Topological Relational Theory: A Simplicial-Complex View of Functional Dependencies, Lossless Decomposition, and Acyclicity

Abstract: We develop a topological lens on relational schema design by encoding functional dependencies (FDs) as simplices of an abstract simplicial complex. This dependency complex exposes multi-attribute interactions and enables homological invariants (Betti numbers) to diagnose cyclic dependency structure. We define Simplicial Normal Form (SNF) as homological acyclicity of the dependency complex in positive dimensions, i.e., vanishing reduced homology for all $n \ge 1$. SNF is intentionally weaker than contractibility and does not identify homology with homotopy. For decompositions, we give a topological reformulation of the classical binary lossless-join criterion: assuming dependency preservation, a decomposition is lossless exactly when the intersection attributes form a key for at least one component. Topologically, this yields a strong deformation retraction that trivializes the relevant Mayer--Vietoris boundary map. For multiway decompositions, we show how the nerve of a cover by induced subcomplexes provides a computable certificate: a 1-cycle in the nerve (detected by $H_1$) obstructs join-tree structure and aligns with cyclic join behavior in acyclic-scheme theory. Finally, we discuss an algorithmic consequence: Betti numbers of the dependency complex (or of a decomposition nerve) can be computed from boundary matrices and used as a lightweight schema diagnostic to localize "unexplained" dependency cycles, complementing standard FD-chase tests.

• The paper introduces a topological framework called dependency complex to encode functional dependencies using simplicial homology for relational schema design.
• Simplicial normal form (SNF) is defined through vanishing reduced homology for higher degree dimensions to diagnose cyclical dependencies.
• The paper provides a nerve-theoretic framework for multiway decompositions to identify structural obstructions through a homology-based test.

Overview and contributions

This paper develops a topological framework for relational schema design by encoding functional dependencies (FDs) as simplices of an abstract simplicial complex. Attributes become vertices, and each FD $X \to A$ in a canonical cover contributes the simplex $X \cup \{A\}$. The resulting "dependency complex" $K_F$ supports homological diagnostics: cycles in the complex correspond to cyclic dependency interactions, and whether those cycles are "filled" by higher-arity determinants is detected by homology. The paper's contributions are: (i) the definition of the dependency complex; (ii) a new normal form, Simplicial Normal Form (SNF), defined as vanishing reduced homology in positive degrees; (iii) a topological reformulation of the classical binary lossless-join criterion via simplicial collapses and Mayer–Vietoris structure; (iv) a nerve-theoretic obstruction to join-tree structure for multiway decompositions; and (v) an algorithmic proposal to use Betti numbers as lightweight schema diagnostics (2602.21213).

The authors are careful about scope. SNF is explicitly defined as homological acyclicity ($\widetilde H_n(K_F)=0$ for all $n\ge 1$), not contractibility, and the paper repeatedly disclaims any implication from vanishing homology to a specific homotopy type. This is a deliberate methodological choice that avoids conflating homology with homotopy.

The dependency complex and SNF

Given a canonical cover $F_c$ (no extraneous attributes, no redundant FDs), the dependency complex $K_F$ is generated by one simplex per FD with single-attribute right-hand side. The encoding is intentionally conservative: it records each FD as a face without closing under logical implication, so $K_F$ serves as a compact combinatorial summary of the canonical cover rather than a complete semantic object.

SNF then requires all positive-degree Betti numbers to vanish. The most substantive claim distinguishing SNF from naive directed-cycle checks is Example 3.4: with $U=\{A,B,C\}$ and $F_c=\{A\to B,\ B\to C,\ C\to A,\ AB\to C\}$, the directed FD graph contains the cycle $X \cup \{A\}$0, yet $X \cup \{A\}$1 contains the 2-simplex $X \cup \{A\}$2 induced by $X \cup \{A\}$3, which fills the triangle so that $X \cup \{A\}$4. SNF therefore distinguishes "unfilled" cyclic interaction—a genuine 1-dimensional hole—from cyclic implication explained by a higher-arity determinant. This is a strictly finer diagnostic than directed-cycle detection on the FD graph.

