CMA Optimization

Use Computalot as the parallel evaluator in a client-driven optimization loop.

Per-generation evaluation


{
  "type": "structured_runner",
  "runner_command": ["python3", "evaluate.py"],
  "payload": {"dataset": "smoke"},
  "fan_out": {
    "items": [
      {"candidate_id": 0, "params": [0.1, 0.5, 0.3]},
      {"candidate_id": 1, "params": [0.2, 0.4, 0.6]}
    ]
  },
  "merge_strategy": "collect",
  "project": "my-project",
  "timeout_s": 600
}

Runner script


import json, os
 
payload = json.load(open(os.environ["COMPUTALOT_TASK_PAYLOAD"]))
fitness = your_objective_function(payload["params"])
 
with open(os.environ["COMPUTALOT_TASK_RESULT"], "w") as f:
    json.dump({"fitness": fitness, "params": payload["params"]}, f)