Real-valued continued fraction of straight lines

Vijay Prakash S

REVIEW 5 major objections 5 minor 1 cited by

Real-valued continued fraction of straight lines

T0 review · 5 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that using the real root of $ay^3 + y = m w^T x$ as the logistic-regression score, computed by a real-valued continued fraction, yields more stable, faster-converging, and more accurate image classifiers on Fashion-MNIST…

desk verdict A correct but classical cubic-root score function, presented with an unfair linear baseline that sinks the only empirical claim. read the letter →

arxiv 2412.16191 v1 pith:FIOHF5NL submitted 2024-12-16 cs.LG

Vijay Prakash S This is my paper · ORCID

classification cs.LG

keywords continuedfractionboundednonlinearitylogisticregressionFashion-MNISTgradientdescentcubicequationparameterestimation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a simple bounded nonlinearity, obtained by replacing a straight-line score with the real root of a cubic equation and expressing that root as a real-valued continued fraction, improves logistic regression for image classification. The proposed score $y$ solves $ay^3 + y = m w^T x$, where $a>0$ and $m$ are learned parameters, and the continued fraction is shown to converge to this real root. On Fashion-MNIST, the paper reports that parameters estimated through the $y$-score have less variance, converge faster and more smoothly, and give higher test accuracy than the paper's linear logistic regression baseline. The motivation is that straight lines with large slope grow faster than the independent variable, while the bounded $y$-curves grow more slowly and therefore produce more stable estimates.

What carries the argument

The load-bearing object is the real root of the cubic equation $ay^3 + y = m w^T x$, expressed as $y = -\frac{1}{3}t^{1/3} + \frac{1}{a}t^{-1/3}$ with $t = -\frac{27mw^Tx}{2a} + \sqrt{\left(\frac{27mw^Tx}{2a}\right)^2 + \frac{27}{a^3}}$. This root is also produced by the continued fraction $y = \frac{m w^T x}{1 + ay^2}$, whose truncated forms all remain bounded by the linear line $y = m w^T x$. The continued-fraction representation is what makes the update rules special: the derivative denominators $1 + 3ay^2$ automatically shrink the gradient steps as $y$ grows, so the optimization is self-bounding. The two extra parameters $a$ and $m$ carry the nonlinearity and the overall scale, and once $a$ converges the other parameters converge.

What would settle it

Run a tuned linear logistic regression on Fashion-MNIST with a chosen learning rate, mini-batch schedule, and optional L2 regularization, and compare test accuracy and weight variance with the $y$-form. If the tuned linear model matches or exceeds the $y$-form accuracy, the central empirical claim would fail.

Watch

Extended reading notes

Core claim

The central discovery is that the standard linear score $w^T x$ of logistic regression can be replaced by the real solution $y$ of $ay^3 + y = m w^T x$ without introducing a step-size hyperparameter for $w$ and $m$. The paper derives this solution through a continued fraction of straight lines, $y = mx/(1 + ay^2)$, and reduces the continued fraction to an infinite series by the standard transformation for continued fractions. Because $1 + 3ay^2$ appears in the denominators of the derivatives, gradient updates for $w$ and $m$ remain bounded and adaptive; only the positivity constraint on $a$ requires a step size. Empirically, after 50 mini-batch iterations on Fashion-MNIST, the $y$-form reaches higher test accuracy and lower loss than the linear form, while the learned weights stay in a small range rather than growing into the hundreds or thousands.

Load-bearing premise

The load-bearing premise is that the linear baseline was optimized fairly; in the paper it is obtained by setting $m=1$ and $a=0$ inside the nonlinear update, which yields an untuned raw gradient step with unit step size and no regularization.

