d-bis/cross-chain-pmm-lps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
main
cross-chain-pmm-lps/scripts
History

README.md

Scripts

size-inventory.cjs

Regenerates I_T^* (inventory target) and suggested D_0 per chain per cW token from config/simulation-params.json. Output includes assumptions (σ, T_refill/T_epoch, β, γ). Keeps configs honest and PRs reviewable.

Run from repo root (cross-chain-pmm-lps):

node scripts/size-inventory.cjs

Options:

  • --sigma 1.8 — override USD stress multiplier (default 1.5)
  • --refill-ratio 0.33 — T_refill / T_epoch (default 0.33)
  • --depth-mult 0.75 — D_0 = depth_mult * I_T^* (default 0.75)
  • --v-epoch '{"1":100000,"56":80000,"137":60000}' — per-chain V_epoch (JSON object)

Environment: V_EPOCH_<chainId> overrides V_epoch for that chain (e.g. V_EPOCH_1=100000).

Output: JSON with assumptions and per-chain per-token I_T_star, D_0, V_epoch, sigma, beta, gamma, and EUR stress flag where applicable.

validate-deployment-status.cjs

Validates config/deployment-status.json for minimum viable deployed graph. Use in CI so deployment-realistic sim cannot run with half-filled state.

Rules:

  • If bridgeAvailable === true on a chain, cwTokens must include at least cWUSDT and cWUSDC (phase 1).
  • For each pmmPool and each pmmPoolsVolatile[] entry: role ∈ {defense, public_routing, truu_routing}; feeBps and k present; base/quote (or tokenIn/tokenOut) exist in cwTokens or anchorAddresses (e.g. mainnet TRUU under anchorAddresses.TRUU). Non-zero poolAddress must not be the zero address.
  • For each gasReferenceVenues[] entry: supported must be explicit, aggregatorOnly is only valid for 1inch, supported=false means the lane is explicitly unsupported (and can never be live), and supported=true with live=false / routingVisible=false means the lane is planned or staged.
  • aggregatorOnly=true rows are routing overlays, not stand-alone live venues.

Run:

node scripts/validate-deployment-status.cjs

Exit code: 0 if valid, 1 if invalid (errors to stderr).

Native protocol-only examples:

node scripts/validate-deployment-status.cjs config/validation-examples/native-protocol-only.valid.json
node scripts/validate-deployment-status.cjs config/validation-examples/native-protocol-only.invalid.json

The invalid example intentionally includes a placeholder scaffold and a non-native live 1inch row to exercise the rejection path.

Parent proxmox repo: live Mainnet cW/TRUU pool deploy and ratio-matched top-up scripts live under scripts/deployment/ (deploy-mainnet-pmm-cw-truu-pool.sh, add-mainnet-truu-pmm-topup.sh); see docs/03-deployment/MAINNET_PMM_TRUU_CWUSD_PEG_AND_BOT_RUNBOOK.md §11.

run-scenario.cjs

Builds the real routing graph from configs, runs epochs with PMM state updates, path enumeration + waterfilling, arb step (implied-price deviation, capped corrective trades, profit gate), bot step (inject/withdraw at 0.5×/1.5× I_T^* with intervention cost β/γ/ρ), and optional bridge shock trades. Emits a real scorecard (PR#2). Runs are deterministic when scenario.seed is set or derived from scenario name.

Scenarios may also include an optional microTradePolicy for gas-budgeted support trades in selected USD wrapper rails. This is meant to model tiny, controlled cWUSDC / cWUSDT turnover-seeding or inventory-corrective activity, not to turn the PMM into a venue.

Configs used: simulation-params.json, token-map.json, routing-controls.json; deployment-status.json only when graphMode = deployed. Pool topology from pool-matrix.json and scenario topology / fullQuoteChains.

Tuning constants (in script):

Constant Value Purpose
PROBE_SIZE 1000 Units for path cost probe (k-shortest by cost)
K_PATHS 5 Max candidate paths per trade for waterfilling
CHUNK_FRACTION 0.05 5% of trade per chunk; marginal-equalization step size
AMM_DEPTH_UNITS 10e6 Background AMM depth (notional; infinite-depth approx in code)
AMM_FEE_BPS 5 Fee for anchor↔anchor stable swaps

Run:

node scripts/run-scenario.cjs hub_only_11
node scripts/run-scenario.cjs --scenario full_quote_1_56_137
node scripts/run-scenario.cjs bridge_shock_137_56
node scripts/run-scenario.cjs micro_support_usd_wrappers_1_56_137
node scripts/run-scenario.cjs crash_40pct_external_asset
node scripts/run-scenario.cjs high_vol_sigma_spike
node scripts/run-scenario.cjs bank_run_redemption_spike
node scripts/run-scenario.cjs --optimizer leverage_sweep_1x_to_4x

Output: JSON scorecard including: capture_mean, churn_mean, drain_half_life_epochs, path_concentration_index; intervention_cost_total / intervention_cost_inject_total / intervention_cost_withdraw_total / intervention_cost_by_chain / intervention_cost_per_1M_volume; micro_trade_count / micro_trade_volume_total / micro_trade_gas_cost_total; peak_deviation_bps (post-arb), peak_deviation_bps_pre_arb, peak_deviation_bps_post_arb, peak_deviation_bps_post_bot; arb_volume_total, arb_profit_total (execution-based, not mid). See docs/12-sim-scorecard.md and config/scorecard-schema.json.

When capitalEfficiency.enabled = true, output also includes Monte Carlo capital-risk metrics: roi_mean, roi_p05, roi_p95, pnl_distribution, max_drawdown_p95, liquidation_probability, peg_deviation_frequency, external_liquidity_floor_violations, volatility_throttle_events, and spread_adjustment_events. Optimizer mode emits ranked parameter candidates and never treats leverage above the configured hard ceiling as deployable.

Orderflow: Trade sizes use distribution: "uniform" by default. Scenario schema supports lognormal / pareto for skewed (many small + occasional whale) flows; implement in sampleTrade() when needed.

validate-capital-efficiency.cjs

Runs CI-style checks for the simulation-only capital efficiency overlay:

  • JSON parse for policy, schemas, and scenarios
  • Baseline scenario remains capital-overlay free
  • Stress scenarios emit capital risk fields
  • Deterministic repeat check
  • Optimizer deployable candidates respect maxDeployableLeverage

Run:

node scripts/validate-capital-efficiency.cjs
Reference in New Issue View Git Blame Copy Permalink