{
  "bench_version": 1,
  "track": "poisson_boundary",
  "model": "mlp_tiny",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.01
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001
        },
        "mean": 0.046625981107354164
      },
      {
        "cfg": {
          "lr": 0.003
        },
        "mean": 0.005821306491270661
      },
      {
        "cfg": {
          "lr": 0.01
        },
        "mean": 0.0026237370329909027
      }
    ],
    "full": {
      "mean": 0.0022902884520590305,
      "std": 0.0008475947478694112,
      "per_seed": [
        0.002140381373465061,
        0.0017085936851799488,
        0.0028188657015562057,
        0.003827107371762395,
        0.0007765397895127535,
        0.0018183346837759018,
        0.002519753761589527,
        0.002712731249630451
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 0.01020078727742657,
    "std": 0.00296426428649209,
    "per_seed": [
      0.009175088256597519,
      0.007033985573798418,
      0.006878904066979885,
      0.015077905729413033,
      0.014966735616326332,
      0.009611713699996471,
      0.009160290472209454,
      0.009701674804091454
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 0.00791049882536754,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      0.007034706883132458,
      0.005325391888618469,
      0.004060038365423679,
      0.011250798357650638,
      0.014190195826813579,
      0.00779337901622057,
      0.0066405367106199265,
      0.006988943554461002
    ],
    "p_value": 0.0081,
    "mde": 0.0027432218817820646,
    "mde_rel_pct": 119.77626134017463,
    "verdict": "idea worse (significant)",
    "system_worked": false
  },
  "mechanism_signature": {
    "idea_config": {
      "lr": 0.003
    },
    "idea_sweep": [
      {
        "cfg": {
          "lr": 0.001
        },
        "result": {
          "mean": 0.052139872685074806,
          "std": 0.010604419760995165,
          "per_seed": [
            0.055915020406246185,
            0.04067733511328697,
            0.048677653074264526,
            0.06918065994977951,
            0.03635910525918007,
            0.0652485340833664,
            0.05410538986325264,
            0.04695528373122215
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.003
        },
        "result": {
          "mean": 0.01020078727742657,
          "std": 0.00296426428649209,
          "per_seed": [
            0.009175088256597519,
            0.007033985573798418,
            0.006878904066979885,
            0.015077905729413033,
            0.014966735616326332,
            0.009611713699996471,
            0.009160290472209454,
            0.009701674804091454
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.01
        },
        "result": {
          "mean": 0.011461915099062026,
          "std": 0.0024442795764689383,
          "per_seed": [
            0.009399672970175743,
            0.0106451865285635,
            0.009206349961459637,
            0.012322325259447098,
            0.01035053189843893,
            0.011077546514570713,
            0.011263057589530945,
            0.01743065007030964
          ],
          "n": 8
        }
      }
    ],
    "math_check": {
      "sphere_second_moment_increment": 0.1600427977741897,
      "predicted_increment": 0.16000000000000003,
      "relative_increment_error": 0.00026748608868542434,
      "variance_scaling": [
        {
          "K": 1,
          "variance": 0.4991379976272583,
          "K_times_variance": 0.4991379976272583
        },
        {
          "K": 2,
          "variance": 0.24831385910511017,
          "K_times_variance": 0.49662771821022034
        },
        {
          "K": 4,
          "variance": 0.12296506762504578,
          "K_times_variance": 0.4918602705001831
        },
        {
          "K": 8,
          "variance": 0.060902342200279236,
          "K_times_variance": 0.4872187376022339
        },
        {
          "K": 16,
          "variance": 0.03029288910329342,
          "K_times_variance": 0.4846862256526947
        },
        {
          "K": 32,
          "variance": 0.014903044328093529,
          "K_times_variance": 0.4768974184989929
        },
        {
          "K": 64,
          "variance": 0.007048895116895437,
          "K_times_variance": 0.451129287481308
        },
        {
          "K": 128,
          "variance": 0.0035142002161592245,
          "K_times_variance": 0.44981762766838074
        }
      ],
      "K_product_cv": 0.03766105478867267,
      "confirmed": true
    },
    "mechanism_signature": {
      "prediction": "WOS stochastic supervision does not create a systematic output mean shift",
      "predicted_shift": 0.0,
      "observed_shift_mean": -0.013755707695963793,
      "observed_shift_abs_mean": 0.028182248890516348,
      "trained_systems": [
        {
          "seed": 0,
          "baseline_mse": 0.005913021042943001,
          "idea_mse": 0.009175088256597519,
          "trained_prediction_mean_shift_idea_minus_baseline": 0.00194565171841532
        },
        {
          "seed": 1,
          "baseline_mse": 0.003503378015011549,
          "idea_mse": 0.007033985573798418,
          "trained_prediction_mean_shift_idea_minus_baseline": -0.02376757189631462
        },
        {
          "seed": 2,
          "baseline_mse": 0.003274704795330763,
          "idea_mse": 0.006878904066979885,
          "trained_prediction_mean_shift_idea_minus_baseline": 0.0016674669459462166
        },
        {
          "seed": 3,
          "baseline_mse": 0.010594122111797333,
          "idea_mse": 0.015077905729413033,
          "trained_prediction_mean_shift_idea_minus_baseline": 0.023475755006074905
        },
        {
          "seed": 4,
          "baseline_mse": 0.0036502110306173563,
          "idea_mse": 0.014966735616326332,
          "trained_prediction_mean_shift_idea_minus_baseline": -0.0916648656129837
        },
        {
          "seed": 5,
          "baseline_mse": 0.004745990503579378,
          "idea_mse": 0.009611713699996471,
          "trained_prediction_mean_shift_idea_minus_baseline": 0.023652706295251846
        },
        {
          "seed": 6,
          "baseline_mse": 0.006178014911711216,
          "idea_mse": 0.009160290472209454,
          "trained_prediction_mean_shift_idea_minus_baseline": 0.0069645848125219345
        },
        {
          "seed": 7,
          "baseline_mse": 0.0056354450061917305,
          "idea_mse": 0.009701674804091454,
          "trained_prediction_mean_shift_idea_minus_baseline": -0.05231938883662224
        }
      ],
      "confirmed": false
    },
    "custom_track": {
      "name": "poisson_boundary",
      "file": "bench/custom_tracks/poisson_boundary.py",
      "domain": "pde"
    },
    "structural_match": "PDE Dirichlet boundary-value regression on the unit disk; WOS is the intervention."
  }
}