Pith. sign in

REVIEW 4 minor 17 references

Sharp Ternary Martingale Isoperimetry and $n$-adic Takagi-Type Lower Bounds

T0 review · 0 major / 4 minor · reviewed 2026-07-14 · grok-4.5

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

Pith's one-line read The sharp ternary martingale isoperimetric profile is an explicit Takagi-type series that is not the classical ternary Takagi function.

desk verdict Exact ternary profile via a new Takagi-type Bellman function, plus clean general lower bounds of the right order; the computer QE step is the only non-hand piece and is framed correctly. read the letter →

arxiv 2607.11069 v1 pith:XMZA2SYP submitted 2026-07-13 math.PR math.CA

Natanael Alpay This is my paper · ORCID
classification math.PRmath.CA MSC 60G4205C3505D0542B2539B6211A63
keywords martingaleisoperimetryn-adicfiltrationsBellmanfunctionsTakagiedge-isoperimetricinequalitiesdigitsumsone-variation
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

The paper asks how small the one-variation of an indicator can be when its measure is fixed, under the regular n-adic martingale filtration on the unit interval. For the ternary filtration it gives the exact answer: the profile equals a self-similar Takagi-type series built from an explicit piecewise-linear generator that records the cheapest one-residual split of a parent density into three children. That function is larger than the classical ternary Takagi–van der Waerden function at some points, so mean-absolute-deviation cost produces a genuinely different profile. For every base n the same profile is bounded from below by the usual n-adic Takagi function, which already yields the sharp logarithmic order |A|* log(1/|A|*). In the quasi-norm range 0<α<1 the logarithm disappears and the distance to {0,1} itself becomes admissible. The results link martingale inequalities to digit-sum combinatorics and edge-isoperimetry on trees, and they identify the precise local cost that must be minimized at every vertex.

What carries the argument

The n-point Bellman inequality for continuous functions F that vanish at 0 and 1: F(¯x)^α ≤ (1/n) ∑ (F(xi)^β + |xi-¯x|^β)^{α/β}. Any such F lower-bounds the L^α norm of the β-variation of every indicator; the sharp ternary profile is the maximal such F for α=β=1, n=3.

What would settle it

Re-run the quantifier-elimination check of the 729 local-debt inequalities on [0,1]^3 (or find a single rational counter-example triple (a,b,r)); if any fails, T3 is not admissible and the exact ternary profile claim collapses.

Watch

Extended reading notes

Core claim

For the ternary filtration the isoperimetric profile V3(x) equals exactly the Takagi-type series T3(x)=∑ 3^{-j} ψ3({3^j x}), where ψ3 is the min of two linear pieces coming from optimal one-residual splits; this T3 is admissible as a Bellman function, is attained by a recursive one-residual construction, and is strictly larger than the classical ternary Takagi function at points such as 1/3.

Load-bearing premise

The local debt inequality that closes the inductive compression for the ternary series must hold on the whole unit cube; it is asserted by exact quantifier elimination of 729 rational piecewise-affine inequalities.

Editorial extensions

If this is right

  • Any measurable set of measure x under the ternary filtration has one-variation at least T3(x), and the bound is attained by an explicit recursive construction.
  • For every base n the one-variation is at least a constant multiple of |A|* log(1/|A|*), and the constant is sharp along the leftmost n-adic intervals.
  • In the range 0<α<1 the same variation is bounded below simply by |A|*, again with matching order along n-adic intervals.
  • The sharp ternary profile is a non-classical Takagi function, so mean-absolute-deviation cost and range cost produce different isoperimetric profiles even in the same base.

Reading between the lines

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

  • The same one-residual generator that works for n=3 already fails the Bellman inequality for n=4, so the sharp profiles for n≥4 may require multi-residual states or a genuinely different finite-state description.
  • Because the lower bound is built from summatory digit sums, the same arithmetic may yield sharp constants for other combinatorial isoperimetric problems on regular trees.
  • The endpoint α-norm bound suggests that the transition from logarithmic to linear growth occurs exactly at α=1; intermediate Orlicz norms could interpolate the two regimes.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 4 minor

