Task 007: Homework for Meeting #10¶
Priority: HIGH Status: DONE Due: Monday 2026-03-23 (Meeting #10) Source: Meeting #9 homework assignment
Assignment¶
From Meeting #9:
- Get agents to improve sparse parity using DMC (Data Movement Complexity) as the energy proxy
- Iterate on prompts and meta-approaches to go from "metric spec + problem spec" to experiments quickly
Checklist¶
DMC Baseline Sweep¶
- Run DMC measurement for GF(2) Gaussian Elimination
- Run DMC measurement for KM Influence Estimation
- Run DMC measurement for SGD (standard config: n=20, k=3, hidden=200, lr=0.1)
- Run DMC measurement for SMT Backtracking
- Run DMC measurement for Fourier/Walsh-Hadamard
- Compile comparison table: method / accuracy / DMC / ARD / wall time
DMC Optimization¶
- Pick the most promising method for DMC optimization
- Run at least one experiment targeting DMC reduction
- Record results in findings format (hypothesis, method, result, key number)
- Update DISCOVERIES.md if findings answer an open question
Presentation Prep¶
- Prepare results summary (table + key insight)
- Note any differences between DMC ranking and ARD ranking
- Record video or prepare demo if presenting async
Context¶
DMC is already tracked in MemTracker alongside ARD. Formula: DMC = sum(sqrt(stack_distance)) for all float accesses. Current baseline from CLAUDE.md: ARD 4,104 / DMC 300,298.
Key question: does optimizing DMC lead to different algorithmic choices than optimizing ARD? If the rankings change, that's a finding.
Files¶
src/sparse_parity/tracker.py-- MemTracker with DMCsrc/sparse_parity/cache_tracker.py-- CacheTracker with LRU simulationsrc/sparse_parity/harness.py-- Locked evaluation harness (DO NOT MODIFY)