d'Alembert's Functional Equation and a Globally Convex Free-Action Principle on Positive Paths

Jonathan Washburn; Sebastian Pardo-Guerra

d'Alembert's Functional Equation and a Globally Convex Free-Action Principle on Positive Paths

T0 review · 0 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read Starting from d'Alembert's functional equation and one step-evaluation postulate, the paper proves the cosh kinetic action is strongly convex, with a unique global minimizer and an exact Bregman gap identity.

desk verdict A clean, honest paper that proves a correct but elementary convexity theorem; the d'Alembert framing is provocative but the load-bearing assumption is clearly labeled, and the mechanics bridge is properly conditional. read the letter →

arxiv 2607.22594 v1 pith:EIKBY2LT submitted 2026-06-10 math.OC math.FA

Sebastian Pardo-Guerra , Jonathan Washburn This is my paper · ORCID

classification math.OCmath.FA MSC 49J0526A5139B2253B1249S05

keywords d'AlembertfunctionalequationstrongconvexityleastactionprincipleBregmandivergenceduallyflatgeometrycoshkineticgeometricinterpolationFriedrichs–Poincaréinequality

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 shows that d'Alembert's functional equation, after calibration, forces a cosh cost, and that a single modeling postulate – evaluating this cost at the log-velocity rather than the log-position – turns that cost into a kinetic action that is strongly convex under geometric (log-space) interpolation of paths. This convexity gives a genuine global least-action principle: for fixed positive endpoints, the unique minimizer is the uniform-log-velocity path, and the action gap to any other path decomposes exactly as an integrated Bregman divergence, with a quantitative Friedrichs–Poincaré lower bound. The derivation needs no Euler–Lagrange equation, derivative, or second variation. The paper carefully limits the physical bridge: recovering Newtonian or rapidity mechanics requires four additional postulates, and the cosh-dual Hamiltonian is not the special-relativistic free Hamiltonian. If the step-evaluation postulate is not accepted, the kinetic action and all its consequences do not follow from d'Alembert's equation.

What carries the argument

The central object is the cosh kinetic action A[γ]=∫(cosh ξ̇−1)dt with ξ=log γ, built from the d'Alembert-calibrated cost via Postulate 2.10. The load-bearing mechanism is the log change of coordinates, which converts multiplicative (geometric) interpolation of positive paths into affine interpolation in log space; pointwise strong convexity of cosh (second derivative ≥1) then propagates through integration to give 1-strong convexity of A with an explicit L² slack. The exact action gap uses the cosh Bregman divergence D_K(v∥w)=cosh v−cosh w−sinh w(v−w), and the Friedrichs (Wirtinger) inequality supplies the quantitative lower bound. The whole package is interpreted dually-flat/Hessian in the

What would settle it

A numerical search over fixed-endpoint positive paths in a high-resolution spline basis that finds any path with action smaller than the uniform-log-velocity path would refute the global-minimizer claim; equivalently, computing the action gap for a non-geodesic path and checking whether A[γ]−A[γ*] is strictly less than the Bregman integral would invalidate the Pythagorean identity. The theorem predicts the gap is exactly that integral, so any discrepancy within numerical precision would be a counterexample.

Watch

Extended reading notes

Core claim

Calibrated d'Alembert's equation H(t+u)+H(t-u)=2H(t)H(u) with H(0)=1 and H''(0)=1 selects H(t)=cosh t. Writing the positive half-line in log coordinates ξ=log x, the induced log-cost is eJ(ξ)=cosh ξ−1. The paper's single postulate (Postulate 2.10) evaluates this cost at the log-velocity ξ̇, producing the kinetic action A[γ]=∫(cosh ξ̇−1)dt. The central theorem (Theorem 4.2) states that A is 1-strongly convex along geometric interpolation of paths, meaning A[interp×(γ1,γ2,s)] ≤ (1−s)A[γ1]+sA[γ2]−s(1−s)/2 ∫(ξ̇1−ξ̇2)²dt. Consequently (Corollary 4.10) the unique fixed-endpoint minimizer is the uniform-log-velocity path γ*(t)=exp(log x_a + (t−a)/(b−a)(log x_b−log x_a)), and the action gap obeys th

Load-bearing premise

The entire kinetic action and its convexity rest on Postulate 2.10, the modeling choice to evaluate the d'Alembert log-cost at the log-velocity ξ̇ rather than at the log-position ξ; d'Alembert's equation itself fixes the function cosh−1 but says nothing about its argument.

Editorial extensions

If this is right

Global minimality without calculus: a one-sided chord condition along geometric interpolation characterizes global minimizers, replacing stationary-action checks with a pure convexity argument.
Closed-form geodesic: for any fixed positive endpoints, the unique minimizer of the free action is the uniform-log-velocity path, with action A*(T,Δ)=T(cosh(Δ/T)−1).
Exact gap identity: A[γ]−A[γ*]=∫D_K(ξ̇∥ξ̇*)dt, so the action gap is a Bregman divergence; the Friedrichs–Poincaré bound gives A[γ]−A[γ*] ≥ π²/(2(b−a)²) ∥log(γ/γ*)∥²_L².
Perspective/subadditivity: A* is jointly convex in (T,Δ) and positively 1-homogeneous, so geodesic concatenation is subadditive: splitting an interval never beats the single geodesic.
Conditional physics bridge: with four additional postulates (kinematic embedding, mass coupling, time calibration, Hamiltonian-primary Legendre structure), the cosh action has the Newtonian small-step limit and rapidity profile m(γ_L−1), but the cosh-dual Hamiltonian is strictly larger than the SR free Hamiltonian.

Reading between the lines

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

If the theorem transfers to higher dimensions (componentwise on R^n_{>0} or symmetric positive-definite matrices), the same log-space convexity might give explicit minimizers for matrix-valued interpolation problems, potentially connecting to Bures–Wasserstein geometry.
The exact Bregman gap suggests a projection/geodesic interpretation: the uniform-log-velocity path is the Bregman projection of any path onto the endpoint constraint, so the action principle is a one-dimensional instance of a more general geodesic-convexity-plus-projection result for Bregman energies.
Because the strong convexity slack is exactly the L² difference of log-velocities, the action gap can serve as a data-driven distance between positive paths, yielding a divergence for shape analysis or time-series alignment on positive data.
The exponential barrier for large log-displacement over short times (A* ~ ½ e^{|Δ|/T}T) suggests the cosh action as a regularizer in optimal control, penalizing sudden jumps; this testable property could guide path-planning on positive variables.

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.

Desk Editor's Note

read the letter

You should know two things up front. First, the mathematics is correct: every proof I checked, from strong convexity under geometric interpolation through the Jensen minimizer, the Bregman/Pythagorean identity, and the Friedrichs–Poincaré bound, is sound. Second, the novelty is real but modest: the central theorem is pointwise strong convexity of cosh plus Jensen, and the authors say so themselves in Remark 4.4 and Section 9.1. The paper is an honest piece of convex analysis dressed in a d'Alembert–provenance costume, not a deep new variational principle.

What the paper does well: it is unusually explicit about what is and is not claimed. Postulate 2.10, evaluating the cosh cost at log-velocity rather than log-position, is named as a modeling postulate and flagged in Remark 2.11 as not forced by d'Alembert. The mechanics bridge is gated behind four named postulates, and Proposition 6.9 flatly says the cosh-dual Hamiltonian is not the SR Hamiltonian. The derivations in the physical section check out, and the small-velocity limit is handled with a sharp error bound. This is careful, honest work.

The soft spots are proportional to how soft they actually are. The weakest point is the gap between the title's promise—a kinetic action induced by d'Alembert's equation—and the actual dependency on Postulate 2.10. If you reject that postulate, the whole convexity package evaporates. The authors admit this, but the framing still leans on 'd'Alembert forces' more than the proof supports. The other soft spot is minor: Proposition 6.19 leaves the k=m boundary case undecided for non-strict local minimality. That does not affect the free-sector result.

