Forge Pool

Appearance

Monte Carlo ​

mc@1 is the canonical probabilistic execution family of the Forge Pool Planetary Kernel.

It runs deterministic shard-based simulation across Forge Pool infrastructure and returns replay-grade results, metrics, and execution traces.

Identity ​

json
{
  "op": {
    "name": "mc",
    "version": 1,
    "profile": "eta.v1"
  }
}

mc@1 is selected through the canonical execute contract, not through a dedicated route.

http
POST /api/v0/ops/execute

What mc@1 is For ​

Monte Carlo is used when outcomes are better expressed as distributions than as single-point predictions.

Typical use cases:

  • logistics ETA under uncertainty
  • insurance frequency/severity modeling
  • climate ensemble simulation
  • portfolio stress distributions
  • compound event modeling

Execution Model ​

Monte Carlo follows the Planetary Kernel lifecycle:

  1. request validation
  2. deterministic planning
  3. shard execution across agents
  4. optional verification
  5. deterministic reduction
  6. replay metadata return
  7. artifact persistence when requested

This means identical inputs remain reproducible across execution runs.

Profiles ​

Profiles define the workload-specific schema and sampling behavior layered on top of mc@1.

Examples include:

  • eta.v1
  • insurance.loss.v1
  • risk.compound.v1
  • climate.ensemble.v1
  • finance.portfolio.v1
  • energy.price.v1

Profiles are not endpoints. They are configuration surfaces inside a stable Kernel family.

Example Request ​

json
{
  "ctx": {
    "job_id": "eta-demo-001",
    "billing": {
      "mode": "test"
    }
  },
  "op": {
    "name": "mc",
    "version": 1,
    "profile": "eta.v1"
  },
  "seed": {
    "mode": "explicit",
    "value": "eta-seed-001"
  },
  "policy": {
    "target": "cpu",
    "min_agents": 1,
    "max_agents": 32,
    "verify": "spotcheck"
  },
  "args": {
    "iterations": 1000000,
    "base_eta": 3600,
    "traffic_mean": 120,
    "traffic_std": 60,
    "weather_prob": 0.2,
    "weather_mean": 300,
    "weather_std": 120,
    "incident_prob": 0.05,
    "incident_mean": 600,
    "incident_std": 200
  }
}

What the Response Contains ​

A successful result returns:

  • job identity
  • execution status
  • hub metrics
  • shard execution metadata
  • reduced probabilistic output
  • replay metadata
  • optional artifact references

The exact result surface depends on the selected profile.

Determinism Rules ​

Monte Carlo on Forge Pool is deterministic by contract:

  • identical workload identity + args + seed -> identical result
  • shard planning must be reproducible
  • reduction ordering must not change the final output
  • agent routing must not alter the final distribution

For full details, see: