Pith. sign in

REVIEW 3 major objections 4 minor 13 references

Expansion of a bivariate symmetric mean in the neighborhood of the first bisector

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper's claim is that a single second derivative, evaluated on the diagonal, determines how a symmetric mean compares to the arithmetic mean near the diagonal and, for homogeneous means, globally.

desk verdict Useful local characteristic-function framework and clean classification results, but the headline global comparison theorem for homogeneous means is false and the homogeneity/additivity characterizations overclaim. read the letter →

arxiv 2506.07601 v1 pith:VSY2WPTZ submitted 2025-06-09 math.NT

Bakir Farhi This is my paper · ORCID
classification math.NT MSC 26E6041A5839B6226B0539B22
keywords bivariatemeanscharacteristicfunctionfirstbisectorexpansioncomparisonofnormaladditiveintegralarithmetic-geometricmean
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

Every infinitely differentiable symmetric mean M has a unique expansion near the diagonal x=y in even powers of (x-y), with the arithmetic mean A as the leading term; the coefficient of (x-y)^2 is half the characteristic function Q_M(x)=∂²M/∂x²(x,x). The paper proves that comparing Q_M1 and Q_M2 pointwise decides which mean is larger near the diagonal, and for homogeneous means the same conclusion holds for all positive x,y. For six natural classes of means—normal, additive, integral of the first and second kind, and weighted integral of the first and second kind—the characteristic function takes a simple differential form in the defining function, and the map from means to characteristic functions is bijective. Concrete consequences include the chain A≥I≥AGM≥L≥G≥H and the fact that the only power means that are normal are A, G, and H. The arithmetic-geometric mean is shown to be an M-mean built from a specific weighted integral mean of the first kind, which recovers Q_AGM(x)=-1/(8x).

What carries the argument

The central object is the characteristic function Q_M(x)=∂²M/∂x²(x,x), the second partial derivative of the mean with respect to its first argument evaluated on the diagonal. Symmetry of M forces the mixed second derivative on the diagonal to be -Q_M(x), which is why the expansion around the diagonal contains only even powers. The workhorse identities are Q_M=1/2 P'/P for normal means with weight P, Q_M=1/4 f''/f' for additive means, Q_M=1/12 f''/f' for integral means of the first kind, Q_M=1/6 ($f^{{-1}}$)''/($f^{{-1}}$)' for integral means of the second kind, and their weighted analogues with the constants c_1(g) and μ(g)=∫ t(1-t)g(t)dt. These identities make Q_M an invariant that can be inverted: an antiderivative of any smooth function can be exponentiated to produce a generator with that characteristic function (completeness), while equality of two such expressions forces the generators to differ by an affine change (minimality).

What would settle it

Take P(x)=$e^{{x^3}}$, form the normal mean M(x,y)=(xP(x)+yP(y))/(P(x)+P(y)), and compute ∂²M/∂x² at (x,x) directly; the paper's formula predicts Q_M(x)=3x²/2, so any deviation would falsify the characteristic-function formula for normal means. A second check: if two distinct C^∞ symmetric means in one of the six classes are found with identical Q_M, the claimed minimality of that class is false.

Watch

Extended reading notes

Core claim

Near the first bisector, the paper establishes that M(x,y)=A(x,y)+f_1(A)(x-y)^2+f_2(A)(x-y)^4+..., with f_1=Q_M/2, for every C^∞ symmetric mean admitting the expansion; for homogeneous means this becomes A+Σ c_n (x-y)^{2n}/$A^{{2n-1}}$, and for additively homogeneous means the coefficients are constants. The defining identity is Q_M=2f_1, and all higher f_n are rational linear combinations of the characteristic functions $Q_M^{{(k)}}$ and their derivatives. The paper computes characteristic functions for the classical means and derives from them the comparison chain A≥I≥AGM≥L≥G≥H as well as parameter-monotonicity of power, Lehmer, Stolarsky, and Gini means. It then proves minimality and completeness of six classes: equality of Q_M characterizes equality of means inside each class, and every smooth one-variable function on (0,∞) is realized as Q_M for some member. This yields classification results—the only homogeneous additive means are power means, the only homogeneous integral means of the first kind are Stolarsky means, and the harmonic mean is not an integral mean of the first kind—and, via the composition rule Q_{$f^{{-1}}$∘M∘f}=f''/(4f')+f'(Q_M∘f), the representation AGM=$h^{{-1}}$∘I_{1/√t,g}∘h with h(t)=t² and g(t)=1/(π√{t(1-t)}).

