Resolve economy organ toolchains: ECLiPSe 7.2 + COIN-OR, M3g to Fortran, BoundedPrediction fields

Hook:
- ECLiPSe upgraded 7.1_13 → 7.2_13, adds ic + eplex (if_osiclpcbc) with
  COIN-OR system dep, sha256 pinned, ECLIPSEDIR exported, correct paths
  (lib/x86_64_linux/eclipse.exe not bin/)
- fpm switched from GitHub binary download to pip (0.12.0) — proxy blocks
  GitHub release downloads in this environment
- Alire download gets sha256 verification on both install and presence check
- Fortran comment updated M3d,M3e → M3d,M3e,M3g
- Foundry section comment clarified as hosted separately

Specs:
- M3g §3: Zig → Fortran 2018 (gfortran/fpm/OpenBLAS, hand-rolled Riccati)
- M3 hub §3: sub-process list Zig(M3g) → Fortran(M3g)
- M3 hub §5: BoundedPrediction adds token_ticker and recent_shift (ground
  truth from M2, same source as correctness scoring and calibration)
- M3b §3: ECLiPSe 7.2 + ic + eplex (COIN-OR CLP/CBC)
- M3f §3: ECLiPSe 7.2 + ic + eplex (COIN-OR CLP/CBC)
- M3c §3: Solidity + Foundry confirmed, hosted separately, same Hub path
This commit is contained in:
Claude
2026-07-18 03:55:43 +00:00
parent 700a1825bf
commit e408409b77
6 changed files with 77 additions and 45 deletions
@@ -15,9 +15,11 @@ optimal execution C5 (industry standard since 2001; crypto adaptations validated
Kurz CMC thesis). DEX-specific microstructure C2 (emerging). Implementation C1.
## 3. Language & location
TBD · `src/economy/sims/microstructure/`. **Zig** — tick-level event-driven simulation with
deterministic memory layout, no GC pauses, and sub-microsecond latency for Riccati solvers and
order-book state updates; comptime generics eliminate runtime dispatch on hot paths.
**Fortran 2018** (gfortran) · `src/economy/sims/microstructure/`. Build: **fpm**. Dependencies:
**OpenBLAS** (LAPACK/BLAS via native Fortran interfaces). Hand-rolled: Riccati ODE solver,
order-book state arrays, JSON I/O against fixed schemas. Almgren-Chriss optimal execution is a
dense ODE (Riccati equations) — Fortran's home turf; LAPACK is native, array intrinsics map
directly to order-book depth vectors, and zero new toolchain is needed (same as M3d/M3e).
## 4. Does / does-not
- **Does:** simulate order flow across venues (DEXs and CEXs); model bid-ask spread dynamics as a