Editorial extensions

If this is right

On Fashion-MNIST, the $y$-form score should reach higher or comparable test accuracy with smaller weight magnitudes than a raw linear logistic regression run with the same schedule.
Because $w$ and $m$ updates need no learning rate, the method reduces the number of tuning hyperparameters to one, namely the step size for $a$.
The learned $(a, m)$ pairs are stable per class, so a multi-dimensional classifier can be summarized by ten curves on the $xy$-plane or on the $(i, ii)$-plane.
The smooth, near-monotonic convergence of the weights is a direct consequence of the bounded gradient denominators and does not require early stopping or regularization.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The comparison that would most directly test the paper's claim is against a tuned linear logistic regression with an optimized learning rate and L2 regularization; the paper's baseline is a single raw gradient substitution, so the reported accuracy gap may shrink under a fair baseline.
The same bounded score could be plugged into other linear classifiers, such as softmax regression or linear support vector machines, since the construction only changes the scalar score before the loss.
The $(i, ii)$-plane view suggests a diagnostic: if categories form separated clusters for particular initial conditions, that could indicate sensitivity of the learned model to initialization.
Because $a \to 0$ recovers the linear case, the parameter $a$ can be read as a measure of how much nonlinear bounding the data require; classes needing larger $a$ are more nonlinear.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The short version: this paper takes a classical cubic equation, solves it with Cardano's formula, rewrites the solution as a continued fraction, and then uses the real root as a bounded score inside logistic regression for Fashion-MNIST. The forward and backward equations are internally consistent, and the implicit differentiation is right. But the paper's central claim—that this y-form is more accurate than the linear counterpart—rests on a baseline that is optimized with a fixed unit step size and no tuning, while the y-form gets an implicit adaptive step from the 1/(1+3ay^2) factor. That is not an apples-to-apples comparison.

What is actually new: applying this specific cubic nonlinearity as a trainable score in logistic regression on one dataset. That is a narrow extension of learnable activations, not a new theory. The continued fraction rewriting is a known technique, and the paper uses the Cardano root for training anyway; the series is only motivational.

What the paper does well: the derivations in Section 2 are careful and the plots of the curves are clear. The implicit gradient computations in Eq. (11) are correct. The observation that the factor 1/(1+3ay^2) gives bounded, adaptive step sizes is mildly interesting, though it is not a substitute for a fair baseline.

The soft spots are substantial. First, the linear comparator in Step 6e is a raw unit-step-size mini-batch update with no learning rate schedule, no regularization, and no tuning. Standard logistic regression on Fashion-MNIST, properly optimized, should be at least as accurate as the reported y-form accuracy. Second, there are no error bars, no multiple seeds, and no code, so we cannot assess run-to-run variance or reproducibility. Third, the 'variance' claim is about the magnitude of learned weights, not statistical variance across runs, and is largely a consequence of the m scaling. Fourth, the convergence of the continued fraction is asserted without conditions, and the initial step size alpha is not reported. These are not minor quibbles; the first one directly undermines the paper's headline result.

Who is this for? Someone interested in bounded score functions might find the idea worth a glance, but the evaluation as it stands does not establish an advantage. A serious referee would need a properly tuned baseline, repeated runs, and code.

My recommendation: this deserves a desk reject in its current form. The mathematical core is sound but classical, and the empirical claim is not supported by the evidence presented. If the author returns with a fair comparison and error bars, it might be worth another look.

Referee Report

5 major / 5 minor

Summary. The paper introduces a parametric nonlinear transformation of a linear regression score, defined by the cubic equation (1+ay^2)y = mx, whose real root is expressed as a continued fraction of straight lines and then as an infinite series following Euler's method. The authors apply this transformation to Fashion-MNIST image classification by replacing the linear predictor w^Tx in logistic regression with the real root y(w^Tx) of the cubic equation, estimating per-class parameters w, a, and m by mini-batch gradient descent with implicit gradients. The central claim is that the continued-fraction-based model yields parameters with lower variance, faster convergence, and higher test accuracy than the plain linear logistic regression baseline.

Significance. If the empirical claims were established, the work would demonstrate that a simple bounded nonlinear transformation of the linear predictor can improve optimization stability and classification accuracy over ordinary logistic regression. The closed-form expression for the cubic root and the implicit-gradient derivation are internally consistent and could be useful for other models. However, the central empirical claim is not currently supported: the linear baseline is given an untuned, unit-step-size update, no error bars or multiple seeds are reported, and the convergence of the continued fraction is asserted without conditions. The theoretical contribution is also limited because the continued fraction is an equivalent rewriting of the cubic equation rather than a new mathematical object. The manuscript contains no code or data availability statement, so the results are not reproducible as presented.

major comments (5)

[Section 3, Step 6e, Eq. (14)] The linear baseline is obtained by substituting m=1 and a=0 into the nonlinear update of Eq. (14), which yields the raw mini-batch update wi <- wi - (1/n) sum_j (sigma_j - p*_j) x_ij, i.e., gradient descent with a fixed step size of 1 and no learning-rate schedule, no regularization, no momentum, and no hyperparameter tuning. In contrast, the nonlinear model's update includes the factor 1/(1+3ay^2), which acts as a state-dependent damping that effectively reduces the step size as y grows. The accuracy gap reported in Figs. 7(c) and 8(c) therefore conflates the model change with a difference in optimization protocols. The authors must compare against a properly tuned linear logistic regression (for example, one with a chosen learning rate, regularization, or a standard solver) under identical training epochs and initialization conditions.
[Section 3, Figs. 7 and 8] All reported accuracy and loss curves are single trajectories: no error bars, no multiple random seeds, and no statistical significance test are provided. The Abstract's claim that the continued-fraction model is 'more accurate than the linear counterpart' is a point estimate from one run for each of two initial conditions. The authors should report means and standard deviations over at least several independent initializations, and ideally a paired test across the same initial weights, before the superiority claim can be evaluated.
[Section 2, Eq. (4)] The convergence of the continued fraction to the real root of Eq. (1) is asserted without any conditions or proof. The derived infinite series in Eq. (4) is not well formed (the third term has unbalanced parentheses), and the convergence properties are not analyzed. For large values of the reduced parameter ahat = a m^2 x^2, a naive fixed-point iteration of the truncations can oscillate rather than converge, so a precise convergence theorem with a stated domain of validity is required. If the continued-fraction representation is only motivational, the claim should be removed and the paper reframed around the closed-form root in Eq. (3), which is correct.
[Section 4.1 and Fig. 9] The 'less variance' claim is based on comparing the raw magnitudes of the learned weight vectors (thousands for the linear model versus about 2 for the nonlinear model). This is not a statistical variance over repeated runs, and it is not scale-invariant: because the nonlinear model has an additional per-class scale parameter m, the magnitude of w is not directly comparable to the magnitude of the linear model's w. The authors should instead report repeated-run variances of test accuracy, of predicted probabilities, or of the resulting decision boundaries.
[Section 2, Property 2 and Section 4.2] The statement 'Once a converges, the other parameters converge' is used as a justification for the optimization procedure, but no argument or theorem is provided. The plotted trajectories of a and m in Figs. 7(a)-(b) and 8(a)-(b) show that a and m settle to nearly constant values in the particular runs, but this does not imply that the full gradient-descent dynamics converge, nor that convergence of a is a general condition for convergence of w. This should be presented as an empirical observation, not as a property of the model.

minor comments (5)

[Section 2, around Eq. (4)] There is a typo 'infnite' in the sentence preceding Eq. (4), and the continued-fraction notation with nested exponents is difficult to parse; please rewrite the derivation more clearly.
[Section 3, Eq. (9)] The loss function in Eq. (9) is written for a scalar output p*_j, but p*_j is a one-hot vector and the model is a multi-class classifier. Please specify how the loss is summed over classes (e.g., one-vs-rest with sigmoid, as implied by the text, or a softmax variant).
[Section 4.3 and Fig. 10] The xy-plane plots in Fig. 10 are simply the learned curves for each class; calling this a representation of a multi-dimensional problem on a plane is an overstatement, since the plot shows the class-specific scalar function y(x) after training, not the full decision boundary. Consider rewording.
[References] Reference [3] (Ruder's overview of gradient descent) is a survey, not a primary source for the specific mini-batch updates; please cite a more standard optimization text. Reference [1] is a translation of Euler's work; please provide the original publication details or a more accessible citation.
[Figures] Several figure captions contain garbled or duplicated text (e.g., Fig. 1 and Fig. 9), making it hard to interpret the markers and shaded regions. Please revise all captions for clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the continued fraction is an equivalent rewriting of the defining cubic equation, and all empirical claims are evaluated on held-out Fashion-MNIST data rather than derived from fitted quantities.

