021).

Numerical stability is achieved by expressing barycentric weights as normalized ratios to suppress potential overflows near degenerate facets.

5. Applications and Theoretical Significance

Barycentric projections underpin geometric finite element methods on manifolds, enabling error-controlled interpolation and Galerkin discretizations whose convergence is rigorously tied to distortion estimates. In computational geometry and computer graphics, generalized barycentric projections provide coordinate systems for mesh parameterization, interpolation over convex polytopes, and robust extension to higher- and exotic-geometry settings, such as toric or Grassmannian polytopes (Deylen et al., 2016, Vaitkus, 2021).

In algebraic geometry, the layer of structure provided by canonical forms, adjoint hypersurfaces, and linear projections elucidates the relationship between algebraic invariants (e.g., Segre classes, moments of polytope measures) and coordinate systems (Kohn et al., 2019). The birationality and invertibility properties of the Wachspress map and barycentric projection are tightly coupled to combinatorial and topological features of the ambient polytope or positive geometry.

6. Domain of Validity and Generalizations

In Riemannian settings, the validity of barycentric projection as a local diffeomorphism is determined by convexity and injectivity radius, simplex thickness, and curvature bounds; explicit radii are provided: $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$4 for weights with $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$5 and support in a ball of radius $E(x; w) = \sum_{i=0}^n w_i\, d^2(x, p_i)$6 (Dyer et al., 2016). The projection remains a valid coordinate chart on neighborhoods controlled by this data, with the Jacobian invertible if and only if the tangent simplex Gram matrix is nondegenerate.

Generalized barycentric projection extends beyond polytopes to any positive geometry with a recursively well-defined canonical form and simple boundary poles, automatically including convex polytopes and admitted by more exotic structures with rational canonical forms (Vaitkus, 2021). In low regularity or high curvature, a plausible implication is that the size of the domain admitting an injective barycentric projection shrinks, and numerical algorithms require additional stabilization.

---

Fundamental research by von Deylen-Glickenstein-Wardetzky (Deylen et al., 2016), Vaitkus (Vaitkus, 2021), and Martinez-Bernstein-Chen-Sottile-Tang (Kohn et al., 2019) has fully characterized the mathematical, algorithmic, and distortion-theoretic properties of barycentric projection in the contemporary literature, supplying both practical criteria for implementation and rigorous guarantees for geometric and numerical applications.