🔢 mathematics

Incidence-based Combinatorial Geometry on Cell Complexes

This paper introduces an incidence-based combinatorial geometry framework for cell complexes that constructs local tangent-like fibers from intrinsic incidence structures to define finite analogues of transport, curvature, and metric properties without relying on smooth manifolds or fixed-rank bundles.

Original authors: Andrey P. Jivkov, Muhammad Azeem, Aneirin Griffiths, Pieter D. Boom

Published 2026-08-07

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Andrey P. Jivkov, Muhammad Azeem, Aneirin Griffiths, Pieter D. Boom

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Geometry of Connections: Why Your Neighborhood Matters

Imagine you are trying to describe the weather in a city. In the old-school way of doing physics, scientists treat the city like a smooth, continuous sheet of fabric. They assume that if you zoom in close enough, the air flows smoothly, and you can measure wind speed at any tiny point using calculus. This works great for things like water flowing in a river or light traveling through space. But what if the world isn't a smooth sheet? What if it's made of distinct, chunky building blocks—like a city made of individual houses, or a material made of tiny grains of sand? In these "discrete" worlds, the smooth rules of calculus break down because there is no "in-between" space to measure.

This is where a field called Combinatorial Mesh Calculus comes in. Think of it as a new language for physics that doesn't rely on smooth curves. Instead, it uses a map of connections—a "mesh"—to describe how things move and interact. In this language, the basic building blocks aren't points, but cells (like rooms, walls, or corners) and the lines connecting them. The big question scientists have been asking is: How do we describe things that have direction, like force or velocity, on this chunky, blocky map? If you have a wind blowing at one corner of a room, how do you compare it to the wind at the next corner if they are built on different "scaffolds"? This paper tackles that exact puzzle, proposing a way to build a geometry of direction and force directly from the shape of the connections themselves, without needing a smooth background to lean on.

Building a Geometry from the Ground Up

The authors of this paper, Andrey Jivkov and his team, are essentially inventing a new set of tools to do physics on a "blocky" world. They are taking an existing framework called Combinatorial Mesh Calculus (CMC), which was already good at handling simple numbers (scalars) on these blocky maps, and upgrading it to handle directions (vectors) and forces.

Here is the core idea: In the smooth world of standard physics, every point has a "tangent space"—a little flat plane that tells you which directions you can move. You can't have that on a blocky map because there are no smooth planes. So, the authors ask: Can we build a "direction space" just by looking at the connections?

The "Fiber" Neighborhood
Imagine you are standing at a specific intersection in a city (a "vertex" in their math). You look around and see all the streets leading away from you. The authors say: "Let's build a little vector space right here, made entirely of those streets." They call this a fiber.

If you are at a corner with three streets, your fiber is a 3D space.
If you are at a corner with four streets, your fiber is a 4D space.
If you are at a dead end with only one street, your fiber is just a line.

This is a huge shift. In traditional math, every point usually has the same number of dimensions (like 3D space everywhere). Here, the "dimension" of your local world changes depending on how many connections you have. It's like living in a city where your personal "universe" expands or shrinks based on how many neighbors you have.

The "Solder" and the "Transport"
Now, imagine you want to walk from your corner to your neighbor's corner. You have a vector (a direction) in your fiber, and your neighbor has a vector in theirs. But their fiber is built from their streets, which might be different from yours. How do you compare them?

The Solder Form: This is like a universal translator that says, "This street in your fiber corresponds to this specific edge in the map." It anchors the abstract directions to the actual physical connections.
The Transport Map: This is the rulebook for moving. It tells you how to take a vector from your fiber, walk it over to your neighbor's fiber, and translate it so it makes sense in their local language.

The "Torsion" and "Curvature" Defects
Once you have these tools, you can start measuring the "weirdness" of the map.

Torsion: Imagine you walk around a square block, carrying a direction with you. You translate it from corner to corner. When you get back to the start, does your direction match what you started with? If not, the "gap" is called torsion. It's like walking around a block and realizing you're facing the wrong way because the streets don't quite line up perfectly.
Curvature: This is similar but looks at how the rules of translation change as you move. If you take a path one way around a loop and get a different result than taking the path the other way, that's curvature.

What They Actually Found (and What They Didn't)
The paper doesn't claim to have solved all the mysteries of the universe. Instead, it builds a scaffold—a mathematical framework.

