Generalized Regular Grids

• Generalized Regular Grids are discrete structures that extend classical lattices by incorporating relaxed constraints and variable cell designs.
• They integrate algebraic, combinatorial, and topological methods, including k-point algorithms, quadrangulations, and aperiodic tilings.
• Their applications span DFT, digital topology, and quasicrystal modeling, while presenting challenges in symmetry reduction and tractability.

A generalized regular grid is a discrete structure extending the classical notion of a regular lattice, either by relaxing algebraic constraints on grid generation, generalizing embedding surfaces, accommodating variable cell or block structures, or implementing advanced symmetry reductions. This concept encompasses a diverse array of mathematical, physical, and computational constructs, ranging from crystal sampling grids in reciprocal space to quadrangulations of arbitrary surfaces, block-regular spatial discretizations, and high-girth regular subgraphs of lattices. The unifying theme is a departure from strict Cartesian regularity while preserving key structural or symmetry properties, enabling enhanced efficiency, flexibility, and mathematical richness across multiple research domains.

1. Foundational Formulations of Generalized Regular Grids

Traditional regular grids, defined as $\mathbb{Z}^n$ subsets with uniform integer spacing, form the basis for generalization along several dimensions:

• Algebraic/Combinatorial Generalization: In reciprocal spaces, generalized regular k-point grids allow any integer, invertible generator matrix $H\in\mathbb{Z}^{3\times 3}$, not just diagonal—yielding sampling grids commensurate with the lattice and enabling efficient irreducible-point reduction in Brillouin zone integration (Hart et al., 2018, Morgan et al., 2018, Morgan et al., 2019, Wang et al., 2019).
• Topological Generalization: On closed surfaces $S$, a graph $G$ embedded in $S$ is called an $S$-grid if all facial walks have length four; such quadrangulations require non-degree-4 "curvature vertices" on non-toroidal surfaces, and have rich structural characterizations (Abrams et al., 2019).
• Block/Hierarchical Generalization: In block-regular grids, space is partitioned into blocks with identical local point arrangements (but not globally uniform), supporting efficient high-dimensional sampling and simulation (Park et al., 2014).
• Girth/Graph-Theoretic Generalization: Regular subgraphs of $\mathbb{Z}^n$ can be characterized by vertex degree and maximal girth, producing high-girth, sparsely-connected substructures not seen in canonical lattice graphs (Haugland, 2021).
• Aperiodic/Projection Generalization: Generalized grid-projection techniques generate quasiperiodic tilings or grids by projecting from a higher-dimensional periodic lattice under irrational embeddings, relevant for quasicrystal modeling (Korepin et al., 2011).
• Coding and Digital Topology: Generic representations for all cell types in finite $n$-dimensional regular grids, supporting digital topology operations, further broaden “grid” to encompass unified algebraic and topological primitives (0906.2767).

This structural flexibility is critical for both theoretical classification and applied computation.

2. Algebraic and Algorithmic Frameworks for Grid Construction

The algebraic construction of generalized regular grids is predominantly governed by integer lattice theory, group actions, and normal forms:

• Generalized k-Point Grid Generation: Given primitive reciprocal vectors $R = (\mathbf b_1\;\mathbf b_2\;\mathbf b_3)$ and invertible $H\in\mathbb{Z}^{3\times 3}$, the grid is $K = R H^{-1}$, with k-points $$\mathbf{k} = K\mathbf{n},\; \mathbf{n}\in \mathbb{Z}^3$$ (Hart et al., 2018, Morgan et al., 2018). Hermite Normal Form (HNF) streamlines classification, as grids equivalent under $\mathrm{GL}(3,\mathbb{Z})$ changes and the point group $G$ are identified.
• Symmetry Reduction: The irreducible set of k-points is obtained by partitioning the grid under the action of $G$. Linear-time algorithms exploit SNF odometer indices and perfect hash tables for high efficiency, in contrast to quadratic brute-force methods (Hart et al., 2018, Morgan et al., 2019, Wang et al., 2019).
• Block-Regular Discretization: The block-circulant embedding defines covariance structures for Gaussian fields over block-regular grids, using block-FFT eigen-decomposition and block-wise parallelism (Park et al., 2014).
• Surface Quadrangulation: The skeleton-immersion characterization partitions all $S$-grids for fixed curvature sequence $L$ according to quadrangular transverse immersions of $L$-degree graphs. This covers both enumeration and generation via immersion, subdivision, and rectangular patching (Abrams et al., 2019).
• Aperiodic Tiling Construction: Generalized grid-projection (strip method) selects lattice points by projecting the fundamental domain onto subspaces and acceptance windows, ensuring non-periodicity and repetitivity (Korepin et al., 2011).

Tables: Key Integer-Matrix Frameworks in Generalized Regular Grids

Domain	Generator Matrix Form	Symmetry Operation
k-point grids	$H$: $\mathbb{Z}^{3\times 3}$ HNF	Crystal point group $G$
Surface S-grid	Skeleton graph on $L$-degree sequence	Surface homeomorphisms
Block-regular	Block structure on $\mathbb{Z}^d$	Block translation

Algebraic normal forms and symmetry-adapted enumeration are foundational to practical algorithms.