Summary. The paper studies the martingale isoperimetric profile V_n(x) = inf{|A|=x} ||S_1(1_A)||_1 for the regular n-adic filtration on [0,1). For n=3 it identifies the profile exactly as the Takagi-type function T_3 built from an explicit one-residual generator ψ_3 (Theorem 2.3), and shows that T_3 coincides with the Bellman envelope B_{1,1,3}. For general n it constructs an admissible digit-sum Bellman function P^{(n)} (Theorem 2.7) that dominates the usual n-adic Takagi–van der Waerden function ω_n, yielding the lower bound ||S_1(1_A)||_1 ≥ ω_n(|A|*) ≍_n |A|* log(1/|A|*) with matching order along leftmost n-adic intervals (Theorem 2.6). A sub-L^1 endpoint ||S_1(1_A)||_α ≥ |A|* for 0<α<1 is also proved and shown to be sharp up to a constant depending only on α and n (Theorem 2.9). The argument rests on a Bellman principle (Theorem 2.2), a recursive one-residual construction that attains T_3, and a finite exact verification of a local-debt inequality that closes the ternary compression.

Significance. The exact ternary profile is a genuine contribution: it shows that the mean-absolute-deviation cost produces a non-classical Takagi-type Bellman function strictly larger than ω_3 at some points (e.g., T_3(1/3)=4/9 > 1/3=ω_3(1/3)). The general-n lower bound of the correct logarithmic order, obtained from the Allouche–Stipulanti summatory digit-sum inequality, cleanly links martingale isoperimetry to digit-sum combinatorics and recovers the known dyadic case. The sub-L^1 endpoint is elementary but sharp. Strengths include an explicit attaining construction (Lemma 4.6), a fully printed Wolfram script that reduces the only non-hand step (Lemma 4.2) to 729 first-order statements decided by exact quantifier elimination, and the absence of free parameters or fitted constants. The open problem for n≥4 is stated honestly.

minor comments (4)
  1. [Figure 1] Figure 1 is referenced for the comparison of candidates and for the failure of the naive one-residual generator when n≥4, but the caption alone does not make the plotted curves fully self-explanatory; a short legend or explicit formula list in the caption would help.
  2. [Abstract / Eq. (10)] In the abstract and Theorem 2.3 the generator is written as a min of two expressions involving |t-1/2|; the piecewise definition (10) used in the proofs is equivalent but not identical in appearance. A one-line remark that the two presentations coincide would remove any momentary confusion.
  3. [Theorem 2.6] The constant c_n = 2(n-1)/n appearing in the sharpness statement of Theorem 2.6 is computed explicitly in Section 6; it would be useful to record it already in the theorem statement.
  4. [References / throughout] A few typographical slips (e.g., missing spaces after commas in displayed formulae, and the duplicated Hart reference [11,12]) should be cleaned in production.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: sharp ternary profile and general lower bounds are derived from explicit constructions, an elementary Bellman principle, and a published external digit-sum inequality.

full rationale

The paper's central claims are self-contained. The Bellman principle (Theorem 2.2) is proved from first principles via an n-point inequality and martingale convergence. For n=3, T3 is defined by an explicit generator ψ3 coming from one-residual mean-deviation costs (Lemma 4.1); admissibility is obtained by a finite inductive compression (Lemma 4.4) whose only non-hand step is a computer-checked local-debt inequality (Lemma 4.2 / Appendix A) consisting of 729 first-order statements over the reals with rational coefficients, decided by exact quantifier elimination rather than by fitting or by self-citation. Attainment is shown by an explicit recursive construction of sets Ex whose one-variation equals T3 term-by-term (Lemma 4.6). For general n the lower bound uses the continuous extension of a digit-sum Bellman function P(n) built from the summatory inequality of Allouche–Stipulanti (external, published), compared pointwise with the classical ωn; sharpness of the logarithmic order is verified by direct computation on leftmost n-adic intervals. The sub-L1 endpoint uses the elementary distance function x∗ as an admissible Bellman function. No parameter is fitted to data and then re-presented as a prediction; no uniqueness theorem is imported from the authors' prior work to force the profile; the only self-reference is ordinary comparison with the dyadic case. The computer verification is a finite exact check of the paper's own inequalities, not a circular reduction of the claim to its inputs. Hence the derivation chain does not collapse by construction.

