{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://snake-lab.local/schemas/simulation-config-v1.schema.json",
  "title": "SnakeLab simulation configuration",
  "description": "Configuration for one serial SnakeLab simulation run.",
  "type": "object",
  "additionalProperties": false,
  "properties": {
    "epochs": {
      "title": "Number of epochs",
      "description": "Number of complete Snake episodes to execute.",
      "type": "integer",
      "default": 100,
      "minimum": 50,
      "maximum": 5000,
      "x-snakelab-unit": "episodes",
      "x-snakelab-sweepable": true
    },
    "seed": {
      "title": "Simulation seed",
      "description": "Master seed used to derive independent model, exploration, replay, and episode seeds.",
      "type": "integer",
      "default": 1970,
      "minimum": 0,
      "maximum": 9223372036854775807,
      "x-snakelab-sweepable": true
    },
    "game": {
      "title": "Game environment",
      "type": "object",
      "additionalProperties": false,
      "properties": {
        "board_width": {
          "type": "integer",
          "default": 20,
          "minimum": 2,
          "maximum": 100,
          "x-snakelab-unit": "cells",
          "x-snakelab-sweepable": true
        },
        "board_height": {
          "type": "integer",
          "default": 20,
          "minimum": 1,
          "maximum": 100,
          "x-snakelab-unit": "cells",
          "x-snakelab-sweepable": true
        },
        "initial_snake_length": {
          "type": "integer",
          "default": 3,
          "minimum": 1,
          "maximum": 100,
          "x-snakelab-unit": "cells",
          "x-snakelab-sweepable": true
        },
        "max_moves_multiplier": {
          "description": "An episode ends when moves exceed this value times the current snake length.",
          "type": "integer",
          "default": 100,
          "minimum": 1,
          "maximum": 10000,
          "x-snakelab-sweepable": true
        },
        "rewards": {
          "type": "object",
          "additionalProperties": false,
          "properties": {
            "food": {
              "type": "number",
              "default": 10.0,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            },
            "wall": {
              "type": "number",
              "default": -10.0,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            },
            "snake": {
              "type": "number",
              "default": -10.0,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            },
            "max_moves": {
              "type": "number",
              "default": -10.0,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            },
            "empty": {
              "type": "number",
              "default": 0.0,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            },
            "closer_to_food": {
              "type": "number",
              "default": 0.1,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            },
            "further_from_food": {
              "type": "number",
              "default": -0.1,
              "minimum": -1000.0,
              "maximum": 1000.0,
              "x-snakelab-sweepable": true
            }
          },
          "required": [
            "food",
            "wall",
            "snake",
            "max_moves",
            "empty",
            "closer_to_food",
            "further_from_food"
          ]
        }
      },
      "required": [
        "board_width",
        "board_height",
        "initial_snake_length",
        "max_moves_multiplier",
        "rewards"
      ]
    },
    "model": {
      "title": "Recurrent model",
      "type": "object",
      "additionalProperties": false,
      "properties": {
        "hidden_size": {
          "type": "integer",
          "default": 192,
          "minimum": 1,
          "maximum": 4096,
          "x-snakelab-unit": "neurons",
          "x-snakelab-sweepable": true
        },
        "layers": {
          "type": "integer",
          "default": 3,
          "minimum": 1,
          "maximum": 16,
          "x-snakelab-sweepable": true
        },
        "dropout": {
          "type": "number",
          "default": 0.1,
          "minimum": 0.0,
          "exclusiveMaximum": 1.0,
          "x-snakelab-unit": "probability",
          "x-snakelab-sweepable": true
        }
      },
      "required": ["hidden_size", "layers", "dropout"]
    },
    "training": {
      "title": "Replay and training",
      "type": "object",
      "additionalProperties": false,
      "properties": {
        "sequence_length": {
          "type": "integer",
          "default": 8,
          "minimum": 1,
          "maximum": 256,
          "x-snakelab-unit": "frames",
          "x-snakelab-sweepable": true
        },
        "batch_size": {
          "type": "integer",
          "default": 64,
          "minimum": 1,
          "maximum": 4096,
          "x-snakelab-unit": "sequences",
          "x-snakelab-sweepable": true
        },
        "replay_max_frames": {
          "type": "integer",
          "default": 150000,
          "minimum": 1,
          "maximum": 10000000,
          "x-snakelab-unit": "frames",
          "x-snakelab-sweepable": true
        },
        "learning_rate": {
          "type": "number",
          "default": 0.002,
          "exclusiveMinimum": 0.0,
          "maximum": 1.0,
          "x-snakelab-sweepable": true
        },
        "gamma": {
          "type": "number",
          "default": 0.96,
          "minimum": 0.0,
          "maximum": 1.0,
          "x-snakelab-sweepable": true
        },
        "tau": {
          "type": "number",
          "default": 0.001,
          "exclusiveMinimum": 0.0,
          "maximum": 1.0,
          "x-snakelab-sweepable": true
        },
        "max_gradient_norm": {
          "type": "number",
          "default": 1.0,
          "exclusiveMinimum": 0.0,
          "maximum": 1000.0,
          "x-snakelab-sweepable": true
        }
      },
      "required": [
        "sequence_length",
        "batch_size",
        "replay_max_frames",
        "learning_rate",
        "gamma",
        "tau",
        "max_gradient_norm"
      ]
    },
    "epsilon": {
      "title": "Epsilon-greedy exploration",
      "description": "Episode-based epsilon-greedy exploration schedule.",
      "type": "object",
      "additionalProperties": false,
      "properties": {
        "initial": {
          "type": "number",
          "default": 0.96,
          "minimum": 0.0,
          "maximum": 1.0,
          "x-snakelab-unit": "probability",
          "x-snakelab-sweepable": true
        },
        "minimum": {
          "type": "number",
          "default": 0.0,
          "minimum": 0.0,
          "maximum": 1.0,
          "x-snakelab-unit": "probability",
          "x-snakelab-sweepable": true
        },
        "decay": {
          "type": "number",
          "default": 0.97,
          "exclusiveMinimum": 0.0,
          "maximum": 1.0,
          "x-snakelab-unit": "multiplier-per-episode",
          "x-snakelab-sweepable": true
        }
      },
      "required": ["initial", "minimum", "decay"]
    }
  },
  "required": ["epochs", "seed", "game", "model", "training", "epsilon"]
}