Technical details

Persistent Homology

Persistent homology measures how connected components and loops appear and disappear as a scale parameter changes. SIMPLEX focuses this idea into an exact, deterministic, auditable H0/H1 engine for operational workloads.

Simplicial complexes

A simplicial complex is a collection of vertices, edges, triangles, and higher-dimensional simplices that is closed under taking faces. If a filled triangle is present, its edges and vertices must also be present.

σ ∈ K,   ∅ ≠ τ ⊆ σ   ⟹   τ ∈ K

Chains, boundaries, and homology

Over a field F, p-chains are formal sums of p-simplices. Boundary maps send a simplex to its faces. Cycles are chains with zero boundary; boundaries are cycles that bound a higher-dimensional chain.

Hp(K; F) = ker ∂p / im ∂p+1

H0 counts connected components. H1 counts independent loops that have not been filled in by triangles.

Filtrations and persistence

A filtration is a nested sequence of complexes indexed by scale. As scale increases, new simplices enter and topological features are born or die.

∅ = Kα0 ⊆ Kα1 ⊆ ··· ⊆ Kαm,   α0 < α1 < ··· < αm

A finite interval [b, d) records a feature that appears at birth scale b and dies at death scale d. Its lifetime is persistence.

ℓ = d − b

Vietoris–Rips filtrations

For metric data, the Vietoris–Rips complex connects points whose pairwise distances are below a scale threshold and fills cliques into higher-dimensional simplices.

VRα(X) = { σ ⊆ X : diam(σ) ≤ α }
diam(σ) = maxu,v∈σ d(u,v)

Stability and operational signals

Persistence diagrams are stable under bounded perturbations of the filtration. Operationally, this motivates long-lived features and summaries such as total persistence, maximum persistence, and baseline-normalized change scores.

dB(Dgmp(f), Dgmp(g)) ≤ ||f − g||

Why SIMPLEX is operational

SIMPLEX does not only return a barcode. It returns a structured run result: intervals, birth/death pairings, stable simplex handles, statistics, and audit metadata. That result can be replayed, compared, archived, and inspected by a technical reviewer.

SIMPLEX’s product boundary is intentionally narrow: exact H0/H1 persistent homology with deterministic ordering and audit-rich output.