Load-bearing premise

The load-bearing premise is that every mean is infinitely differentiable, so the diagonal Taylor expansion and the second-derivative characteristic function exist; if means are only continuous or piecewise smooth, the expansions, comparison criterion, and classification theorems are not guaranteed.

Editorial extensions

If this is right

  • If the paper is right, comparing most classical bivariate means reduces to comparing one-variable functions: for homogeneous means, Q_M1>Q_M2 everywhere implies M1≥M2 everywhere.
  • Within each of the six classes, the characteristic function is a complete invariant; this means a mean in such a class can be identified uniquely by the single function Q_M rather than by its defining generator.
  • The classification theorems rule out many intersections: the only normal power means are A, G, H; the only homogeneous additive means are power means; the only homogeneous integral means of the first kind are Stolarsky means; and H is not an integral mean of the first kind.
  • The arithmetic-geometric mean acquires a new finite description as h^{-1}∘M∘h, with h(t)=t² and M a fixed weighted integral mean of the first kind, from which its characteristic function -1/(8x) follows.
  • The expansion technique gives explicit fifth-order integral approximations, one for each integral-mean kind, that involve only f and |f'| evaluated at the endpoints.

Reading between the lines

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

  • Since the map M↦Q_M is bijective on each class, Q_M can be viewed as a coordinate on the space of means; a natural next step is to describe intersections of classes such as normal ∩ additive by equating their differential expressions for Q, which the paper leaves as an open problem.
  • The approximation formulas of Propositions 3.29 and 3.36 suggest practical quadrature rules for integrals with close bounds; testing the O((x-y)^5) error numerically on monotone functions would be a direct extension of the paper's results.
  • The almost-arithmetic example in Section 2.6.1 shows Q_M=0 does not force M=A among C^∞ means; extending the theory to weaker regularity would require replacing Q_M by a finite-difference or distributional analogue, since the paper's uniqueness of A as the only homogeneous and additively homogeneous mean breaks down without differentiability (Remark 2.10).
  • The AGM representation as an M-mean may open a route to new compound-mean iterations: perturbing the integrality function or the symmetric distribution changes Q_M and shifts the local comparison class while preserving the structure of Gauss-type recursion.
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

3 major / 4 minor

Summary. The paper studies the behavior of a C^∞ symmetric bivariate mean M near the diagonal by deriving an expansion M(x,y)=A(x,y)+f_1(A)(x-y)^2+f_2(A)(x-y)^4+..., introducing the characteristic function Q_M(x)=∂^2M/∂x^2(x,x), and proving that, locally, Q_{M1}>Q_{M2} implies M1≥M2 near the first bisector. It computes Q_M for many classical means, proves minimality/completeness (bijectivity of M↦Q_M) for several classes of means, and introduces M-means, showing that the arithmetic-geometric mean is an M-mean of a weighted integral mean of the first kind. The paper also claims, in Corollary 2.28, that for homogeneous means the local comparison becomes global.

Significance. If corrected, the paper's local theory is attractive and useful: the characteristic function provides a univariate invariant for comparing means near the diagonal, the explicit formulas for classical means are checkable, and the completeness/minimality results for normal, additive, and integral mean classes give a clean classification picture. The proof of the AGM as an M-mean is a nice constructive result. However, the advertised global comparison theorem for homogeneous means is false, and two remarks claiming that local expansion forms characterize global homogeneity or additive homogeneity are also false; these issues affect the abstract and Section 2.6.4.

