{
 "artifacts": null,
 "category": "architecture",
 "description": "Replace unconstrained token-to-expert routing with a nonnegative mixture of a small set of feasible routing or communication patterns. Enforce resource limits using the paper's one-sided positive-contribution bound, which gives a conservative certificate without enumerating all joint token activation scenarios. Expand the pattern set only when a separation procedure finds a routing pattern that improves the router objective while adding useful capacity information.",
 "formulas_latex": [
  "$$nr=\\sum_{i=1}^{T}p_i f_i,\\qquad f_i\\geq 0$$",
  "$$\\sum_{i\\in T}W_k^i f_i\\leq \\widetilde{F_k^{max}},\\qquad \\forall k\\in CNE$$",
  "$$W_k^{n,z}:=\\max\\left(PTDF_{k,n}-PTDF_{k,z},0\\right)$$",
  "$$r=\\sum_{i=1}^{T}p_i f_i,\\qquad \\sum_{i=1}^{T}\\max\\left(0,a_k^{\\top}p_i\\right)f_i\\leq b_k\\quad\\forall k$$"
 ],
 "id": 2971,
 "implementation": "Integrate the method at the router output of a sparse MoE Transformer, before dispatching tokens to experts. Let E be the number of experts and let K index resource constraints such as per-device capacity, all-to-all bandwidth, memory, or latency. Maintain a pattern matrix P = [p_1,...,p_T], where each p_i is a normalized expert-load or signed traffic pattern, and nonnegative coefficients f. For each batch, compute router logits g(x), aggregate the desired load vector r_target, and solve a small constrained projection: minimize ||P f - r_target||_2^2 + tau ||f||_2^2 subject to f \u003e= 0 and sum_i W_ki f_i \u003c= b_k for every k. Here W_ki = max(0, a_k^T p_i). Convert the projected vector r = P f into top-1 or top-2 dispatch probabilities. A simple implementation can use 10 to 30 projected-gradient iterations or a differentiable QP layer. Every 100 to 500 training steps, perform column generation: sample candidate patterns by perturbing overloaded expert assignments, compute each candidate's reduced-cost proxy using the current router residual and capacity slack, and append the best nonduplicate pattern. Estimate a_k from recent token counts, transmitted bytes, or measured communication time; b_k is the hardware budget. The mathematical part is the one-sided W bound; resource estimates and pattern generation are empirical. First test on an 8-expert, 12-layer MoE language model with WikiText-103 or C4, comparing against Switch top-1 and top-2 routing at equal model size and FLOPs. Track perplexity, dropped-token rate, peak memory, communication volume, and per-step latency. Success means at least 50 percent fewer dropped tokens and lower communication variance with no more than 5 percent perplexity degradation, or equal perplexity with lower peak memory and stable constraint satisfaction.",
 "math_summary": "Theorem 1 uses exchange-pattern columns p_i and nonnegative coefficients f_i to represent a net-injection vector as nr = sum_i p_i f_i, with f_i \u003e= 0. For each network constraint k, the paper defines a secure per-pattern capacity W_k^i and imposes sum_i W_k^i f_i \u003c= F_k^max. In the power-flow specialization, W_k^(n,z) = max(PTDF_(k,n) - PTDF_(k,z), 0) for the exchange e_n - e_z. The crucial one-sided inequality is sum_z PTDF_(k,z) nrAct_(z,s) \u003c= sum_i W_k^i f_i \u003c= F_k^max, which certifies every activation scenario represented by the same nonnegative pattern coefficients. Adapt this to a router as follows: p_i in R^E is a routing or traffic pattern over E experts, f_i \u003e= 0 is its batch-level usage coefficient, r in R^E is the resulting expert-load vector, a_k in R^E is the measured resource-incidence vector for constraint k, b_k is its capacity, and W_k^i = max(0, a_k^T p_i). The conservative feasible router is r = sum_i p_i f_i with sum_i W_k^i f_i \u003c= b_k. The max prevents unsafe cancellation between individually costly patterns, while the active column pool avoids enumerating all joint routing configurations.",
 "math_tags": [
  "optimization",
  "convex-analysis",
  "linear-algebra",
  "combinatorics"
 ],
 "ml_areas": [
  "moe-routing",
  "transformer",
  "training",
  "inference-speedup"
 ],
 "paper": {
  "arxiv_id": "2609.00439",
  "arxiv_url": "https://arxiv.org/abs/2609.00439",
  "summary_what_math_gives_to_ml": "The paper develops a constructive way to replace exponentially many activation scenarios by a conservative polyhedral inner approximation built from reusable exchange patterns. Its key asset is a one-sided decomposition inequality: if a feasible activation is represented as a nonnegative combination of patterns, then bounding the positive contribution of every pattern separately certifies all combined activations without enumerating every joint scenario. This structure can transfer to constrained mixture-of-experts routing, where expert and device traffic patterns must remain feasible under arbitrary token activation. The practical adaptation is a pattern-based router with a small active column pool and periodic separation that adds only useful patterns violating capacity or communication constraints.",
  "title": "Models and Algorithms for Reserve Deliverability in Cross-Zonal Balancing Capacity Markets",
  "year": "2026"
 },
 "ratings": {
  "difficulty": 6,
  "novelty": 7,
  "usefulness": 6
 },
 "solves": [
  "speedup",
  "stability",
  "memory"
 ],
 "title": "Polyhedral Column-Generation MoE Router",
 "url": "https://synthcore.org/idea/2971/polyhedral-column-generation-moe-router",
 "verification": {
  "peer_reviewed": false,
  "status": "unverified",
  "status_label": "Unverified",
  "verdict_source": "deterministic test code (paired-seed permutation statistics)",
  "verification_axes": {
   "benchmark_mechanism": {
    "confirmed": null,
    "tested": false
   },
   "practical_benchmark": {
    "beats_baseline": null,
    "tested": false
   },
   "toy_mechanism_gate": {
    "confirmed": null,
    "tested": false
   }
  }
 }
}