full rationale

The paper's only derivational step is rewriting Eq. (1), (1 + ay^2)y = mx, as y = mx/(1 + ay^2), and then recursively substituting y into the right-hand side to obtain a continued fraction and, via Euler's method, the infinite series in Eq. (4). This is an algebraic representation identity for the real root of the same cubic; it does not fit any parameter and then relabel that fit as a prediction. In the classification section, the parameters w, a, and m are estimated from training batches using gradient-descent updates in Eqs. (12)-(14), and accuracy is measured on the Fashion-MNIST test set, which is external to the fitted values. The linear case sigma(w^T x) is correctly identified as the limiting case m = 1 and a -> 0, and the comparison is empirical rather than tautological. The fact that Step 6e gives the linear baseline a unit-step-size, untuned mini-batch update is a possible fairness or optimization concern, but it is not circularity: the claimed advantage of the nonlinear model is a testable empirical claim and could in principle fail against a properly tuned baseline. The references are Euler's classical paper, a machine-learning textbook, a gradient-descent survey, and the dataset paper; no load-bearing self-citation is present. Since no fitted quantity is renamed as a prediction and the derivation is self-contained with respect to the external benchmark, the circularity score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical or mathematical entities are postulated. The free parameters a, m, and w are all fitted to the dataset, and the continued fraction representation is an equivalent rewriting of the defining cubic rather than an independent object.

free parameters (5)

a (per-class saturation parameter) = values in Table 1, e.g., 0.0093 to 0.2626
Learned by gradient descent; controls the strength of the cubic bounding term and is central to the model.
m (per-class scale parameter) = values in Table 1, e.g., -5.77 to 4.67
Learned by gradient descent; scales the linear score before bounding.
w (per-class regression weights) = 10 x 785 matrix, not fully listed; ranges plotted in Fig. 9
Standard logistic-regression weights, learned for both the nonlinear and linear forms.
alpha (step size for a) = unreported
Introduced in Eqn (10) to enforce a > 0; initial value and schedule are never stated, so replication requires guessing.
a(0) initial value = 1 or 5
Chosen by hand for each experiment; results depend on this choice (Table 1).

assumptions (4)

standard math Cardano's formula gives the unique real root of the cubic for all a > 0
Used in Eqns (3) and (8); standard but the text does not discuss branch selection when t <= 0.
ad hoc to paper The continued fraction iteration converges to the real root for the parameter ranges considered
Section 2 and Fig. 5 assert convergence without a proof or explicit conditions; the map f(y)=c/(1+a y^2) is not globally convergent.
ad hoc to paper Gradient descent with the stated unit-step-size updates reaches a good minimum for the nonlinear model
Section 3, Eqns (10)-(14); no convergence analysis or learning-rate tuning is provided for w and m.
domain assumption Sigmoid output with a cubic score is a valid probabilistic classifier
Standard logistic-regression assumption carried over to the nonlinear score; no calibration analysis is given.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Real-valued continued fraction of straight lines." pith.science (2026). https://pith.science/paper/FIOHF5NL

@misc{pith2026241216191,
  author       = {Pith},
  title        = {Pith review of: Real-valued continued fraction of straight lines},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FIOHF5NL}},
  note         = {Machine review of arXiv:2412.16191}
}

abstract

In an unbounded plane, straight lines are used extensively for mathematical analysis. They are tools of convenience. However, those with high slope values become unbounded at a faster rate than the independent variable. So, straight lines, in this work, are made to be bounded by introducing a parametric nonlinear term that is positive. The straight lines are transformed into bounded nonlinear curves that become unbounded at a much slower rate than the independent variable. This transforming equation can be expressed as a continued fraction of straight lines. The continued fraction is real-valued and converges to the solutions of the transforming equation. Following Euler's method, the continued fraction has been reduced into an infinite series. The usefulness of the bounding nature of continued fraction is demonstrated by solving the problem of image classification. Parameters estimated on the Fashion-MNIST dataset of greyscale images using continued fraction of regression lines have less variance, converge quickly and are more accurate than the linear counterpart. Moreover, this multi-dimensional parametric estimation problem can be expressed on $xy-$ plane using the parameters of the continued fraction and patterns emerge on planar plots.

Figures

Figures reproduced from arXiv: 2412.16191 by the authors.