major comments (3)
  1. [§2.6.4, Corollary 2.28] The global comparison theorem for homogeneous means is false. Let M1(x,y)=A(x,y)+1/4·(x-y)^2/(x+y) and M2(x,y)=A(x,y)+1/2·(x-y)^4/(x+y)^3. Both are C^∞, symmetric, homogeneous, and satisfy min≤Mi≤max on (0,∞)^2; their characteristic functions are Q_M1(x)=1/(4x) and Q_M2(x)=0, so Q_M1>Q_M2 pointwise. Yet M1(10,1)=323/44≈7.34 while M2(10,1)=21202/2662≈7.96, contradicting the claimed global conclusion. The proof fails because Proposition 2.27 supplies, for each center x, a neighborhood whose width may depend on x; homogeneity then only yields a conical neighborhood |x-y|/A<ε, not all of (0,∞)^2. This invalidates the abstract's claim that characteristic inequalities become global for homogeneous means and the applications in Examples 2.29, whose conclusions may be true but are not established by this theorem.
  2. [Remarks 2.4 and 2.6] The converses asserted in Remark 2.4 (an expansion of the homogeneous form characterizes homogeneity) and Remark 2.6 (an expansion with constant coefficients characterizes additive homogeneity) are not valid. The expansion near the first bisector only records the germ of M on some neighborhood of the diagonal; it cannot determine the global scaling or translation behavior of M. For example, one can take a homogeneous mean with the prescribed expansion and add a smooth symmetric perturbation supported away from the diagonal; the expansion is unchanged while the homogeneity is destroyed. These remarks should be weakened to one-way implications, or restated under the much stronger hypothesis that the expansion holds on all of (0,∞)^2.
  3. [Theorem 2.1 and Proposition 2.27] The phrase 'in a neighborhood of the first bisector' is used ambiguously. If each φ_x admits a Taylor expansion around 0, the radius of convergence or the neighborhood on which the expansion represents φ_x may depend on x, so no single uniform neighborhood of the diagonal need exist. The statements are correct when read pointwise ('for each point of the diagonal there is a neighborhood'), but the uniformity is exactly what is needed for the false globalization in Corollary 2.28. The paper should state the local nature explicitly and should not use the nonuniform formulation to draw global conclusions.
minor comments (4)
  1. [Throughout] There are several typographical errors: 'bissector' in Section 2.6.1, 'si strictly contained' in Remark 4.3, 'integral mans' in Remark 3.38, and the page range '707-7024' in reference [6].
  2. [Section 1] The definition of a mean as a C^∞ function is stated clearly, but it would help to note explicitly that the Taylor-expansion framework relies on this regularity and that means which are merely continuous are outside the scope of the paper.
  3. [Examples 2.29] After the correction of Corollary 2.28, the displayed chain A≥I≥AGM≥L≥G≥H and the parameter-mean inequalities need a different justification; the author should either cite the known global results or state these comparisons as local ones.
  4. [Remark 2.18] The notation Q_M[1]'' in equation (2.22) would be clearer as (Q_M[1])'' or d²Q_M[1]/dx², especially since the superscript brackets also index the order of the characteristic function.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all results are derived from definitions and standard calculus; the only self-citation is introductory context.

full rationale

The paper's central object Q_M is defined as the second partial derivative of M with respect to its first variable evaluated on the diagonal (Definition 2.12), and the expansion near the first bisector is proved from Taylor's theorem (Theorem 2.1, Corollaries 2.3 and 2.5), so the link between the second derivative and the leading deviation from A is a proved identity rather than an input. Each classification result begins from the defining formula of the corresponding class and differentiates it to obtain Q_M (Propositions 3.4, 3.14, 3.23, 3.33, 3.46, and 3.52); minimality and completeness then follow by solving elementary ordinary differential equations, with no fitted parameter and no external datum. The arithmetic-geometric mean result (Proposition 4.8) is verified by substituting the explicit weighted integral mean and changing variables, reducing it to the classical Gauss elliptic-integral formula (1.3); it does not assume that AGM is an M-mean. The only self-citation is reference [9], which supplies the pre-existing definition and terminology of normal means; this is used as a class definition rather than as authority for the new conclusions, and the uniqueness of weight functions up to a constant is re-derived from equality of characteristic functions in Corollary 3.5. No step renames a known result as a prediction, and no conclusion is forced by the definition of Q_M alone. Even if the global homogeneity extension in Corollary 2.28 were mathematically questionable, that would be a correctness concern rather than circularity. Accordingly, no circular step is identified.

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

The central results rest only on standard calculus, smoothness assumptions, and classical facts about means. No free parameters are fitted, no new entities are postulated, and no circular steps appear.

assumptions (5)
  • domain assumption Means are C^infinity and symmetric on (0,infty)^2, satisfying the betweenness property.
    This is the definition of a mean in Section 1; it is required for Taylor expansions and for defining Q_M.
  • standard math Taylor's theorem and the binomial theorem apply to the smooth functions involved.
    Used throughout, in particular in Theorem 2.1 and Corollary 2.7.
  • domain assumption Gauss's integral representation of the arithmetic-geometric mean, formula (1.3).
    Quoted from Borwein and Borwein; used in Proposition 4.8 to identify the AGM as an M-mean.
  • domain assumption Bullen's theorem that additivity functions of a quasi-arithmetic mean are unique up to affine transformation.
    Cited as [3, Theorem 5] and used in Proposition 3.12 and Corollary 3.16.
  • domain assumption The definition of normal means from the author's earlier paper [9].
    Section 3.1 uses this definition as the starting point for the class analysis.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Expansion of a bivariate symmetric mean in the neighborhood of the first bisector." pith.science (2026). https://pith.science/paper/VSY2WPTZ

