{
 "artifacts": null,
 "category": "regularization",
 "description": "Use the determinant and trace-power identities of the rules matrix as a spectral diagnostic for recurrent or state-space training. Penalize unstable or excessively resonant modes through a truncated log-zeta objective, while retaining selected eigenvalues near the unit circle when long memory is desired. This gives a falsifiable transition criterion based on closed-walk growth rather than only gradient norms.",
 "formulas_latex": [
  "$$\\zeta_R(z)=\\frac{1}{\\det(I-zR)}=\\prod_j(1-z\\lambda_j)^{-1},\\qquad \\log\\zeta_R(z)=\\sum_{n=1}^{\\infty}\\frac{z^n\\operatorname{tr}(R^n)}{n}.$$",
  "$$L_{\\mathrm{zeta}}(\\theta)=-\\sum_{n=1}^{K}\\frac{\\alpha^n}{n}\\operatorname{tr}(R(\\theta)^n),\\qquad 0\u003c\\alpha\u003c\\frac{1}{\\rho(R(\\theta))}.$$",
  "$$\\widehat{\\operatorname{tr}(R^n)}=\\frac{1}{J}\\sum_{j=1}^{J}v_j^{\\top}R^nv_j,\\qquad \\mathbb{E}[v_jv_j^{\\top}]=I.$$",
  "$$\\rho(R)\u003c1\\ \\Rightarrow\\ \\|R^n\\|\\leq C(\\rho(R)+\\varepsilon)^n\\ \\text{for sufficiently large }n.$$"
 ],
 "id": 3006,
 "implementation": "Integration point: apply this monitor to the recurrent Jacobian or transition matrix at every optimizer step. For an explicit linear state-space layer, use its state matrix A directly. For an RNN, estimate a local Jacobian R=partial h_{k+1}/partial h_k on a minibatch and several sequence positions; this is not the full parameter Hessian. Choose K=4 to 8 and J=2 to 8 Rademacher vectors. For each vector, set u_{j,0}=v_j and recursively compute u_{j,n}=Ru_{j,n-1}; estimate each trace by J^(-1) sum_j v_j^T u_{j,n}. Add lambda L_zeta to the task loss, and separately log power growth and a few Lanczos eigenvalue estimates. If the estimated spectral radius exceeds a target r_*\u003c1, reduce recurrent gain or apply spectral normalization. If memory is required, constrain only unstable modes while allowing selected modes near the unit circle. The determinant/zeta identity and trace expansion are imported from the paper; Hutchinson estimation, Jacobian selection, and the penalty are adaptations. First experiment: compare GRU, orthogonal RNN, and zeta-regularized RNN on Copy Memory and sequential-MNIST with equal parameter counts. Sweep recurrent gain across its instability boundary. The predicted signature is a sharp increase in G(alpha) as alpha rho(R) approaches one from below, and exploding long-horizon gradients when measured rho(R)\u003e1. The estimated boundary should agree with power iteration within 10-20 percent.",
 "math_summary": "Let R(theta) be a recurrent transition matrix, reduced Jacobian, or learned finite-state transfer matrix parameterized by neural parameters theta. The paper identifies the symbolic-dynamics zeta function with zeta_R(z)=det(I-zR)^(-1). Its power-series coefficients are tr(R^n), aggregating all length-n closed walks, and its singularities occur at z=1/lambda_j for eigenvalues lambda_j of R. For a contraction radius alpha satisfying 0\u003calpha\u003c1/rho(R), use the truncated objective L_zeta=-sum_{n=1}^K alpha^n tr(R^n)/n, or monitor G(alpha)=sum_{n=1}^K alpha^n|tr(R^n)|/n. Hutchinson random vectors estimate traces without explicitly forming dense powers. The relevant stability boundary is rho(R)=1, and the truncated zeta should increase sharply as alpha rho(R) approaches one from below.",
 "math_tags": [
  "dynamical-systems",
  "spectral-theory",
  "linear-algebra",
  "combinatorics",
  "optimization"
 ],
 "ml_areas": [
  "rnn",
  "ssm",
  "optimizer",
  "regularization"
 ],
 "paper": {
  "arxiv_id": "2609.01399",
  "arxiv_url": "https://arxiv.org/abs/2609.01399",
  "summary_what_math_gives_to_ml": "The paper provides a constructive symbolic-dynamics mechanism: a quadratic rotation-symmetric Boolean rule induces a signed binary de Bruijn transfer matrix, and Fourier transformation in an auxiliary parity coordinate block-diagonalizes the finite-type shift into unsigned and signed components. Its dynamical zeta function is exactly the reciprocal characteristic determinant, so periodic-orbit counts are encoded by traces of transfer-matrix powers and the eigenvalues determine recurrence behavior. This can transfer to neural sequence models as a finite-memory parity/automaton module and as a spectral-zeta monitor or regularizer for recurrent dynamics. The transfer is most credible for RNNs, state-space models, and discrete latent sequence architectures rather than generic feedforward networks.",
  "title": "The Quadratic Easy Coefficients Conjecture via Finite-Type Shifts and Zeta Functions",
  "year": "2026"
 },
 "ratings": {
  "difficulty": 5,
  "novelty": 6,
  "usefulness": 6
 },
 "solves": [
  "stability",
  "generalization",
  "accuracy"
 ],
 "title": "Zeta spectral monitor for recurrent training",
 "url": "https://synthcore.org/idea/3006/zeta-spectral-monitor-for-recurrent-training",
 "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
   }
  }
 }
}