Who is this for? A reader working in Bregman geometry, convex variational problems, or the cost-first ledger programme will find a clean worked example with complete proofs. A reader looking for fundamentally new mathematics will be disappointed. I would send it to a serious referee, because it is correct, clearly delimited, and the conditional structure is a useful template. My own verdict is a qualified accept—not because the theorem is deep, but because the paper says exactly what it proves and proves it well.

Referee Report

0 major / 5 minor

Summary. The paper studies the kinetic action A[γ]=∫_a^b (cosh ξ̇ −1)dt with ξ=log γ, interpreted as the d'Alembert-calibrated cost evaluated at the log-velocity (Postulate 2.10). The central result (Theorem 4.2) is that A is 1-strongly convex under geometric/log-space interpolation of positive paths, with explicit quadratic slack. This yields a chord-form characterization of global minimality (Theorem 4.7), the unique uniform-log-velocity minimizer for fixed endpoints (Corollary 4.10), an exact Bregman/Pythagorean gap identity (Theorem 4.13) with Friedrichs–Poincaré lower bound (Corollary 4.15), and the joint convexity/1-homogeneity of the minimum-action profile A*(T,Δ) (Proposition 4.16), together with a dually-flat reading. The authors then build a conditional bridge to Newtonian and rapidity mechanics via four additional postulates, prove the Newtonian small-velocity limit, and stress (Proposition 6.9) that the cosh-dual Hamiltonian is not the special-relativistic free-particle Hamiltonian. Finally, they show that adding a non-affine strictly convex potential destroys joint convexity and restores the classical stationary-action/conjugate-point picture (Section 8).

Significance. If the results hold, the paper offers a clean, fully proven example where a functional equation, supplemented by one explicitly named modeling postulate, produces a globally—not just locally—minimizing free action with a closed-form geodesic and exact quantitative gap. The authors are unusually careful in delimiting what is forced by d'Alembert's equation (the function cosh−1) and what is postulated (evaluation at the log-velocity), and in separating the mathematical theorem from the conditional physical bridge. The central proofs are complete and self-contained, with explicit constants in the strong-convexity slack, the Friedrichs constant, and the sharp quartic remainder. The paper also provides a useful dually-flat/Hessian interpretation. The main limitations—the conditional status of the mechanics bridge and the open k=m boundary in Proposition 6.19—are transparently acknowledged and do not affect the free-sector claim.

minor comments (5)

[Remark 4.6] The heading 'Vacuity of the boundary cases 0 = 1' appears to contain a typo: it should read 's0 = 1'. Also, 'Ats0 = 1' is missing a space.
[Section 4.2 / Theorem 4.2] In the equality clause, 'equivalently, γ2/γ1 is constant on [a,b]' might be clearer as 'equivalently, ξ2 − ξ1 is constant on [a,b]' (equivalently γ2/γ1 is constant), since the proof works with ξ. This is purely stylistic.
[Section 6.3 / Proposition 6.19] The k=m boundary case is explicitly left open. The manuscript handles this honestly; a sentence indicating that a second-order transverse analysis would be required would further help the reader, but this is not necessary.
[Section 9.1] The statement that 'the only one forced by the d'Alembert calibration is the native cosh–sinh Lagrangian' could be misread, since L_nat also requires the Hamiltonian-primary Legendre structure and the independent binding coupling k. Suggest adding 'together with the Hamiltonian-primary and additive-cost postulates of §6' for precision.
[Abstract / Introduction] The phrase 'Calibrated d'Alembert forces the cosh cost' is accurate, but since the kinetic action also depends on Postulate 2.10, consider a small rewording such as 'forces the cosh cost function' to avoid any appearance that the action itself is forced without the postulate. The paper's own Remark 2.11 already makes this clear.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central convexity theorem is conditional on an explicitly named postulate, and no derivation step reduces to its own input.