Assumptions & free parameters 0 free parameters · 3 assumptions · 2 invented entities

The paper works entirely inside standard real analysis and martingale theory. The only external non-elementary input is the Allouche–Stipulanti summatory digit-sum inequality, used as a black-box lemma. No free parameters are fitted; the generators ψ3 and ηn are derived from one-step mean-deviation costs. The computer verification is an exact decision procedure, not a numerical fit.

assumptions (3)
  • standard math Allouche–Stipulanti summatory base-n digit-sum inequality (Lemma 5.4 / [3, Thm 4.2])
    Used verbatim to prove the n-point Bellman inequality for the digit-sum function P^{(n)} (Proposition 5.5).
  • standard math Martingale convergence / Lebesgue differentiation for the regular n-adic filtration
    Invoked in the proof of the Bellman principle (Section 3) to pass to the limit M o∞.
  • standard math Existence and uniqueness of the continuous extension of P^{(n)} from the n-adic rationals
    Follows from uniform convergence of the Takagi-type series once the generator ηn is continuous and bounded.
invented entities (2)
  • Ternary generator ψ3 and the associated Takagi-type function T3
    purpose: Serves as the exact Bellman function for the ternary one-variation profile.
    Defined explicitly from the one-residual mean-deviation cost; shown admissible by finite verification and attained by construction. No independent physical or combinatorial existence claim beyond the math.
  • Digit-sum Bellman function P^{(n)} with generator ηn
    purpose: Provides an admissible lower bound dominating the classical ωn for every n.
    Constructed from the summatory function Gn of base-n digit sums; continuous extension and Bellman inequality proved directly.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sharp Ternary Martingale Isoperimetry and $n$-adic Takagi-Type Lower Bounds." pith.science (2026). https://pith.science/paper/XMZA2SYP

@misc{pith2026260711069,
  author       = {Pith},
  title        = {Pith review of: Sharp Ternary Martingale Isoperimetry and $n$-adic Takagi-Type Lower Bounds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XMZA2SYP}},
  note         = {Machine review of arXiv:2607.11069}
}
abstract

Let $S_1$ be the one-variation associated with the regular $n$-adic martingale filtration on $[0,1)$. We study the martingale isoperimetric profile \[ V_n(x):= \inf_{\substack{A\subset[0,1)\ {\rm measurable}\\ |A|=x}} \|S_1(\mathbbm 1_A)\|_1 . \] For the ternary filtration we determine this profile exactly. Namely, \[ V_3(x)=T_3(x):= \sum_{j=0}^{\infty}3^{-j}\psi_3(\{3^j x\}), \] where \[ \psi_3(t)= \min\left\{ \frac{1+2\left|t-\frac12\right|}{3}, \frac{2-4\left|t-\frac12\right|}{3} \right\}, \qquad 0\le t\le1 . \] Thus the sharp ternary profile is a Takagi-type Bellman function. It is, however, not the usual ternary Takagi--van der Waerden function $\omega_3$; for example, \[ T_3(1/3)=4/9, \qquad \omega_3(1/3)=1/3 . \] For general $n\ge2$, we prove that every measurable $A\subset[0,1)$ satisfies \[ \|S_1(\mathbbm 1_A)\|_1 \ge \omega_n(|A|^*) \asymp_n |A|^*\log\frac1{|A|^*}, \qquad |A|^*:=\min\{|A|,1-|A|\}. \] Moreover, this logarithmic order is sharp up to a constant depending only on $n$. Finally, for every $0<\alpha<1$, we prove the endpoint estimate \[ \|S_1(\mathbbm 1_A)\|_\alpha\ge |A|^*, \] and show that it is sharp up to a constant depending only on $\alpha$ and $n$.

Figures

Figures reproduced from arXiv: 2607.11069 by the authors.

Figure 1
Figure 1. Comparison of several Takagi-type candidates on n-adic grids. For n = 2 the natural candidates coincide. For n = 3, the sharp ternary Bellman function T3 is strictly larger than the usual Takagi–van der Waerden function ω3 at some points. For n = 4, 5, the picture shows the lower bound ωn, together with experimental Bellman profiles and the naive one-residual extension; the latter fails to be admissible for n ≥ 4. 2… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 2 linked inside Pith

  1. [1]

    Cambridge University Press, 2014

    PieterCAllaart.Hausdorffdimensionoflevelsetsofgeneralizedtakagifunctions.InMathematical Proceedings of the Cambridge Philosophical Society, volume 157, pages 253–278. Cambridge University Press, 2014

    2014

  2. [2]

    Allaart and Kiko Kawamura

    Pieter C. Allaart and Kiko Kawamura. The takagi function: a survey.Real Analysis Exchange, 37(1):1–54, 2011

    2011

  3. [3]

