{
 "artifacts": null,
 "category": "architecture",
 "description": "Replace independent softmax routing or attention normalization with a differentiable approximate distribution over one-to-one assignments, using the Bethe permanent as the partition-function surrogate. Constrain the allowed token-to-expert or query-to-key support graph to have high girth, which gives an explicit bound on the approximation error and reduces short-cycle-induced correlations.",
 "formulas_latex": [
  "$$\\operatorname{per}(A)=\\sum_{\\sigma\\in\\operatorname{Sym}(n)}\\prod_{i=1}^{n}A_{i,\\sigma(i)}=\\sum_{M\\in\\mathcal{M}(G_A)}\\prod_{e\\in M}A_e.$$",
  "$$F_B(\\beta;A)=\\sum_{(i,j)\\in E}\\left[\\beta_{ij}\\log\\frac{\\beta_{ij}}{A_{ij}}-(1-\\beta_{ij})\\log(1-\\beta_{ij})\\right],\\qquad \\operatorname{Bethe}(A)=\\exp\\left(-\\min_{\\beta\\in\\mathcal{B}_n}F_B(\\beta;A)\\right).$$",
  "$$\\operatorname{Bethe}(A)\\leq\\operatorname{per}(A)\\leq 2^{2n/g}\\operatorname{Bethe}(A)\\quad\\text{if }\\operatorname{girth}(G_A)\\geq g\\text{ and }g\\geq4\\text{ is even}.$$",
  "$$0\\leq\\log\\operatorname{per}(A)-\\log\\operatorname{Bethe}(A)\\leq\\frac{2n}{g}\\log 2.$$"
 ],
 "id": 3034,
 "implementation": "Integrate this as a sparse assignment-normalization layer between token/query representations and experts/keys. For a batch of $n$ tokens and $n$ experts, compute logits $s_{ij}$ only on edges of a fixed sparse bipartite graph $E$, set $A_{ij}=\\exp(s_{ij}/\\tau)$ on those edges and $A_{ij}=0$ elsewhere, and require each row and column to sum to one through the Bethe optimization variable $\\beta$. Use projected mirror descent or damped belief-propagation-style updates to minimize $F_B(\\beta;A)$: initialize positive feasible $\\beta$ by sparse Sinkhorn scaling; repeatedly compute the gradient $\\partial F_B/\\partial\\beta_{ij}=\\log(\\beta_{ij}/A_{ij})+\\log(1-\\beta_{ij})+2$; take a mirror or exponentiated-gradient step; then approximately project onto the sparse doubly-stochastic constraints with 5--20 Sinkhorn iterations. Return $\\beta$ as routing weights, optionally with top-1 or top-2 dispatch after balancing. The theorem is used to choose the graph: construct a random or algebraically generated sparse bipartite graph, measure its shortest even cycle with BFS, and retain patterns with girth $g\\ge6$ or $8$. The mathematical quantity computed exactly is the certified bound $(2n/g)\\log2$; the Bethe minimizer and its gradients are numerical approximations. For a first experiment, compare a 12-layer small Transformer MoE with standard independently normalized top-2 routing, sparse Sinkhorn routing, and high-girth Bethe routing on CIFAR-100 or WikiText-2. Match active edges and FLOPs. Measure training loss at equal steps, expert-load coefficient of variation, routing entropy, and the gap between Bethe and exact permanent on small $n\\leq10$ instances. Success means lower load imbalance and more stable loss, or improved loss at equal dispatch cost; the strongest falsifiable signal is that the observed log-permanent error decreases as the support girth increases in accordance with the $(2n/g)\\log2$ ceiling.",
 "math_summary": "For a nonnegative matrix $A\\in\\mathbb{R}_{\\ge 0}^{n\\times n}$, the permanent is $\\operatorname{per}(A)=\\sum_{\\sigma\\in\\operatorname{Sym}(n)}\\prod_{i=1}^n A_{i,\\sigma(i)}$, equivalently the weighted sum over perfect matchings of the bipartite support graph $G_A$. Let $\\mathcal{B}_n=\\{\\beta\\in\\mathbb{R}_{\\ge0}^{n\\times n}:\\sum_j\\beta_{ij}=1,\\ \\sum_i\\beta_{ij}=1\\}$ be the Birkhoff polytope, and let $E=\\{(i,j):A_{ij}\u003e0\\}$. The Bethe approximation can be written as $\\operatorname{Bethe}(A)=\\exp(-\\min_{\\beta\\in\\mathcal{B}_n,\\ \\beta_{ij}=0\\ \\text{if }A_{ij}=0}F_B(\\beta;A))$, with Bethe free energy $F_B(\\beta;A)=\\sum_{(i,j)\\in E}\\left[\\beta_{ij}\\log\\frac{\\beta_{ij}}{A_{ij}}-(1-\\beta_{ij})\\log(1-\\beta_{ij})\\right]$. The paper proves the universal comparison $\\operatorname{Bethe}(A)\\leq\\operatorname{per}(A)\\leq2^{n/2}\\operatorname{Bethe}(A)$ and, if the bipartite support graph has even girth at least $g\\ge4$, the sharper bound $\\operatorname{per}(A)\\leq2^{2n/g}\\operatorname{Bethe}(A)$. Thus the log partition-function error is at most $(2n/g)\\log2$. Here $n$ is the number of left and right vertices, $g$ is the length of the shortest cycle in the support graph, and $A_{ij}$ is the positive edge weight produced by the neural logits.",
 "math_tags": [
  "combinatorics",
  "graph-theory",
  "optimization",
  "probability"
 ],
 "ml_areas": [
  "moe-routing",
  "attention",
  "graph-nn"
 ],
 "paper": {
  "arxiv_id": "2609.02017",
  "arxiv_url": "https://arxiv.org/abs/2609.02017",
  "summary_what_math_gives_to_ml": "The paper gives a sharp, structural error guarantee for replacing the permanent of a nonnegative matrix by its polynomial-time Bethe approximation: the multiplicative gap improves from $2^{n/2}$ to $2^{2n/g}$ when the bipartite support graph has girth at least $g$. This suggests a matching-aware neural normalization or routing layer whose partition function is approximated by Bethe inference, with a known worst-case log-error controlled by the sparsity pattern rather than only its size. The most promising transfer is to design fixed high-girth sparse assignment graphs for MoE routing, set matching, or permutation-equivariant attention, and use the Bethe free-energy optimizer as a differentiable surrogate for the intractable permanent.",
  "title": "Optimal girth-dependent bounds for the Bethe approximation of the permanent",
  "year": "2026"
 },
 "ratings": {
  "difficulty": 6,
  "novelty": 6,
  "usefulness": 6
 },
 "solves": [
  "scalability",
  "stability",
  "accuracy"
 ],
 "title": "High-girth Bethe assignment layer",
 "url": "https://synthcore.org/idea/3034/high-girth-bethe-assignment-layer",
 "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
   }
  }
 }
}