Multiparadigm Engine
Documentation status: reference — see Maturity and evidence.
The multiparadigm engine provides the execution mechanisms used by logiCells models. It allows several evaluation styles to cooperate without moving business meaning out of the conceptual layer.
Execution families
- functional: computation, transformation, and function composition;
- logical: queries, constraints, derivation, and reasoning;
- parallel / process: tasks, processes, channels, and orchestration;
- neural: tensors, embeddings, and models supported by the target version.
These families may cooperate in one application, but their exact capabilities must be read according to documentation status and Runtime version.
FPL and FPLN
FPL provides expressions and scripts used by several artifacts. FPLN extends the execution model toward richer processes and evaluation. Application code should remain above private interpreter or machine classes.
Contexts and Runtime values
An execution has a context, values, optionally processes and channels, and a shutdown policy. Runtime identities must not be confused with business identities in the conceptual model.
Local and remote
An SDK may present a coherent surface across local and remote Runtimes. This does not make operational properties identical: latency, cancellation, recovery, streaming, and lifetime remain explicit concerns.
Public boundary
The public surface consists of models, published capabilities, SDK/bindings, and Runtime contracts. Internal evaluation classes and constructors are not application APIs.
Maturity
A roadmap page or design note must not be interpreted as a shipped capability. Neural and Symbolic[Neuro] features are documented with their maturity level where known.