**Figure 1.** y for different a and m values. As a → ∞ we approach the x− axis. 0° 45° 90° 135° 180° 225° 270° 315° 0.5 1.0 1.5 2.0 2.5 3.0 (a) a ∈ [1e − 1, 1] and m ∈ interval [-200,200]. 20 15 10 5 0 5 10 15 20 m 1.5 1.0 0.5 0.0 0.5 1.0 1.5 tan = m/(1 + sin 2 ) tan = m (b) Reduced phase when compared to the linear case [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 2.** Polar plots of Eqn. (1) with r = 1/ √ a 2 Real solution and its properties Eqn. (1) has two complex roots and a real root for all a > 0. The real solution is given by y = − 1 3 t 1 3 + 1 a t − 1 3 , where t = − 27mx 2a + s 27mx 2a 2 + 27 a 3 . (3) The solution of y is sum of two components i: − 1 3 t 1 3 and ii: 1 a t − 1 3 . The plots of these two components are shown in Fig. (3) for various values of a and m… view at source ↗

**Figure 3.** Plots of individual components i and ii of [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Figures from the paper (7 more)

**Figure 4.** Figure 4: (a) shows i and ii for a fixed a. (b) With i and ii as axes, we get unique curves for each a ∈ [1e − 3, 5e − 3, 1e − 2, 0.1, 1] and for various values of m. 2. From Fig. (4b) with components i and ii as axes, it can be seen that we only have to determine a in order to …

**Figure 5.** Figure 5: Continued fraction series converging to y = 1/at1/3 − t 1/3/3 or y = i + ii values. All the curves including the truncated Eqn. (4) are bounded by y = x line. a = 1, m = 1 being A/A. Further we have, B B = 1 1 + ˆa 1 (1+ˆa) 2 = (1 + ˆa) 2 (1 + ˆa) 2 + a = A2 A2 + ˆa , …

**Figure 6.** Figure 6: (a) One of the sample images of Fashion-MNIST dataset, (b) A section of the sample image with pixel [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

**Figure 7.** Figure 7: Estimation of parameters of continued fraction [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

**Figure 8.** Figure 8: Estimation of parameters of continued fraction [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]

**Figure 9.** Figure 9: w ′ i s for 0th, 1st and 2nd categories of both σ(w T x) and σ(y), with a (0) = 0, a (0) = 5 are shown. Boundedness: The range of w (50) i (w ′ i s after 50 iterations) are shown above. (a), (d) and (g) have more variance. Convergence: The dark dots (‘ ’) are w (50) i …

**Figure 10.** Figure 10: For various initial conditions, the final values of [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Two-parameter superposable S-curves
stat.ME 2025-04 reject novelty 3.0 of 10

An algebraic S-curve family is fitted to iris CDFs and proposed as a universal representation of non-uniform patterns, but only visual evidence is provided and the derivation contains inconsistencies.

Reference graph

Works this paper leans on

4 extracted references · 2 canonical work pages · cited by 1 Pith paper

[1]

(2004) On the Transformation of Infinite Series to Continued Fractions (D

Euler, L. (2004) On the Transformation of Infinite Series to Continued Fractions (D. W. File, Trans.). Read- ing Classics: Euler. https://people.math.osu.edu/sinnott.1/ReadingClassics/continuedfractions.pdf (Original work published 1785)

work page 2004
[2]

Abu-Mostafa, Y.S., Magdon-Ismail, M., and Lin, H.T. (2012). Learning from data (Vol. 4, p. 4). New York: AMLBook

work page 2012
[3]

Ruder, S. (2017). An overview of gradient descent optimization algorithms. arXiv preprint arXiv:1609.04747

arXiv 2017
[4]

Xiao, H., Rasul, K., Vollgraf, R. (2017). Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747. 12

arXiv 2017

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

(2004) On the Transformation of Infinite Series to Continued Fractions (D

Euler, L. (2004) On the Transformation of Infinite Series to Continued Fractions (D. W. File, Trans.). Read- ing Classics: Euler. https://people.math.osu.edu/sinnott.1/ReadingClassics/continuedfractions.pdf (Original work published 1785)

work page 2004

[2] [2]

Abu-Mostafa, Y.S., Magdon-Ismail, M., and Lin, H.T. (2012). Learning from data (Vol. 4, p. 4). New York: AMLBook

work page 2012

[3] [3]

Ruder, S. (2017). An overview of gradient descent optimization algorithms. arXiv preprint arXiv:1609.04747

arXiv 2017

[4] [4]

Xiao, H., Rasul, K., Vollgraf, R. (2017). Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747. 12

arXiv 2017