They built: A way to define these fibers, transport maps, and "solder" forms purely from the connections of the mesh. They showed how to calculate "torsion" and "curvature" as specific defects in these connections. They also defined new ways to multiply these directional quantities (using something called an "evaluation cup product") to get energy-like numbers.
They ruled out: They explicitly state that this is not just a smooth world that has been chopped up. They are not approximating a smooth surface; they are saying the geometry is the connections. They also rule out the idea that you can just copy-paste the old smooth math rules onto this new system; the rules are fundamentally different.
The Confidence Level: The authors are very careful. They call this an exploratory note. They have proposed and analyzed these mathematical definitions, showing that the pieces fit together logically as a framework, but they explicitly admit that many questions remain open. For instance, they have not yet proven a perfect "Cartan equation" (a famous formula in smooth geometry) that links everything together; they have the pieces, but the final puzzle is still being assembled. They also note that their new operators don't always behave like the smooth ones (specifically, they aren't always "nilpotent," meaning applying them twice doesn't always zero out), which is a major open problem they are flagging for future work.

Why It Matters
Why should a curious teenager care? Because this approach allows scientists to model things that are naturally "chunky" without forcing them into a smooth mold. Think of:

Crystals: Where atoms are arranged in a grid.
Biological tissues: Where cells are distinct units with different shapes.
Computer networks: Where data moves between specific nodes.

In these systems, the "smooth" assumption of traditional physics is often wrong. This new framework offers a way to do physics that respects the actual, discrete structure of the world. It suggests that the geometry of space isn't just a background stage; it's built directly from the relationships between the things on it.

The Open Questions
The paper ends by listing the things they still need to figure out. They need to find the "best" ways to move vectors around (admissible transport classes), figure out how to make the math work with specific physical materials (constitutive operators), and see if these discrete "curvature" and "torsion" values eventually look like the smooth versions we know when you zoom out.

In short, this paper is a blueprint. It says, "Here is how you can build a geometry of directions and forces out of pure connections. It works, it's consistent, and it opens the door to a whole new way of seeing the world." The rest of the journey—turning this blueprint into a working machine for real-world physics—is just beginning.

Technical Summary: Incidence-based Combinatorial Geometry on Cell Complexes

Problem Statement
Combinatorial Mesh Calculus (CMC) provides a framework for formulating conservation laws directly on cell complexes using scalar-valued combinatorial differential forms. However, many physical theories involving directional quantities (e.g., displacement, velocity, momentum, force) require local coefficient spaces, comparison of states at different locations, and geometric structures capable of representing compatibility, transport, curvature, and torsion. Standard smooth differential geometry relies on tangent bundles and differentiable structures, which do not transfer directly to explicit cell complexes. Furthermore, existing discrete approaches (such as Discrete Exterior Calculus) often rely on fixed-rank bundles prescribed independently of the complex's topology. The problem addressed is the development of an incidence-based combinatorial geometry that intrinsically generates local coefficient spaces, transport structures, and geometric invariants directly from the organization represented by the cell complex, without assuming an underlying smooth manifold.

Methodology
The paper constructs a geometric framework on a computational complex $K$ (a Forman subdivision of a physical complex $M$ ) by replacing scalar coefficients with locally varying vector spaces (fibres) generated by the complex's incidence structure.

Incidence-Generated Fibres: For each vertex $a \in K_0$ , a local fibre $F_a$ is defined as the vector space spanned by the oriented edge directions incident to $a$ . The dimension of $F_a$ ( $n_a$ ) is determined by the local valence of the vertex, allowing for variable-rank bundles that reflect local topology.
Fibre-Valued Incidence Cohains: Instead of assigning values to cells alone, the framework defines cochains on incidences $(a, \sigma)$ where a vertex $a$ is a face of a cell $\sigma$ . The coefficient space for such a cochain is the fibre $F_a$ (or its dual $F_a^*$ ). This anchors the coefficient to the specific vertex while the cell $\sigma$ provides the geometric domain.
Transport and Connections: Since neighbouring fibres are distinct, a directed edge-transport structure $P = \{P_{ab}\}$ is introduced, assigning linear maps between fibres of adjacent vertices. Path transport is defined via composition. Reversible discrete connections (isomorphisms with inverses) are identified as a special case.
Solder Forms and Torsion: A solder form $\theta$ (a vector-valued 1-cochain) relates the combinatorial 1-cells to the local fibre directions. The torsion $\Theta$ is defined as the transported-solder defect around a 2-cell, measuring the failure of local edge directions to close when transported to a reference fibre.
Curvature and Holonomy: Holonomy $R(a, f)$ is defined as the composition of transport maps around the boundary of a 2-cell. The finite holonomy defect $\Delta = R - \text{Id}$ and the orientation-odd curvature $\Omega = \frac{1}{2}(R - R^{-1})$ are introduced to characterize transport failure.
Metric Structures: A fibre metric $g$ is introduced as independent data, inducing inner products on fibre-valued cochains. This allows for the definition of bundle Hodge operators ( $\star^\flat, \star^\sharp$ ) that map between vector- and covector-valued cochains, and evaluation cup products that pair kinematic and force-like quantities to form scalar energetic fields.
Weighted Covariant Operators: A weighted covariant incidence operator $\delta^\nabla$ is constructed to raise the degree of cochains. It aggregates transported values from neighbouring vertices using anchor sets, paths, and normalizing weights to ensure well-defined summation in the common reference fibre.
Locality-Dependent Algebras: The paper analyzes the algebra of operators acting on scalar 0-cochains, defining a filtration based on combinatorial distance. It demonstrates that vector-field induced operators possess a quadratic Leibniz defect and that their commutators expand the locality radius, forming a filtered Lie algebra.

Key Contributions and Results

Intrinsic Geometric Scaffold: The framework establishes a combinatorial scaffold where local coefficient spaces, admissible directions, and transport structures are generated directly from the cell complex's organization, rather than being prescribed externally.
Variable-Rank Fibres: Unlike fixed-rank bundle approaches, the fibre dimension varies with vertex valence, providing a more flexible representation of local topology.
Operational Definitions of Torsion and Curvature: Torsion and curvature are defined operationally through finite transport operations (transported solder defects and holonomies) rather than as derivatives of a connection. The paper distinguishes between the finite holonomy defect and the orientation-odd curvature cochain.
Evaluation Cup Product: A generalized cup product is defined that pairs covector-valued and vector-valued cochains via the canonical evaluation of dual fibres, enabling the construction of scalar energetic fields without transport.
Bundle Hodge Operators: The paper defines vector-to-covector and covector-to-vector Hodge stars induced by the fibre metric, splitting the scalar Hodge star into distinct operational maps.
Leibniz Defect: The paper proves that the pointwise algebra of scalar 0-cochains on a finite vertex set admits no non-zero derivations. Consequently, operators induced by combinatorial vector fields exhibit a quadratic Leibniz defect, a distinct feature from smooth differential geometry.
Filtered Operator Algebras: The construction of a locality filtration on operator spaces reveals that commutators of radius-one operators generally result in radius-two operators, establishing a filtered Lie algebra structure.

Significance and Claims
The paper positions itself as an exploratory note that establishes a mathematical framework and identifies principal questions for future development. It claims to provide:

A combinatorial analogue of local coefficient spaces, transport, curvature, and torsion that does not rely on smooth manifolds or differentiable structures.
A unified setting for formulating mechanics and transport directly on evolving cellular organization.
A distinction from cellular sheaf theory: while the vertex data resembles sheaf stalks, the construction is not automatically a cellular sheaf or local system without additional compatibility conditions (e.g., constant rank, functoriality).
A modest scope regarding differential equations: The paper explicitly states that it does not yet establish a full Cartan-type structure equation or a nilpotent covariant derivative ( $\delta^\nabla \delta^\nabla \neq 0$ is not asserted). The relationship between the operational definitions (finite holonomy) and differential definitions (curvature forms) remains an open problem.

The work concludes by identifying open questions, including the identification of admissible transport classes, metric-compatible connections, constitutive operators, and the formulation of coarse-graining projections to continuous tangent bundles in the macroscopic limit. The significance lies in providing the foundational "scaffold" for these future developments in fibre-valued Combinatorial Mesh Calculus.

The Geometry of Connections: Why Your Neighborhood Matters

Building a Geometry from the Ground Up

Technical Summary: Incidence-based Combinatorial Geometry on Cell Complexes

More like this