3. Structural, Analytical, and Complexity Properties

• Combinatorics and Geometry: For $S$-grids, the crucial identity is $\sum_i (4-i)v_i=4\chi(S)$, tightly linking vertex degrees, Euler characteristic, and necessary grid "curvature." The unique skeleton-immersion mapping partitions equivalence classes (Abrams et al., 2019).
• Grid Discrepancy and Tractability: For multidimensional regular grids with variable mesh size, the weighted star discrepancy $D^*_{N,\gamma}$ is characterized exactly for arbitrary product weights and grid sizes. Tractability is controlled by the decay of weight sequence $\gamma_j$, with strong, polynomial, and weak tractability criteria rigorously established (Pillichshammer, 2017).
• Girth in Subgraph Grids: The maximum girth $g(n,k)$ of $k$-regular subgraphs in $\mathbb{Z}^n$ is bounded above linearly in $n$, with explicit constructions achieving girth 12 or higher at prescribed dimension and degree. Subgraphs in classical lattices (BCC, FCC, $D_4$) exhibit distinct extremal properties (Haugland, 2021).
• Aperiodic Order: Generalized projection grids are provably non-periodic, repetitive, and possess pure-point diffraction spectra, confirming suitability for quasicrystal structure modeling (Korepin et al., 2011).

These properties directly impact sampling accuracy, numerical integration, spectral analysis, and discretized PDE simulation.

4. Implementation and Practical Applications

• Brillouin Zone Integration in DFT: Generalized regular k-point grids, including Generalized Monkhorst–Pack (GR) and Moreno–Soler grids, enable finer sampling, improved symmetry reduction, and typically 20–60% decrease in computational cost at fixed target accuracies (e.g., 1 meV/atom), compared to standard Monkhorst–Pack grids (Morgan et al., 2018, Hart et al., 2018, Morgan et al., 2019, Wang et al., 2019). Open-source libraries such as GRkgridgen and kpLib implement these algorithms with sub-second average grid generation.
• Sampling on Block-Regular/Irregular Domains: Block-circulant embedding methods (BCEM) efficiently simulate stationary Gaussian fields on block-regular grids found in MLMC and FEM settings, outperforming classical CEM by factors of 3–5 due to reduced node counts and two-level parallelism (Park et al., 2014).
• Digital Topology and Image Analysis: Bit-coded representations support dimension-independent algorithms for morphological analysis, hypersurface extraction, and homology computation, achieving memory and per-operation efficiency near theoretical optimality (0906.2767).
• Aperiodic and Quasiperiodic Materials: Projection methods are directly applied to the mathematical modeling and analysis of quasicrystal symmetries, fluorescence spectra, and statistical mechanics of interacting spins and tiles (Korepin et al., 2011).
• Girth Optimization in Graph-Theoretic Design: High-girth regular subgraph constructions are fundamental in error-correcting code design, combinatorial optimization, and extremal graph theory (Haugland, 2021).

Generalized regular grids thus underpin efficiency and accuracy in numerous computational and analytical settings.

5. Connections to Medial, Radial, and Overlay Constructions

• Medial and Radial Graphs: Classical constructs such as radial graphs $R(H,H^*)$ (bipartite, degree-4) and overlay graphs $O(H,H^*)$ (bipartite, with degree-4 and curvature vertices) are subsumed within the $S$-grid framework, providing a unified analytic and constructive theory for self-dual and current–voltage coverings on surfaces (Abrams et al., 2019).
• Surface and Lattice Generalization: Quadrangulations, medial graphs, and the diverse topological types of embedded grids are instances of structural generalization, with applications ranging from map coloring to topological quantum field theory.
• Digital Complexes: The cell coding technique in regular grids admits seamless translation between discrete homological algebra and generic computational geometry (0906.2767).

This structural unification enables the cross-pollination of methods between combinatorial, algebraic, and topological research communities.

6. Limitations, Open Problems, and Future Research Directions

• Limits of Classical Tractability: Arbitrary product weights in star discrepancy preclude strong or polynomial tractability unless the tail vanishes; only weight sequences rapidly decaying to zero allow such properties, restricting practical high-dimensional QMC grid constructions (Pillichshammer, 2017).
• Generalization to Non-Euclidean Lattices and Surfaces: For $S$-grids, enumeration and classification depend on existence of suitable quadrangular immersions for a given degree sequence and Euler characteristic, leaving open the systematic construction for complex topologies (Abrams et al., 2019).
• Girth Extremality and Dimension: The maximum girth for $k$-regular induced subgraphs as $k<n$ remains unresolved for many parameter regimes, and the full spectrum of construction methods for high-girth, low-degree, low-dimension lattices is the focus of ongoing research (Haugland, 2021).
• Optimal Grid Construction Algorithms: Despite dramatic advances, symmetry-preserving enumeration becomes expensive in very low-symmetry settings (triclinic, monoclinic cells); further improvements in hashing, reduction, and parallel workflow integration remain active areas (Wang et al., 2019).
• Quasicrystal Extensions: Generalized projection techniques can be extended via alternative cut spaces, linear transformations, or tiling subspaces, suggesting avenues for novel aperiodic orderings and corresponding physical models (Korepin et al., 2011).

Further cross-disciplinary study on the algebraic, topological, and computational aspects of generalized regular grids is likely to yield additional foundational and applied advances.