An important caveat, acknowledged implicitly through the conservative encoding: because $X \cup \{A\}$5 depends on the choice of canonical cover and does not close under implication, homological conclusions are statements about the cover representation, not invariantly about the FD theory itself. The paper does not prove cover-independence of $X \cup \{A\}$6, which leaves open whether two distinct canonical covers of the same FD set can yield different Betti numbers.

Lossless join topology

For binary decompositions, the paper restates the classical criterion—$X \cup \{A\}$7 or $X \cup \{A\}$8 in $X \cup \{A\}$9—in topological terms. Proposition 4.1 shows that if the intersection keys one side, repeated FD application induces elementary simplicial collapses of the keyed component onto the intersection subcomplex, preserving reduced homology in degrees $K_F$0. The Homological Lossless Join Theorem packages this into a Mayer–Vietoris statement: when the key condition holds, the connecting morphism $K_F$1 is trivial on the portion of $K_F$2 represented inside the keyed component.

The equivalence between losslessness and the key condition is, as the paper concedes, exactly the classical result of Aho, Beeri, and Ullman; the topological content is interpretive packaging rather than a new criterion. Its value lies in bridging to multiway covers and computational diagnostics.

Nerve obstructions for multiway decompositions

For $K_F$3, the paper defines the nerve $K_F$4 of the cover of $K_F$5 by induced subcomplexes $K_F$6: a set of indices forms a simplex iff the corresponding subcomplexes intersect. Proposition 4.5 states that if $K_F$7, the cover cannot admit a join tree, since a tree-shaped nerve has trivial $K_F$8. This certifies global cyclic cover interaction even when every pairwise intersection is nonempty—the three-component example with pairwise overlaps but empty triple overlap yields $K_F$9.

The paper is explicit that this does not replace the chase, which remains exact for losslessness; the nerve test is a fast structural warning that acyclic-scheme join simplifications cannot hold. A worked six-relation join query illustrates the practical payoff: when the nerve is tree-like, Yannakakis-style two-pass semijoin reduction along separators applies directly, whereas a nontrivial $\widetilde H_n(K_F)=0$0 signals multiple competing join paths and a larger optimization search space.

Algorithmic consequences

Betti numbers are computable from boundary matrices over a field via $\widetilde H_n(K_F)=0$1. In practice only $\widetilde H_n(K_F)=0$2 (and occasionally $\widetilde H_n(K_F)=0$3) need be computed, since these capture the first obstructions to acyclicity. Two uses are proposed: cycle localization within $\widetilde H_n(K_F)=0$4, flagging unfilled cyclic interactions as candidates for redesign or decomposition; and cover obstruction via $\widetilde H_n(K_F)=0$5, ruling out join-tree plans cheaply before more expensive chase-based analysis. No complexity analysis or empirical evaluation of these computations on realistic schemas is provided, so the claimed "lightweight" character remains an assertion rather than a measured result.

Limitations and open questions

Several limitations are conceded or evident. First, the binary lossless-join theorem adds no new decision power over the classical test; its contribution is conceptual. Second, the nerve condition is sufficient but not necessary for detecting all failures relevant to multiway losslessness, and the paper does not characterize precisely when $\widetilde H_n(K_F)=0$6 correlates with actual join behavior beyond tree-likeness. Third, the dependency complex is defined relative to a canonical cover, and cover-independence of the homology is not established. Fourth, the connection between SNF and classical normal forms (3NF, BCNF) is left unexplored: it is unknown whether SNF implies, is implied by, or is incomparable to standard normal forms in general. Finally, the framework handles FDs only; extensions to multivalued or join dependencies are not addressed.

Conclusion

The paper offers a mathematically clean translation of relational design concepts—FDs, lossless join, acyclicity—into simplicial homology and nerve theory. Its main technical results are a homological reformulation of the binary lossless-join criterion and a nerve-based certificate excluding join-tree structure, both accompanied by honest scoping remarks separating new content from classical results. The framework's practical significance hinges on open questions of cover-independence, relation to classical normal forms, and empirical cost of Betti-number computation on production schemas.