ColdMkh

ColdMkh is a two-sided clearing exchange for quantum cold-capacity. Quantum algorithms commoditize, but the binding physical input is cryogenic cooling — dilution-refrigerator millikelvin capacity. ColdMkh aggregates fragmented idle cold-capacity from independent fridge operators into one metered exchange that trades guaranteed millikelvin-hours (mK·h) at a temperature grade. For an operator book and a job book it answers:

Which developer jobs can be matched to cold-enough, affordable, reliable supply, at what price, and does the market clear?

It clears each job against the cheapest eligible supply, allocates over finite capacity, and emits a posture verdict: cleared, thin, or blocked.

The Insight

Once an abstraction commoditizes, advantage moves to the scarce substrate underneath it (L10_Generative). Two lens moves make a market of it:

• L11 domain transplant — transplant "capture the scarce substrate" into quantum cryogenics. The traded good is millikelvin-hours at a temperature grade, not the fridge.
• L19 aggregation — aggregate fragmented idle cold-capacity across many operators into one metered clearing exchange. Operators monetize idle cryo time; developers buy substrate without owning fridges.

This is the first genuinely two-sided exchange among its IdeaFirst siblings: supply offers clear against demand bids.

Clearing Mechanics

• Temperature grade is a hard gate. A fridge reaches a coldest base_temp_mk; a job tolerates a warmest max_temp_mk. Eligible only if base_temp_mk ≤ max_temp_mk — deeper cold is scarcer.
• Two-sided price discovery. Operators post an ask; jobs post a bid ceiling. A fill happens only where ask ≤ bid, and each job fills from the cheapest eligible supply first.
• Reliability gate. A fridge's reliability must meet the job's min_reliability.
• Finite, fragmented supply. Capacity is consumed in a deterministic clearing order (jobs in bid priority), so aggregation over scarce real capacity is explicit.

Quickstart

python -m coldmkh sample --output examples/market_sample.json
python -m coldmkh clear  --input examples/market_sample.json
python -m coldmkh report --input examples/market_sample.json --output examples/coldmkh_report.md

Market Spec

{
  "policy": { "min_clear_fill_rate": 0.5, "platform_fee_rate": 0.02 },
  "operators": [
    { "operator_id": "fridge-A", "base_temp_mk": 8.0, "capacity_mkh": 1000.0,
      "price_per_mkh": 5.0, "reliability": 0.95 }
  ],
  "jobs": [
    { "job_id": "annealer", "demand_mkh": 700.0, "max_temp_mk": 10.0,
      "max_price_per_mkh": 9.0, "min_reliability": 0.95 }
  ]
}

How It Works

jobs clear in bid priority: sort by (max_price_per_mkh desc, job_id)
for each job:
  eligible operators = remaining capacity AND
      base_temp_mk <= max_temp_mk        # cold enough (temperature grade)
      price_per_mkh <= max_price_per_mkh # affordable (ask <= bid)
      reliability >= min_reliability     # stable enough
  allocate from cheapest eligible (price asc, operator_id) until demand met
    or eligible capacity exhausted; decrement operator capacity per fill
  status:
    allocated == demand  -> filled
    0 < allocated < demand -> partial (insufficient_capacity)
    allocated == 0       -> unfilled (too_warm / over_budget / below_reliability)
  each fill priced pay-as-bid at the operator's ask
platform_fee = platform_fee_rate × total traded value

The market posture aggregates:

• matched ≥ min_clear_fill_rate share of demand → cleared
• some matched but below threshold (illiquid) → thin
• nothing matched → blocked (exit 1)

An unknown field, a negative magnitude, or a non-positive temperature/demand is a blocking issue with a nonzero exit, never a stack trace.

MVP Scope

ColdMkh produces a clearing plan and market-liquidity verdict under a declared policy, not a binding cryo-capacity contract, an SLA, or a settlement of funds. It matches and flags; it does not reserve fridges or move money.

Provenance

ColdMkh is the concrete beachhead of IDEA-006 (Cold-Capacity Market for Quantum Substrate — EVX-20260607-005 winner after derivative exclusion), reached via the L11_DomainTransplant → L19_Aggregation lens stack (CIX-20260607-001 ← IDX-20260607-001 ← TCX-20260606-001). It is a sibling of WasteStack, FailureFutures, PowerRoam, CoverGate, RoboTrace, InferMesh, CertMesh, SettleMesh, SpendMesh, ReleaseMesh, ClimateMesh, SovMesh, FlowMesh, and SignalMesh.

License

MIT — see LICENSE.