{
 "artifacts": null,
 "category": "memory",
 "description": "Replace MSE-calibrated scalar quantization with a conservative left-edge quantizer whose thresholds are denser where activation probability and task utility slope are both high. For a monotone utility function, this should preserve high-impact activation regions better than uniform or MSE-optimal bins at the same number of codes, while retaining an explicit rate-versus-quality design rule.",
 "formulas_latex": [
  "$$i_B(B)=j\\;\\Longleftrightarrow\\;R_j\\leq B\u003cR_{j+1},\\qquad R_0=0,\\quad R_{n+1}=\\infty$$",
  "$$Q(R)\\in[0,1],\\qquad Q(0)=0,\\qquad Q(\\infty)=1,\\qquad Q'(R)\u003e0$$",
  "$$n\\,\\mathcal{G}_n^{*}\\to C_w=\\frac{1}{2}\\left(\\int\\sqrt{p(x)Q'(x)}\\,dx\\right)^2$$",
  "$$\\hat{x}=R_j\\ \\text{for }x\\in[R_j,R_{j+1}),\\qquad \\int_0^{R_j}\\sqrt{p(x)Q'(x)}\\,dx\\approx \\frac{j}{n}\\int_0^{\\infty}\\sqrt{p(x)Q'(x)}\\,dx$$"
 ],
 "id": 3148,
 "implementation": "Integrate this at the scalar activation-quantization point, initially for nonnegative post-ReLU activations or absolute values of weights in one Transformer MLP block. During calibration, collect a histogram \\(\\{(x_k,c_k)\\}\\) of values in the chosen tensor, normalize it to an empirical density \\(p_k=c_k/\\sum_l c_l\\), and estimate a monotone utility curve \\(Q\\). The cheapest first utility is \\(Q(x)=\\log(1+x/s)/\\log(1+x_{\\max}/s)\\), where \\(s\\) is the tensor median; a stronger version estimates \\(Q'(x)\\) by measuring validation-loss increase after selectively attenuating values in each histogram bin. Smooth the nonnegative weights \\(g_k=\\sqrt{p_kQ'_k+\\epsilon}\\), compute \\(G_k=\\sum_{l\\le k}g_l\\), and choose \\(n-1\\) internal thresholds at cumulative targets \\(jG_K/n\\). At inference, locate the largest threshold \\(R_j\\le x\\) and emit the left edge \\(R_j\\), using integer code indices and a per-tensor scale/table. Pseudocode: `hist=calibrate(x); p=normalize(hist); Qprime=utility_slope(hist); g=sqrt(p*Qprime+eps); G=cumsum(g); thresholds=invert_cdf(G,[1/n,...,(n-1)/n]); codes=searchsorted(thresholds,x)-1; xhat=thresholds[codes]`. The paper-derived quantities are the \\(\\sqrt{pQ'}\\) allocation and \\(C_w=0.5(\\sum_k\\sqrt{p_kQ'_k}\\Delta x)^2\\); density and utility slope are estimated empirically. First test on a small 6-layer Transformer language model with INT4 activation quantization on WikiText-2, comparing uniform, percentile, and MSE Lloyd-Max quantizers at equal 4-bit memory and kernel cost. Measure validation perplexity, calibration loss, saturation rate, and throughput. Success is lower perplexity at the same bitwidth, or equal perplexity with fewer codes; also verify the predicted diminishing-return curve \\(\\mathcal{G}_n\\approx C_w/n\\) for \\(n\\in\\{4,8,16,32\\}\\).",
 "math_summary": "The paper models a monotone quality curve \\(Q(R)\\in[0,1]\\), with \\(Q(0)=0\\), \\(Q(\\infty)=1\\), and strictly increasing \\(C^1\\) behavior. A threshold ladder \\(0=R_0\u003cR_1\u003c\\cdots\u003cR_n\u003cR_{n+1}=\\infty\\) uses the left-edge reproduction rule \\(i_B(B)=j\\Longleftrightarrow R_j\\le B\u003cR_{j+1}\\), so an input \\(B\\) in cell \\([R_j,R_{j+1})\\) is represented by \\(R_j\\). Let \\(p(x)\\) be the empirical density of a nonnegative activation or weight magnitude \\(x\\), and let \\(Q(x)\\) be a monotone task-utility curve: the ideal utility at value \\(x\\), while the quantized utility is \\(Q(R_j)\\). The asymptotic quality-gap theorem states \\(\\mathcal{G}_n^*=\\Theta(1/n)\\) and \\(n\\mathcal{G}_n^*\\to C_w\\), where \\(C_w=\\frac{1}{2}(\\int\\sqrt{p(x)Q'(x)}\\,dx)^2\\). The optimal threshold density is proportional to \\(\\sqrt{p(x)Q'(x)}\\), unlike the classical squared-error rule proportional to \\(p(x)^{1/3}\\). For a finite codebook, define \\(g(x)=\\sqrt{p(x)Q'(x)}\\), \\(Z=\\int g(x)dx\\), and place thresholds so that \\(\\int_0^{R_j}g(x)dx\\approx jZ/n\\). The left-edge rule is conservative: \\(\\hat{x}=R_j\\le x\\), avoiding upward rounding when overestimation is harmful.",
 "math_tags": [
  "probability",
  "approximation-theory",
  "measure-theory",
  "optimization"
 ],
 "ml_areas": [
  "quantization",
  "transformer",
  "inference-speedup"
 ],
 "paper": {
  "arxiv_id": "2609.03745",
  "arxiv_url": "https://arxiv.org/abs/2609.03745",
  "summary_what_math_gives_to_ml": "The paper recasts a one-sided, threshold-based representation ladder as a nonstandard scalar quantizer: each cell reconstructs at its left edge, so the local error is first-order rather than squared-error. Its asymptotically optimal allocation places representation thresholds with density proportional to \\(\\sqrt{p(x)Q'(x)}\\), producing an \\(O(1/n)\\) quality gap and an explicit constant. This transfers naturally to activation or weight quantization when downstream utility is monotone but reconstruction must be conservative, such as low-bit inference or thresholded routing. A practical implementation is a calibration-time quantizer whose codebook is obtained from the empirical activation density and a task-derived utility slope, rather than from standard MSE or uniform spacing.",
  "title": "A Quantization Problem Posed by Adaptive Streaming",
  "year": "2026"
 },
 "ratings": {
  "difficulty": 4,
  "novelty": 7,
  "usefulness": 7
 },
 "solves": [
  "memory",
  "accuracy",
  "speedup"
 ],
 "title": "Utility-Weighted Left-Edge Quantization",
 "url": "https://synthcore.org/idea/3148/utility-weighted-left-edge-quantization",
 "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
   }
  }
 }
}