{
 "artifacts": null,
 "category": "architecture",
 "description": "Translate a recursive neural architecture grammar into a combinatorial specification and use its generating function to sample architectures uniformly conditional on size. This removes the severe shape and branching biases of naive random tree generation, producing a better-covered architecture-search population at the same evaluation budget.",
 "formulas_latex": [
  "$$\\mathcal{C}_n=\\{c\\in\\mathcal{C}\\mid |c|=n\\},\\qquad c_n=|\\mathcal{C}_n|$$",
  "$$C(z)=\\sum_{c\\in\\mathcal{C}}z^{|c|}=\\sum_{n=0}^{\\infty}c_nz^n,\\qquad [z^n]C(z)=c_n$$",
  "$$\\mathcal{A}=\\mathsf{Z}+\\mathsf{Z}\\times\\mathcal{A}+\\mathsf{Z}\\times\\mathcal{A}\\times\\mathcal{A}+\\mathsf{Z}\\times\\mathrm{CYC}_3(\\mathcal{A})$$",
  "$$\\mathrm{CYC}_3(A)(z)=\\frac{A(z)^3+2A(z^3)}{3}$$"
 ],
 "id": 3094,
 "implementation": "Integrate this at the neural architecture-search population generator, before model training. Define an architecture grammar with leaf L, unary node U(X) such as a residual or graph-convolution block, binary node B(X,Y) such as concatenation followed by projection, and a three-way invariant node T(X,Y,Z) such as sum aggregation. Let size count trainable blocks, and let a[n] be the number of canonical architecture trees of size n. Compute all counts up to budget N by dynamic programming. For each constructor r with child sizes n1,...,nk, compute its completion count w_r as the product of the child counts. For a requested total size n, enumerate feasible constructor and child-size tuples and choose each tuple with probability proportional to its number of completions. Recurse until leaves, then assign operation labels from a finite set and canonicalize commutative children by sorting their serialized hashes. For the symmetric three-way merge, enumerate unordered triples directly or use the cycle-index correction (A(z)^3+2A(z^3))/3. Cache sampled subtrees and reject only tensor-shape-invalid combinations, or include shape and type as additional DP states. The exact mathematical components are the coefficient recursion, symmetry quotient, and completion-weighted sampling; operation validity and downstream accuracy are empirical. First compare 1,000 architectures on CIFAR-10 using a small GraphNAS-style GNN or eight-block image model against uniform random mutation and ordered random trees at equal training FLOPs. Measure distinct canonical structures, validation accuracy after a fixed budget, and best accuracy versus evaluated architectures. Success is higher best and median accuracy with lower variance at small search budgets.",
 "math_summary": "The paper defines a combinatorial class \\(\\mathcal{C}\\), its size-\\(n\\) subset \\(\\mathcal{C}_n=\\{c\\in\\mathcal{C}:|c|=n\\}\\), and the generating function \\(C(z)=\\sum_{c\\in\\mathcal{C}}z^{|c|}=\\sum_{n\\ge0}c_nz^n\\), where \\(c_n=|\\mathcal{C}_n|\\) and \\([z^n]C(z)=c_n\\). Its acyclic-structure specification has the recursive form \\(\\mathcal{A}=\\mathsf{Z}+\\mathsf{Z}\\times\\mathcal{A}+\\mathsf{Z}\\times\\mathcal{A}\\times\\mathcal{A}+\\mathsf{Z}\\times\\mathrm{CYC}_3(\\mathcal{A})\\), where \\(\\mathsf{Z}\\) contributes one node and \\(\\mathrm{CYC}_3\\) identifies ordered triples under cyclic permutation. Adapt this as a grammar for rooted neural computation graphs: a node is either a leaf, a unary composition, a binary merge, or a three-branch permutation-invariant merge. Dynamic programming computes \\(a_n\\), the number of valid architectures of size \\(n\\), and recursive completion-weighted decomposition samples a structure uniformly within a fixed size. For an unlabeled 3-cycle constructor, use the cycle-index generating function \\(\\mathrm{CYC}_3(A)(z)=(A(z)^3+2A(z^3))/3\\).",
 "math_tags": [
  "combinatorics",
  "graph-theory",
  "group-theory",
  "generating-functions"
 ],
 "ml_areas": [
  "graph-nn",
  "cnn"
 ],
 "paper": {
  "arxiv_id": "2609.02932",
  "arxiv_url": "https://arxiv.org/abs/2609.02932",
  "summary_what_math_gives_to_ml": "The paper's transferable asset is its symbolic-method view of recursively defined structured objects: a compact grammar is translated into generating functions whose coefficients count objects exactly, including quotienting by symmetries. For neural networks, this can become a principled generator of tree-, DAG-, or module-shaped architectures and training curricula with controlled size and structural diversity, rather than biased random mutation or ad hoc sampling. Multivariate generating functions also provide exact control over auxiliary architectural properties such as branching count, number of special modules, or symmetry markers. The payoff is most plausible for neural architecture search and synthetic graph-model generation, not for ordinary dense Transformers.",
  "title": "Enumerating Chemical Structures, Stereoisomers, and Stereogenic Units with Analytic Combinatorics",
  "year": "2026"
 },
 "ratings": {
  "difficulty": 5,
  "novelty": 7,
  "usefulness": 6
 },
 "solves": [
  "sample-efficiency",
  "scalability",
  "accuracy"
 ],
 "title": "Uniform Species Architecture Sampler",
 "url": "https://synthcore.org/idea/3094/uniform-species-architecture-sampler",
 "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
   }
  }
 }
}