| """ |
| DCA-01 MVP Runner - End-to-End without Docker |
| """ |
| import sys |
| sys.path.append("src/router") |
| sys.path.append("src/decomposer") |
| sys.path.append("src/blackboard") |
| sys.path.append("src/experts/physics") |
| sys.path.append("src/experts/code") |
| sys.path.append("src/experts/language") |
| sys.path.append("src/fusion") |
| from tmr_router import Router, tmr_vote |
| from dag_builder import build_dag |
| from blackboard import Blackboard |
| from physics_expert import infer as physics_infer |
| from code_expert import infer as code_infer |
| from language_expert import infer as lang_infer |
| from fusion import fuse |
| import json |
| def run(request: str): |
| print(f"\n=== DCA-01 MVP Run ===\nRequest: {request}\n") |
| print("[1] TMR Router Cluster") |
| ra = Router("router_a","sha256:a").infer(request) |
| rb = Router("router_b","sha256:b").infer(request) |
| rc = Router("router_c","sha256:c").infer(request) |
| vote = tmr_vote(ra, rb, rc) |
| print(f" -> {vote['flag']}") |
| print("\n[2] Task Decomposer") |
| dag_nodes = build_dag(vote["validated"], request) |
| dag_id = dag_nodes[0]["dag_id"] |
| print(f" -> DAG {dag_id} nodes: {[n["node_id"] for n in dag_nodes]}") |
| print("\n[3] Shared Context Board") |
| board = Blackboard() |
| pattern = {"pattern_id":"CLOSED_LOOP_01","version":"1.0","dag_id":dag_id,"source_module":"physics_expert","source_model_hash":"sha256:physics_v2.1","ttl_ms":5000,"pattern_type":"topological","abstraction":{"name":"closed loop","properties":{"conservation":True}},"vector_embedding":[0.1,0.9],"access_control":{"write":["physics_expert"],"read":["code_expert"]},"audit":True} |
| board.publish(pattern) |
| print(f" -> Published {pattern['pattern_id']}") |
| print("\n[4] Experts (parallel)") |
| phys_out = physics_infer(dag_nodes[0], board.subscribe()) |
| code_out = code_infer(dag_nodes[1], phys_out, board.subscribe()) |
| lang_out = lang_infer({"node_id":"T_lang_01"}) |
| print(f" -> Physics c={phys_out['c_i_calibrated']} Code c={code_out['c_i_calibrated']}") |
| print("\n[5] Fusion Layer") |
| result = fuse([phys_out, code_out, lang_out], dag_id) |
| print(f" -> Flag: {result['flag']} conf={result['confidence_final']:.2f} iters={result['provenance_graph']['fusion_iterations']}") |
| print("\n=== Final Answer ===\n", result["final_answer"][:800]) |
| return result |
| if __name__ == "__main__": |
| run("Write Python simulation for heat exchanger water/water with Q_dot 15kW") |
|
|
