Skip to content

Simulation algorithm backlog (agent todos)

Status: Active
Registry: benchmarks/competitive/algo_registry.json

todos: - id: gap-sim-sim-p1-num-dot-axpy content: "sim: pending plan todo sim-p1-num-dot-axpy" status: pending gap_orchestrator: true - id: gap-sim-sim-p1-md-neighbor-cell content: "sim: pending plan todo sim-p1-md-neighbor-cell" status: pending gap_orchestrator: true - id: gap-sim-sim-p2-qm-dft-scf content: "sim: pending plan todo sim-p2-qm-dft-scf" status: pending gap_orchestrator: true

  • id: sim-p0-md-lj-li-parity content: "Tier-2 Li md_lennard_jones checksum parity vs native (fix runtime sink / driver)" status: pending

  • id: sim-p0-heat-li-smoke content: "Pure-Li heat_equation_2d smoke with harness verify row" status: pending

  • id: sim-p1-num-dot-axpy content: "sim: pending plan todo sim-p1-num-dot-axpy — gap orchestrator" status: pending

  • id: sim-p1-md-neighbor-cell content: "sim: pending plan todo sim-p1-md-neighbor-cell — gap orchestrator" status: pending

  • id: sim-p2-qm-dft-scf content: "sim: pending plan todo sim-p2-qm-dft-scf — gap orchestrator" status: pending

Agent instructions

  • One todo per loop iteration (sim-plan-loop.py).
  • After each slice: ./scripts/sim-plan-gates.sh (uses bench-package.sh, not full tier-12).
  • Update implemented_smoke in registry when a composable or tier-2 row exists.