full rationale

The derivation chain is self-contained conditional on Postulate 2.10. The d'Alembert classification (Theorem 2.2) is recalled from external sources (Aczél, Stetkær), and the calibration H''(0)=1 is an explicitly stated unit choice, not a hidden input. The kinetic cost K(v)=cosh(v)-1 is not claimed to follow from d'Alembert alone: Postulate 2.10 names the step-evaluation choice, and Remark 2.11 says plainly it is a modeling postulate, not a consequence of d'Alembert's equation. Strong convexity of A (Theorem 4.2) is proved directly from pointwise 1-strong convexity of cosh under geometric interpolation, with an explicit equality clause; it does not presuppose the minimizer or the action gap. Corollary 4.10 is a Jensen-inequality argument; Theorem 4.13 is an exact Bregman rearrangement whose linear term vanishes because the endpoints are shared; Corollary 4.15 uses the standard one-dimensional Friedrichs inequality. None of these steps is fitted, renamed, or defined in terms of its conclusion. The self-citations [2,3,4] provide the interpretive 'comparison ratio' framing, but the paper explicitly states these upstream choices are inputs, not consequences of the free-sector analysis; the convexity theorem stands independently of them. The paper also flags its own possible circularity concern in Remarks 4.6 and 4.9 by requiring an interior chord parameter s0 in (0,1) so the hypothesis does not degenerate into the conclusion. The physical bridge is explicitly conditional (Section 6, Remark 6.2, Proposition 6.9) and disclaims identity with the special-relativistic Hamiltonian. The unresolved k=m boundary case in Proposition 6.19 is a limitation, but it is not central to the free-sector theorem and does not create circularity. No circular step is exhibited.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The free-sector theorem relies only on the standard functional-equation classification and the explicitly named Postulate 2.10. The mechanical bridge adds mass/binding couplings (m,k) and kinematic/Hamiltonian postulates, none fitted to data. No new particles, forces, or dimensions are introduced.

free parameters (2)

mass coupling m
Introduced in Definition 6.4 as a species-specific prefactor multiplying the dimensionless d'Alembert kinetic cost; its numerical value is not derived, only the form m(cosh φ−1). It is needed only for the conditional mechanical bridge, not for the free-sector convexity theorem.
binding coupling k
Introduced in Definition 6.11 as a prefactor for the native static cost k(cosh ξ−1); like m, its provenance is declared outside the paper's scope. Used only in the physical bridge layer.

assumptions (5)

standard math d'Alembert's functional equation with continuity and calibration H(0)=1, H''(0)=1 forces H(t)=cosh t (Aczél classification).
Theorem 2.2; classical result used to fix the cost function. Assumes continuity, which the paper restates via Aczél's smoothness theorem.
ad hoc to paper Postulate 2.10: the d'Alembert log-cost is evaluated at the log-velocity ξ̇ rather than the log-position ξ.
This is the single modeling postulate that turns the static cosh cost into the kinetic action. It is explicitly not derived from d'Alembert (Remark 2.11) and is the load-bearing assumption for the entire convexity result.
ad hoc to paper Kinematic embedding q=ξ (or q=c t₀ ξ in standard units) and φ=dξ/dτ identified as rapidity.
Definition 6.1; needed for the conditional bridge to mechanics; it supplies the physical interpretation but is not implied by the free-sector math.
ad hoc to paper Hamiltonian primacy: H=T_H(p)+V(ξ) is primary; the Lagrangian is its Legendre dual.
Definition 6.7; this postulate, plus mass coupling and time calibration, makes the Legendre sign and Newtonian limit follow; part of the conditional bridge.
domain assumption Interpretation of R_{>0} as comparison ratios in the cost-first ledger framework.
The paper takes the positive half-line and its log-coordinate from the authors' own prior framework (refs [2,3]); not needed for the proof, only for motivation.

how reviews work

0 comments

Cite this review

Pith. "Pith review of d'Alembert's Functional Equation and a Globally Convex Free-Action Principle on Positive Paths." pith.science (2026). https://pith.science/paper/EIKBY2LT

