catalyst-q 0.1.3

Closed-source freemium Python SDK for the Catalyst-Q API

Catalyst-Q SDK

Closed-source freemium Python SDK for building against the Catalyst-Q API. The public package is catalyst-q; the import package is catalyst_q. RainProtocol portable state keys are included for stateless workflow continuity.

Docs, benchmarks, pricing, and how-to guides: https://catalyst-q-sdk.strategic-innovations.ai/docs

pip install catalyst-q

# Controlled hosted index:
pip install --index-url https://catalyst-q-sdk.strategic-innovations.ai/simple catalyst-q

from catalyst_q import CatalystQClient, MaxCutProblem, QuantumCircuit, RainProtocolKey, RainQRAM, SATProblem

client = CatalystQClient()  # no account or API key required for free-tier starts
rain_key = RainProtocolKey.create(workflow_id="demo")

circuit = QuantumCircuit(2).h(0).cx(0, 1).measure(0, 0).measure(1, 1)
request = client.prepare_execute(circuit, rain_key=rain_key, shots=1024, calls_this_month=0)

sat_request = client.prepare_sat(
    SATProblem(clauses=[[1, -2, 3], [-1, 2]], variables=3),
    rain_key=rain_key,
    solver_runs_this_month=0,
)

maxcut_request = client.prepare_maxcut(
    MaxCutProblem(edges=[(0, 1, 1.0), (1, 2, 2.0), (0, 2, 0.5)], nodes=3),
    rain_key=rain_key,
    solver_runs_this_month=1,
)

qram = RainQRAM.create(workflow_id="solver-memory").store("candidate:greedy", "101010")
maxcut_with_qram = client.prepare_maxcut(
    MaxCutProblem(edges=[(0, 1, 1.0), (1, 2, 2.0), (0, 2, 0.5)], nodes=3),
    rain_key=rain_key,
    rain_qram=qram,
    solver_runs_this_month=2,
)

On first SDK request the client creates a local anonymous install ID and sends it as X-Catalyst-Install-ID. The API can register that install as a free user and log/rate-limit requests without requiring account signup during pip install.

The free policy is intentionally generous for integration, testing, and local workflow design. Production workflows require a paid license token and server-side permission from Catalyst-Q.

Default policy tiers:

• Free developer: 10,000 circuit runs/month, 500 solver runs/month, 50-qubit circuits, no production use.
• Starter production: $29/month, 100,000 circuit runs/month, 5,000 solver runs/month, 100-qubit circuits.
• Team Pro: $199/month, 1,000,000 circuit runs/month, 50,000 solver runs/month, 250-qubit circuits.

Included NP solver helpers: SAT, TSP, single-constraint and multidimensional knapsack, portfolio optimization, QUBO, and Max-Cut. RainQRAM can be attached to solver requests as a RainProtocol preconditioning payload for candidate resonance, nonlinear activation, sharding, and exact payload-state recombination.

Public benchmark harness:

catalyst-q-benchmark --output-dir catalyst-q-public-benchmarks
catalyst-q-benchmark --execute-api --output-dir catalyst-q-live-api-benchmarks
catalyst-q-benchmark --execute-api \
  --base-url https://api.strategic-innovations.ai/v3turbo \
  --output-dir catalyst-q-demo-api-benchmarks

The harness generates JSON and Markdown artifacts covering QED-C, SupermarQ, QASMBench, MQT Bench, SATLIB, TSPLIB, OR-Library-style, and Biq Mac-style SDK workloads. It validates public request generation and billing estimates. With --execute-api, it also captures status_code, latency_ms, response_bytes, and response_sha256 for each live API request. The SDK docs domain exposes a public demo execution API for zero-credential smoke tests. It does not disclose private execution internals.

Proof harness:

catalyst-q-prove --output-dir catalyst-q-proof-results

The proof harness emits deterministic JSON and Markdown artifacts for fixed TSP and VQE cases. It compares Catalyst-Q candidate rows against public baselines, records reproducibility hashes, and keeps claims benchmark-limited rather than making broad universal performance claims.