Library / Circuits / lumen-lattice
Lumen Lattice
Lumen Lattice is the Data-T product family for representing, evolving, and reading quantum circuits. Four engines share one register model, one observable algebra, and one document set on data-t.uk. Amplitudes, tensor lattices, open-system drift, and Pauli traces stay inside Data-T contracts so Argo Mesh, risk desks, and research programmes can call them without borrowing another vendor’s manual.
Documents on this site
- Lumen State/library/lumen-state
- Lumen Weave/library/lumen-weave
- Lumen Drift/library/lumen-drift
- Lumen Trace/library/lumen-trace
- Reference contract/library/reference
What the family is for
A quantum programme at Data-T is a list of gates, a layout of qubits, a noise story when the device is open, and a set of questions: a probability, an expectation, a sample, or a reduced state on a subset of wires. Lumen Lattice answers those questions with four specialised products rather than one opaque simulator.
Lumen State keeps the full amplitude vector when the register is small enough that 2ⁿ complex numbers fit in accelerator memory. Lumen Weave discards that vector and stores the circuit as a tensor lattice, contracting only the indices a question actually needs. Lumen Drift lifts a pure state into a density operator so channels, leakage, and thermal baths have a first-class object. Lumen Trace pushes Pauli strings backward through the circuit when the question is an expectation and the light-cone stays sparse.
- One identity model: DATA-T LTD, London seat, programme pages under /products/lumen-lattice.
- One access path: API keys issued from the existing Get API access flow, then scoped to a Lumen engine.
- One numbering scheme: product 1.0 specifications in this document set. Runtime binaries are deployed by Data-T, not downloaded as a renamed third-party library.
How a question moves through the lattice
Callers submit a circuit in the Data-T gate alphabet (H, S, T, RX, RY, RZ, CX, CZ, SWAP, measure, reset) plus an optional layout map from logical wires to device wires. They attach a query: statevector snapshot, probability of a bitstring, sample of shots, reduced density on a subsystem, or the expectation of a Pauli string.
A router picks the engine. If the width is at most the State budget and the query needs amplitudes, Lumen State applies gates as in-place updates on a complex buffer. If the width is larger but the entanglement across cuts stays modest, Lumen Weave builds a contraction plan and returns only the scalar or the small tensor the query asked for. If a channel is present, Lumen Drift is mandatory. If the query is a Pauli expectation and the backward cone is thin, Lumen Trace skips the state entirely.
What this document set is
These pages are the product specification and the conceptual manual for Lumen Lattice. They define objects, cost models, failure modes, and the URL of every Data-T library contract. They do not copy another company’s CUDA manual, header files, or example suite. Where an accelerator is used in production, Data-T names the deployment in the runtime note returned by the API, not by pasting an upstream install guide.
Each concept below has its own URL so a desk can link a model card, a risk memo, or an investor note directly to the object it depends on.