REVIEW 3 major objections 6 minor 3 references
Estimations of Fourier Coefficients For Controlled Distortion of a Periodic Function
T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash
Pith's one-line read A variation bound turns harmonic reduction into a bandwidth choice
desk verdict A classical Fourier bound with a repairable regularity gap: the main inequality is true with the total variation, but the paper's identification with ∫|f'| fails for continuous singular functions. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the variation-based coefficient inequality $|a_j| \leq V_0^{2\pi}(f)/(\pi j)$. The proof mechanism is a sign-constancy partition: the period is cut into $4j$ segments on which $\cos(jx)$ and $\sin(jx)$ do not change sign, and the generalized mean-value theorem writes each coefficient as an alternating sum of function samples; because the sample differences sit on non-overlapping intervals, their absolute sum is bounded by the total variation. The same machinery, after the substitution $x=\cos\theta$, yields the Chebyshev analogue and then the abstract form $|a_j| \leq V_{\mathrm{ort}}(f)/(j\|N\|^2)$ for other orthogonal bases with $j$ sign-constancy intervals.
What would settle it
Construct the Cantor-Lebesgue function on $[0,2\pi]$, extended periodically: it is continuous, of bounded variation, and not absolutely continuous, with $\int|f'|dx=0$ while its Fourier coefficients are nonzero; computing $a_1$ numerically would violate $|a_j| \leq 0$ under the paper's stated assumptions, settling that the absolute-continuity gap is real.
Extended reading notes
Core claim
On its own terms, the paper's central claim is the coefficient estimate in Eq. (0.6): $|a_j| \leq V_0^{2\pi}(f)/(\pi j)$, where $V_0^{2\pi}(f)=\int_0^{2\pi}|f'(x)|dx$, and the analogous Chebyshev estimate in Eq. (0.18), $|a_j| \leq 2V_{-1}^1(f)/(\pi j)$. The proof partitions $[0,2\pi]$ into $4j$ intervals of constant sign for $\cos(jx)$, applies the generalized mean-value theorem on each piece, and telescopes the signs into differences $f(x_{2k-1})-f(x_{2k})$; summing absolute differences bounds the coefficient by the total variation. From there the paper derives sufficient conditions for a $q$-fold amplitude reduction: if the distorted function stays within a band of width $\delta$ around a target with zero $j$-th harmonic, and if the distortion has $N$ extrema, then $\delta = |a_j^0|\pi j/(qN)$, or in the simpler form $\delta = |a_j^0|\pi/(2q)$, guarantees $|a_j| \leq |a_j^0|/q$.
Load-bearing premise
The load-bearing premise is that total variation equals $\int|f'|dx$, which requires the function to be absolutely continuous, together with the unstated assumption that the distortion has finitely many extrema for the bandwidth formula; if either fails, the sufficiency bounds as written do not follow.
Editorial extensions
If this is right
- A distortion confined to a band of width $\delta$ with at most $N$ extrema cuts the $j$-th harmonic by at least $q = |a_j^0|\pi j/(\delta N)$.
- A larger allowed band $\delta$ yields the same harmonic reduction with less deviation from the original function, so the formulas expose a direct trade-off between control effort and required accuracy.
- The Chebyshev analogue extends the same guarantee to polynomial expansions on $[-1,1]$ through the substitution $x=\cos\theta$.
- The abstract form (0.19) suggests that any orthogonal basis whose $j$-th member has $j$ sign-constancy intervals will admit a similar variation bound.
Reading between the lines
- The stated assumptions need a repair: continuity and bounded variation alone do not make $V_0^{2\pi}(f)$ equal to $\int|f'|dx$; replacing $V$ with the sum of the variation of the continuous part plus jumps would make the bound valid for all bounded-variation functions.
- The band condition controls variation and extrema, not pointwise error, so a practical controller should verify the variation budget of the actuator, not just the geometric band.
