{
  "design": "B: expected edge per side; blocked during instrumented tuning",
  "blocked": {
    "step": "gas-valid baseline and candidate tuning",
    "error": "The exact requested V2 source (SHA-256 28417e116f57f1029fdd1803832e852574902c1ae7253c05657f317db76da4ab), compiled with the pinned repository compiler, incurred 3 OutOfGas callbacks across seeds 30000..30255 at the unchanged 250000-gas call limit. The initial Design B prototype also incurred OutOfGas. Any OOG invalidates a result under the brief, so no valid paired gain or frozen candidate is submitted."
  },
  "paired": {
    "n": 0,
    "mean_diff": null,
    "se": null,
    "worst5": []
  },
  "tuning": [],
  "diagnostics": {
    "simulator_commit": "6b6921f8d463314aa7c96234e1be25a25245f704",
    "compiler": {
      "solc": "0.8.24",
      "optimizer_runs": 200,
      "viaIR": true,
      "evmVersion": "paris"
    },
    "baseline": {
      "sha256": "28417e116f57f1029fdd1803832e852574902c1ae7253c05657f317db76da4ab",
      "n": 256,
      "seed_start": 30000,
      "seed_end": 30255,
      "gas": {
        "max_ok": 243625,
        "oog": 3,
        "other_halts": 0
      },
      "normalizer_gas": {
        "max_ok": 22592,
        "oog": 0,
        "other_halts": 0
      }
    },
    "prototype_checkpoint": {
      "variant": "B_s1_v1_r1",
      "completed_n": 96,
      "seed_start": 30000,
      "seed_end": 30095,
      "gas": {
        "max_ok": 245494,
        "oog": 204063,
        "other_halts": 0
      },
      "normalizer_gas": {
        "max_ok": 22592,
        "oog": 0,
        "other_halts": 0
      },
      "scope": "Last completed batch checkpoint, not a total for the interrupted run; subsequent in-flight calls are not included."
    },
    "counter_method": "Atomic maximum successful-call gas and separate OutOfGas/other-failure counters were inserted in EVMStrategy.call in amm_sim_rs/src/evm/strategy.rs, immediately after transact_commit and before the original ExecutionResult match. Counters were reset before each separate run_batch(X,VanillaStrategy,configs,8), then read afterward. Successful gas includes initialization, name and swap calls; deployment gas is excluded. Fee limits and execution behavior were unchanged. Normalizer identified by Vanilla strategy name.",
    "model": "Prototype replaced the original fee mapping and removed direction/activity/toxicity decay and state updates. It retained slow/fast price, sigma, arrival and size estimators. Per side it searched 5,15,25,35,45,55,65,75 bps. Router share used the rules equation with estimated normalizer reserves at estimated fair price and matching invariant; normalizer reserves are unobservable. Expected retail flow was lambdaHat*size/2. Arb probability approximated 1-exp(-1/(1+lambdaHat)) using a 16-fold rational exponential approximation. Conditional normalized arb loss used a five-node sigma quadrature of positive(gap-fee)^2/4. This is an approximate model, not an exact expectation or validated optimization.",
    "selection": "Eight parameter variants were compiled; only the first reached simulations before the sweep was interrupted. No top-two stage, freeze or fresh confirmation was performed. Seeds 4060000..4061999 were not used.",
    "baseline_reproduction": [
      {
        "seed": 30105,
        "max_ok": 242780,
        "oog": 2,
        "other_halts": 0
      },
      {
        "seed": 30238,
        "max_ok": 236377,
        "oog": 1,
        "other_halts": 0
      }
    ]
  },
  "notes": [
    "The prescribed setup script completed successfully once in a scratch checkout. Both V2 and the initial prototype passed amm-match validate, which did not prevent later callback OOG in simulations.",
    "Each strategy was simulated separately against compiled VanillaStrategy, never against one another. Configs drew size, arrival rate and sigma in that order from numpy.default_rng(seed), using the exact requested ranges and 10000 steps. Rust price seed and retail seed behavior were unchanged.",
    "No score, mean paired gain, standard error or leaderboard improvement is claimed. Empty tuning and paired n=0 mean no gas-valid comparison, not zero economic difference.",
    "The pinned baseline was not changed to hide OOG, the gas limit was not raised, and failing calls were not discarded from the gas counters. No invalid prototype is offered as deployable code."
  ]
}