@misc{pith2026250607601,
  author       = {Pith},
  title        = {Pith review of: Expansion of a bivariate symmetric mean in the neighborhood of the first bisector},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VSY2WPTZ}},
  note         = {Machine review of arXiv:2506.07601}
}
abstract

In this paper, we investigate the behavior of a bivariate mean $M$ near the first bisector by establishing, in several significant cases, an important expansion of $M$ derived from the Taylor expansion of a single-variable function. These expansions are made explicit for a number of classical means. This motivates the introduction of the concept of the characteristic function $Q_M$ of a mean $M$, defined as the second partial derivative of $M$ with respect to its first variable, evaluated along the diagonal. The function $Q_M$ measures the proximity of $M$ to the arithmetic mean near the first bisector and provides a univariate analytic framework for comparing and classifying means. We prove that inequalities between characteristic functions yield local inequalities between the corresponding means, and that in the case of homogeneous means, such inequalities hold globally. We also examine several important classes of means, both classical and novel, including: normal means, additive means, integral means of the first kind, integral means of the second kind, weighted integral means of the first kind, and weighted integral means of the second kind. For each class, we determine the specific form taken by the characteristic functions $Q_M$ of the means $M$ it contains, and we then study the injectivity and the surjectivity of the mapping $M \mapsto Q_M$ within the class. We also use characteristic functions to investigate intersections between certain classes of means, highlighting one of the key strengths of this concept. Finally, we introduce and study, for a given mean $M$, the class of $M$-means, and show, in particular, that the arithmetic-geometric mean $\mathrm{AGM}$ is an $M$-mean for a specific weighted integral mean of the first kind $M$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

  1. [1]

    Alzer & S

    A. Alzer & S. Ruscheweyh. On the intersection of two-parameter mean value families, Proc. Amer. Math. Soc. , 29 n°9 (2001), p. 2655-2662

    work page 2001

  2. [2]

    J. M. Borwein & P. B. Borwein . Pi and the AGM , A Study in Analytic Number Theory and Computational Complexity, Wiley, New York, 1987

    work page 1987

  3. [3]

    P. S. Bullen. Handbook of means and their inequalities, 560, Springer Science & Business Media, 2013

    work page 2013

  4. [4]

    Buri ´c & N

    T. Buri ´c & N. Elezovi ´c. Computation and analysis of the asymptotic expansions of the compound means, Appl. Math. Comput. , 303 (2017), p. 48-54

    work page 2017

  5. [5]

    Elezovi ´c

    N. Elezovi ´c. Asymptotic inequalities and comparison of classical means, J. Math. In- equal., 9, n°1 (2015), p. 177-196

    work page 2015

  6. [6]

    Elezovi ´c & L

    N. Elezovi ´c & L. Vuk ˇsi´c. Asymptotic expansions of bivariate classical means and related inequalities, J. Math. Inequal. , 8, n°4 (2014), p. 707-7024

    work page 2014

  7. [7]

    Elezovi´c & L

    N. Elezovi´c & L. Vukˇsi´c. Asymptotic expansions and comparison of bivariate param- eter means, Math. Inequal. Appl. , 17, n°4 (2014), p. 1225-1244

    work page 2014

  8. [8]

    Elezovi ´c & L

    N. Elezovi ´c & L. Vuk ˇsi´c. Asymptotic expansions of integral means and applications to the ratio of gamma functions, Appl. Math. Comput. , 235 (2014), p. 187-200

    work page 2014

Show all 13 references
  1. [9]

    B. F arhi. Algebraic and topological structures on the set of mean functions and general- ization of the AGM mean, Colloq. Math. , 132 (2013), p. 139-149

    2013

  2. [10]

    C. Gini. Di una Formula Compressiva delle Medie, Metron, 13 (1938), p. 3-22

    1938

  3. [11]

    D. H. Lehmer . On the compounding of certain means, J. Math. Anal. Appl. , 36 (1971), p. 183-200

    1971

  4. [12]

    Mihokovi `c

    L. Mihokovi `c. Coinciding mean of the two symmetries on the set of mean functions, Axioms, 12 n°3 (2023), 238

    2023

  5. [13]

    K. B. Stolarsky. Generalizations of the logarithmic mean, Math. Mag., 48, n°2 (1975), p. 87-92. 47

    1975

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.