dydact-lab 0.3.0

Official Python SDK for DOLAS Lab — universal-substrate reasoning API

dydact-lab

Official Python SDK for DOLAS Lab — universal-substrate reasoning API.

Zero-parameter, zero-training geometric compilation across domains: symbolic math, constraint satisfaction (Sudoku, SAT), chemistry (SMILES), and circuit equivalence — all served from a single op-routed endpoint.

pip install dydact-lab

Quickstart

from dydact_lab import Lab

lab = Lab(api_key="sk-dol-...")           # or set env DOLAS_API_KEY

# Math — LaTeX, sympy, or Python syntax
r = lab.math.evaluate("sqrt(3**2 + 4**2)")
print(r.result["value"])                   # → 5.0

# Sudoku — any perfect-square N (4, 9, 16, 25)
grid = [[5,3,0,0,7,0,0,0,0], ...]
r = lab.sudoku.solve(grid)
print(r.result["solved"], r.elapsed_ms)    # → True, ~30 ms

# Chemistry — SMILES → geometric signature + zero-shot properties
r = lab.smiles.encode("CN1C=NC2=C1C(=O)N(C(=O)N2C)C")  # caffeine
print(r.result["properties"]["molecular_weight"])

# Circuit equivalence — ISCAS'85 .bench format
r = lab.circuit.equivalence(bench_a, bench_b)
print(r.result["equivalent"])

# Boolean SAT — CNF as list of [(var_idx, negated), ...]
clauses = [[(0,False),(1,True),(2,False)], [(0,True),(1,False),(2,True)]]
r = lab.sat.solve(clauses, n_vars=3)
print(r.result["sat"], r.result["assignment"])

Async

import asyncio
from dydact_lab import AsyncLab

async def main():
    async with AsyncLab(api_key="sk-dol-...") as lab:
        r = await lab.math.evaluate("2+2")
        print(r.result["value"])

asyncio.run(main())

Capability discovery

lab.ops()    # → list of all ops + schemas supported by the deployed backend

Unified compute (low-level)

Every namespace call routes through the same single endpoint:

r = lab.compute("math.evaluate", {"math": "2+2"})

This is by design — minimizes API surface, makes adding new ops a backend-only change.

Reproducibility

Every response includes a trace_id (SHA-256 of op + input + options) you can cite in publications. The compiler is deterministic given input + seed.

Tiers + auth

• Academic — free for verified .edu researchers (citation required)
• Pro Researcher — $79/mo
• Lab Enterprise — $2,500/mo
• DOLAS Strategic — contracted

Sign up: https://lab.dydact.io/signup

License

Apache-2.0.