- The same partitioning argument should transfer to other oscillatory bases, such as Haar or wavelet-type systems, giving a general recipe for suppressing individual coefficients.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves inequalities bounding the j-th Fourier coefficients of a continuous periodic function of bounded variation in terms of its variation or range: |a_j| ≤ V_0^{2π}(f)/(πj) and an analogue for Chebyshev coefficients, |a_j| ≤ 2V_{-1}^1(f)/(πj), and uses them to derive sufficient conditions on the width δ of a permitted distortion band that guarantee a requested q-fold reduction of a selected harmonic, namely δ = |a_j^0|πj/(qN) or δ = |a_j^0|π/(2q). The proof uses subdivision of the period into 4j sign-constancy intervals of the trigonometric kernel, the generalized mean value theorem for integrals, and a total-variation estimate.
Significance. If the stated inequalities are read with V as the total variation and with a finite-extrema assumption added, the core result is correct and the control application is coherent. The paper's sufficient conditions are falsifiable and directly usable: any distortion confined to the indicated band is guaranteed to produce the requested harmonic reduction. Credit is due for deriving the bounds from first principles via the mean value theorem and for explicitly linking the inequalities to a real control target; the self-reference to the author's earlier engineering report [3] is motivational and not circular. The mathematical novelty is modest, however, as the coefficient bound is a standard partition/integration-by-parts estimate, and the main obstacles are technical hypotheses that the manuscript currently leaves implicit.
major comments (3)
- [Eq. (0.6)] The definition V_0^{2π}(f)=∫_0^{2π}|f'(x)|dx is used as the right-hand side of the bound, but the paper only assumes continuity and bounded variation. For a continuous singular function such as a periodically reflected Cantor function, f'=0 almost everywhere, so the displayed right-hand side is zero, while the Fourier coefficients are not all zero; the inequality as stated fails. The proof up to (0.5) actually establishes |a_j| ≤ V_0^{2π}(f)/(πj) with V the total variation, so the fix is to define V as total variation or to add absolute continuity and estimate the total variation separately. This is load-bearing because formulas (0.9)–(0.12) inherit the definition.
- [Eq. (0.17)] The chain of equalities proving the Chebyshev bound is incorrect: d(cosθ)=-sinθ dθ, so ∫_0^π |f'(cosθ)|dθ is not equal to ∫_{-1}^1 |df(x)|. The correct identity is ∫_0^π |f'(cosθ)| sinθ dθ = V_{-1}^1(f) for absolutely continuous f, or V_0^π(f(cosθ))=V_{-1}^1(f) directly by the monotone change of variable for total variation. Although the claimed inequality (0.18) can be repaired by this argument, the proof as printed is invalid and must be rewritten.
- [Eqs. (0.7), (0.11)] The manuscript assumes without stating it that f and f△ have a finite number N of extrema. A continuous function of bounded variation may have infinitely many local extrema (for example, a convergent sum of scaled triangular waves), in which case the sum over extrema in (0.7) is not defined and the formula δ=|a_j^0|πj/(qN) in (0.11) has no meaning. The finiteness of N, or an alternative variation estimate, should appear as an explicit hypothesis before these formulas are used.
minor comments (6)
- [Eqs. (0.3), (0.15)] The notation (-1)^{k\2} is ambiguous; use (-1)^{⌊k/2⌋} or an explicit alternation rule.
- [Eq. (0.14)] After the substitution x=cosθ, the coefficient formula should show the Jacobian: dx=-sinθ dθ, so the equality as displayed skips a step; the final expression is correct but the writing should not omit the transformation details.
- [Eq. (0.8)] It would help to state explicitly that this bound follows from (0.5) because there are 2j terms each bounded by Δ.
- [Eq. (0.11)] The formula divides by N; the N=0 case is discussed in the text but the formula itself should be qualified as N≥1.
- [Abstract and introduction] The hypotheses should be stated precisely, since 'continuous periodic function of bounded variation' is insufficient for the later integral expression ∫|f'|dx; the authors should either assume absolute continuity or define V as total variation.
