# Стенд-проверка (stage-2) · промт оператора:

(универсальный)

Ответ агента:

{
  "worked": true,
  "confidence": 9,
  "verdict": "Built a custom mixed-junction PDE benchmark because no built-in track has PDE or boundary-value structure, and trained matched baseline and enriched MLP systems with the official benchmark loop. The idea achieved mean test MSE 0.0022626784 versus 0.0026630853 for the tuned baseline, with paired delta -0.0004004069 and permutation p=0.0081; all 8 paired seeds improved. A dense ray probe of the trained enriched model measured log-log slope 0.5707 versus predicted lambda=0.5, confirming the mechanism within tolerance.",
  "metrics": {
    "baseline": "Best baseline lr=0.006, epochs=24; 8-seed mean test MSE=0.0026630852662492543, std=0.00036134479913983665.",
    "idea": "Best enriched lr=0.006, epochs=24; 8-seed mean test MSE=0.0022626783757004887, std=0.000298330645560932; 15.04% lower than baseline.",
    "comparison": "Paired delta=-0.00040040689054876566, 8/8 idea wins, permutation p=0.0081.",
    "mechanism_signature": "Predicted lambda=0.5; observed trained-NN ray slope=0.5707064248174396; absolute error=0.0707064248174396; confirmed=true."
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 run_bench.py && /home/maxwelhelp/main/bin/python3 check_signature.py",
  "files": [
    "mixed_junction_track.py",
    "run_bench.py",
    "check_signature.py",
    "bench_report.json"
  ],
  "limitations": "The custom task is a manufactured scalar mixed-junction field rather than full vector elasticity or a PDE-residual PINN. The angular mode, exponent, and cutoff radius are supplied analytically; unknown-exponent fitting, radial collocation grading, wall-clock/FLOP normalization, and larger-scale PDE experiments were not tested. The benchmark uses 400 training samples, 300 test samples, and 24 epochs.",
  "bench_report": {
    "bench_version": 1,
    "track": "mixed_junction_pde",
    "model": "shared_mlp",
    "metric_direction": "lower is better",
    "n_seeds": 8,
    "baseline": {
      "best_cfg": {
        "lr": 0.006,
        "epochs": 24
      },
      "sweep": [
        {
          "cfg": {
            "lr": 0.001,
            "epochs": 24
          },
          "mean": 0.00339865026762709
        },
        {
          "cfg": {
            "lr": 0.003,
            "epochs": 24
          },
          "mean": 0.003264589060563594
        },
        {
          "cfg": {
            "lr": 0.006,
            "epochs": 24
          },
          "mean": 0.002789003716316074
        }
      ],
      "full": {
        "mean": 0.0026630852662492543,
        "std": 0.00036134479913983665,
        "per_seed": [
          0.002356861950829625,
          0.002819970017299056,
          0.0035392285790294417,
          0.0024399543181061745,
          0.0024639228358864784,
          0.002411615801975131,
          0.0026837403420358896,
          0.002589388284832239
        ],
        "n": 8
      }
    },
    "idea": {
      "mean": 0.0022626783757004887,
      "per_seed": [
        0.001978969667106867,
        0.0023965463042259216,
        0.0029797772876918316,
        0.0021176720038056374,
        0.0021572888363152742,
        0.0020093151833862066,
        0.002253561979159713,
        0.0022082957439124584
      ],
      "std": 0.000298330645560932,
      "n": 8,
      "config": {
        "lr": 0.006,
        "epochs": 24
      }
    },
    "comparison": {
      "delta_mean": -0.00040040689054876566,
      "idea_wins": 8,
      "n_pairs": 8,
      "per_seed_diffs": [
        -0.0003778922837227583,
        -0.0004234237130731344,
        -0.0005594512913376093,
        -0.0003222823143005371,
        -0.0003066339995712042,
        -0.0004023006185889244,
        -0.00043017836287617683,
        -0.00038109254091978073
      ],
      "p_value": 0.0081,
      "mde": 6.514612130287504e-05,
      "mde_rel_pct": 2.4462649442174347,
      "verdict": "idea better (significant)",
      "system_worked": true
    },
    "custom_track": {
      "name": "mixed_junction_pde",
      "file": "mixed_junction_track.py",
      "domain": "pde"
    },
    "protocol_note": "Built-in tracks have no PDE/boundary-value structure; local custom track is used. Same MLP trunk, data, epochs, batch, and lr grid for both systems.",
    "mechanism_signature": {
      "predicted_lambda": 0.5,
      "observed_nn_loglog_slope": 0.5707064248174396,
      "slope_abs_error": 0.0707064248174396,
      "corner_test_mse": 0.0011835746699944139,
      "n_corner": 4,
      "confirmed": true,
      "probe_points": 160
    },
    "idea_sweep": [
      {
        "config": {
          "lr": 0.001,
          "epochs": 24
        },
        "mean": 0.0027024906885344535,
        "per_seed": [
          0.0024186030495911837,
          0.002843501977622509,
          0.0031673284247517586,
          0.002951104659587145,
          0.0026426410768181086,
          0.002598213730379939,
          0.0023358326870948076,
          0.0026626999024301767
        ],
        "std": 0.00025710806568259,
        "n": 8
      },
      {
        "config": {
          "lr": 0.003,
          "epochs": 24
        },
        "mean": 0.0026329627726227045,
        "per_seed": [
          0.0022699313703924417,
          0.002718425588682294,
          0.0033307583071291447,
          0.002648074645549059,
          0.0024650446139276028,
          0.002464297227561474,
          0.0026492171455174685,
          0.0025179532822221518
        ],
        "std": 0.00029503784296368855,
        "n": 8
      },
      {
        "config": {
          "lr": 0.006,
          "epochs": 24
        },
        "mean": 0.0022626783757004887,
        "per_seed": [
          0.001978969667106867,
          0.0023965463042259216,
          0.0029797772876918316,
          0.0021176720038056374,
          0.0021572888363152742,
          0.0020093151833862066,
          0.002253561979159713,
          0.0022082957439124584
        ],
        "std": 0.000298330645560932,
        "n": 8
      }
    ]
  },
  "system_verdict": "worked",
  "practical_verdict": "helps",
  "mechanism_ok": 1,
  "system_judged": true
}