    Numeration, Liège 2023

    Jean-Paul Allouche and Manon Stipulanti. Summing the sum of digits.Communications in Mathematics, 33(2), 2025. Published online May 9, 2024; special issue “Numeration, Liège 2023”

    2025

  4. [4]

    Sharp lower bounds for dyadic square functions of indicator functions of sets.arXiv preprint arXiv:2502.16045, 2025

    Natanael Alpay and Paata Ivanisvili. Sharp lower bounds for dyadic square functions of indicator functions of sets.arXiv preprint arXiv:2502.16045, 2025

    arXiv 2025

  5. [5]

    On sharp isoperimetric inequalities on the hypercube, 2023

    David Beltran, Paata Ivanisvili, and Jose Madrid. On sharp isoperimetric inequalities on the hypercube, 2023

    2023

  6. [6]

    A class of functional inequalities and the takagi function.Journal of Inequalities and Applications, 6(4):379–393, 2001

    Gábor Boros and Zsolt Páles. A class of functional inequalities and the takagi function.Journal of Inequalities and Applications, 6(4):379–393, 2001

    2001

  7. [7]

    D. L. Burkholder, B. J. Davis, and R. F. Gundy. Integral inequalities for convex functions of operators on martingales. In L. Le Cam, J. Neyman, and E. L. Scott, editors,Proceedings of the Sixth Berkeley Symposium on Mathematical Statistics and Probability, volume 2 ofProbability Theory, pages 223–240. University of California Press, 1972

    1972

  8. [8]

    D. L. Burkholder and R. F. Gundy. Extrapolation and interpolation of quasi-linear operators on martingales. Acta Mathematica, 124:249–304, 1970

    1970

Show all 17 references
  1. [9]

    On thelp norms of stochastic integrals and other martingales.Duke Mathematical Journal, 43(4):697–704, 1976

    Burgess Davis. On thelp norms of stochastic integrals and other martingales.Duke Mathematical Journal, 43(4):697–704, 1976

    1976

  2. [10]

    Sharp isoperimetric inequalities on the hamming cube near the critical exponent, 2024

    Pavel Durcik, Paata Ivanisvili, and Joris Roos. Sharp isoperimetric inequalities on the hamming cube near the critical exponent, 2024

    2024

  3. [11]

    A note on the edges of then-cube.Discrete Mathematics, 14(2):157–163, 1976

    Sergiu Hart. A note on the edges of then-cube.Discrete Mathematics, 14(2):157–163, 1976

    1976

  4. [12]

    A note on the edges of the n-cube.Discrete Mathematics, 14(2):157–163, 1976

    Sergiu Hart. A note on the edges of the n-cube.Discrete Mathematics, 14(2):157–163, 1976

    1976

  5. [13]

    Ein beispiel einer nicht analytischen funktion.Mathematische Zeitschrift, 3(1–2):93–97, 1919

    Konrad Knopp. Ein beispiel einer nicht analytischen funktion.Mathematische Zeitschrift, 3(1–2):93–97, 1919

    1919

  6. [14]

    Edge–Isoperimetric problem for cayley graphs and generalized takagi function.Israel Journal of Mathematics, 2023

    Vsevolod Lev. Edge–Isoperimetric problem for cayley graphs and generalized takagi function.Israel Journal of Mathematics, 2023. arXiv:2303.01234

    2023 arXiv

  7. [15]

    A simple example of a continuous function without derivative.Proc

    Teiji Takagi. A simple example of a continuous function without derivative.Proc. Phys. Math. Soc. Japan, Ser. II, 1:176–177, 1903. English translation reprinted inThe Collected Papers of Teiji Takagi, Springer, 1990

    1903

  8. [16]

    Sharp inequalities for the conditional square function of a martingale.The Annals of Probability, 19(4):1679–1688, 1991

    Gang Wang. Sharp inequalities for the conditional square function of a martingale.The Annals of Probability, 19(4):1679–1688, 1991

    1991

  9. [17]

    Sharp square-function inequalities for conditionally symmetric martingales.Transactions of the American Mathematical Society, 328(1):393–419, 1991

    Gang Wang. Sharp square-function inequalities for conditionally symmetric martingales.Transactions of the American Mathematical Society, 328(1):393–419, 1991. (NA) Department of Mathematics, University of California, Ir vine, Ir vine, CA 92697, USA Email address:nalpay@uci.edu

    1991

Pith tools

Reviewed July 14, 2026 · model on record in the stance chip above.