- [Throughout] There are numerous typographical issues, including the title 'ESTIMA TIONS', inconsistent dashes, and garbled spacing in words such as 'difference'; a careful proofreading is needed.
Circularity Check
No circular derivation: the Fourier-coefficient bound is proved from total variation via partition and mean-value estimates, and the only self-citation [3] is a non-load-bearing application note. The main defect is an unstated absolute-continuity assumption in V=∫|f'|, which is a correctness issue, not a circular reduction.
full rationale
The central inequality (0.6) is derived self-containedly: Eq. (0.2) partitions [0,2π] into 4j intervals where cos(jx) has constant sign, Eq. (0.3) applies the generalized mean value theorem, and Eq. (0.4)–(0.5) rewrite the coefficient as a sum of differences over disjoint subintervals. The bound by the full variation then follows directly from the triangle inequality and the definition of total variation. No fitted parameter is involved, no target harmonic amplitude is inserted into the derivation, and the control formulas (0.9)–(0.12) are merely sufficient conditions obtained by substituting the desired reduction |a_j^0|/q into (0.7); they do not assume the conclusion. The only self-reference, [3], appears in the closing note 'This approach ... was used for the reduction of sound generated by rotating bodies in the fluid flow [3]', which is not used to prove (0.6), (0.18), or the control formulas. Thus there is no circular step in the derivation chain. Two genuine correctness gaps should be noted, but they are not circularity: the written identity V_0^{2π}(f)=∫_0^{2π}|f'(x)|dx in (0.6) is valid only for absolutely continuous functions, not for all continuous functions of bounded variation; and Eq. (0.11) silently assumes a finite number N of extrema of f△. Both are repairable by adding assumptions and do not make any result equivalent to its input.
Assumptions & free parameters
assumptions (5)
- domain assumption Total variation equals the integral of |f'| over [0,2π].
- ad hoc to paper The number N of extrema of f (and later f△) is finite.
- standard math Generalized mean value theorem for integrals applies on each subinterval.
- domain assumption For the Chebyshev case, V_0^π(f(cosθ)) equals V_−1^1(f(x)).
- ad hoc to paper Every member of a generic orthogonal basis has j zeros and a norm that makes Eq. (0.19) valid.
Cite this review
Pith. "Pith review of Estimations of Fourier Coefficients For Controlled Distortion of a Periodic Function." pith.science (2026). https://pith.science/paper/SLUADN63
@misc{pith2026241210376,
author = {Pith},
title = {Pith review of: Estimations of Fourier Coefficients For Controlled Distortion of a Periodic Function},
year = {2026},
howpublished = {\url{https://pith.science/paper/SLUADN63}},
note = {Machine review of arXiv:2412.10376}
}
read the original abstract
There is a class of physical filtration processes where the input is adequately modeled by a continuous periodic function f (x) of bounded variation over its period, and the output depends only on certain harmonics of the Fourier expansion of f (x) in the orthogonal basis of trigonometric functions. One example is the discrete spectrum sound generation by a revolving body in a steady fluid flow. This type of sound can be controlled through the amplitudes of certain harmonics of the circular distribution of the inflow velocity. Attainable goals of such a controlled distortion of f (x) are formulated as fuzzy targets and required inequalities relating the integral characteristics of f (x) with the coefficients of the corresponding Fourier expansion are proven.
Reference graph
Works this paper leans on
-
[3]
V.L. Sluchak. Method of control of the harmonics of the ve locity field upstream of a propeller. Shipbuilding Industry, RUMB, Leningrad , page 6, 1986
work page 1986
-
[1]
V.F. Bavin, Y.N. Zavadovsky, Y.L. Levkovsky, and V.G. Mi shkevich. Marine Propellers: Mod- ern Methods of Calculation . Sudostroenie Publishing House, Leningrad, 1983. English trans- lation by UK Ministry of Defense, 1991
work page 1983
-
[2]
W alter Rudin. Real and Complex Analysis . McGraw-Hill, 3 edition, 1987
work page 1987
Reviewed August 12, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.