Gativus Network

GNET (Gativus NETwork) is a specification for a digital network modeled on biological neural networks: nodes (analogous to neurons) and relations (analogous to synapses). There are no clients or servers — only datagrams for one-way value transfer. Nodes are permanent, exchange signals via connectors, and resource creation and security are governed by quotas and a rights matrix.

The article describes the Gativus Notation (GNOT) — an engineering standard for the Gativus project consisting of 32 rules for naming, visualization, and documentation. Key requirements: all terms are exactly four characters; multi‑level entities use a three‑character base plus a one‑character suffix; NDDI is visualized as a polyhedron with each face containing one component type, and UNON displayed on the frame; each component type has a unique geometric shape and color; connectors are designed as pins and sockets typed by shape; relationships are shown as lines with an arrow, color‑coded by relationship type, with solid or dashed lines indicating reliable or unreliable delivery; DOM levels are represented as horizontal layers, TRL as a record tape; vectors have colors and thickness proportional to magnitude; documents use the SPC prefix and a three‑digit number. The notation is mandatory across all project documents, ensuring uniformity, strong typing, scalability, and compatibility among the GTOM, GNET, and GATE books.

NDDI is the fundamental and only entity in the GNET network. The node consists of components: D-components (genes containing object code), V-components (values), A-components (activities), and higher-level components (L, R, B, S, N, C, W). D-components are primary and define the entire node architecture through the linking process. Each NDDI has permanent persistence, is stored in non-volatile memory, and is activated in RAM only when needed. Connectors (POCN) provide connections between nodes, while node visualization is represented as a polyhedron with facets by component types. The minimally living node consists of D + V + A and is capable of forming sensory-reflexive networks.

Section 4 describes the architecture of the Asset Vector (AVEC) and the network immune system of GNET. It is built on Consistency Spaces (CNST) defined by issuers (GLAI, GATE, LOAI). Each node receives an AVEC vector as part of a G‑relation, and action validation reduces to multiplying the action vector by the AVEC matrix. The hierarchical asset registry is a tree (GLAI → LRAI → owner), and asset type names are 64‑bit identifiers that match the issuer’s UNON. D/G isomorphism separates fixed structure (D‑level) from dynamic behaviour (G‑level). Immunity is ensured by the mandatory presence of AVEC₀ and the CRL system. Synaptic consensus is reached locally via notaries, without global mining. Smart contracts are defined as MAP9 narratives bound to AVEC. Issuance management evolves from unitary (DOM0) to civic (MAP8/MAP9). The entire system relies on identification (UNON) and does not require computational consensus.

Section 5 describes the GNET security model based on AVEC₀ — a vector of authority in the CNST₀ space. Each axis of CNST₀ corresponds to a quota or right (NRGNQUOTA, SYGNQUOTA, LEGITIMACY, DLGT, etc.). Operation validation is performed by multiplying the action vector by the AVEC₀ matrix; delegation is the reduction of coordinates when passing from parent to child. The security hierarchy is GLAI → LRAI → GATE → ANOD → NDDI, where each level is simultaneously a registry level and a delegation level. Control is divided into three modes: synchronous (node creation NRGN), licensed (relation establishment SYGN), and asynchronous (mass traffic). Three policy filters are added: physical (AVEC₀), symbolic (KLOM rules), and axiological (W‑vector). The immune system uses AVEC₀ as MHC‑I, and CRL ensures revocation of compromised nodes. The GANN namespace is zoned (system, ordinary, temporary). Encryption and audit are integrated into the overall architecture.