@misc{pith2026260722594,
  author       = {Pith},
  title        = {Pith review of: d'Alembert's Functional Equation and a Globally Convex Free-Action Principle on Positive Paths},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EIKBY2LT}},
  note         = {Machine review of arXiv:2607.22594}
}

abstract

We study the kinetic action that d'Alembert's functional equation induces on positive paths in $\Rplus$, and prove it strongly convex. Calibrated d'Alembert forces the cosh cost $\Jcost(x)=\tfrac12(x+x^{-1})-1$, i.e.\ $\Jlog(\xi)=\cosh\xi-1$ in the log coordinate $\xi=\log x$. Evaluating this log-cost at the log-\emph{velocity} $\dot\xi$ rather than the log-position -- a single postulate (Postulate~\ref{post:step}) -- yields $\actionA[\gamma]=\int_a^b(\cosh\dot\xi-1)\,dt$, strongly convex under geometric (log-space) interpolation. This convexity has three consequences, none requiring an Euler--Lagrange equation, a Fr\'echet derivative, or a second variation. First, a one-sided chord condition characterizes global minimality. Second, the unique fixed-endpoint minimizer is the uniform-log-velocity path. Third, the action gap obeys an exact Bregman / Pythagorean identity $\actionA[\gamma]-\actionA[\gamma_*]=\int D_\Kkin(\dot\xi\,\|\,\dot\xi_*)\,dt$, sharpened by a quantitative Friedrichs--Poincar\'e bound on $\log(\gamma/\gamma_*)$. It has a dually-flat / Hessian-manifold reading in the additive coordinate $\xi$. \\ This theorem is purely mathematical, and we delimit it. The bridge to Newtonian and rapidity mechanics is \emph{conditional}, requiring structure beyond Postulate~\ref{post:step}: a kinematic embedding, a mass coupling, a time calibration, and a Hamiltonian-primary Legendre structure. Granted these, the cosh action recovers the Newtonian small-step limit and the rapidity profile $\Kkin_m(\phi)=m(\gamma_L-1)$; yet the cosh-dual Hamiltonian is \emph{not} the special-relativistic free-particle Hamiltonian (Proposition~\ref{prop:not-SR}), the agreement being one of profile, not an identity of Hamiltonians. Global minimality is a free-sector phenomenon: once a non-affine strictly convex potential is added, joint convexity is lost and the classical stationary-action picture returns.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references

[1]

V. I. Arnold.Mathematical Methods of Classical Mechanics. Graduate Texts in Mathemat- ics, vol. 60. Translated by K. Vogtmann and A. Weinstein. Springer-Verlag, New York, 2nd edition, 1989

1989
[2]

Coherent Comparison as Informa- tion Cost: Axiomatic Foundations for Discrete Ledger Dynamics

Pardo-Guerra, S.; Thapa, A.; Simons, M.; Washburn, J. Coherent Comparison as Informa- tion Cost: Axiomatic Foundations for Discrete Ledger Dynamics. Foundations 2026, 6, 17. https://doi.org/10.3390/foundations6020017

doi:10.3390/foundations6020017 2026
[3]

Uniqueness of the Canonical Reciprocal Cost

Washburn, J.; Zlatanović, M. Uniqueness of the Canonical Reciprocal Cost. Mathematics 2026, 14, 935. https://doi.org/10.3390/math14060935

doi:10.3390/math14060935 2026
[4]

Multidimensional Cost Geometry

Washburn, J.; Zlatanović, M.; Beltracchi, P. Multidimensional Cost Geometry. Axioms 2026, 15, 378. https://doi.org/10.3390/axioms15050378

doi:10.3390/axioms15050378 2026
[5]

Aczél.Lectures on Functional Equations and Their Applications

J. Aczél.Lectures on Functional Equations and Their Applications. Academic Press, New York, 1966

1966
[6]

Stetkær.Functional Equations on Groups

H. Stetkær.Functional Equations on Groups. World Scientific, Singapore, 2013

2013
[7]

L. M. Bregman. The relaxation method of finding the common point of convex sets and its application to the solution of problems in convex programming.USSR Computational Mathematics and Mathematical Physics, 7(3):200–217, 1967

1967
[8]

Amari and H

S. Amari and H. Nagaoka.Methods of Information Geometry. Translations of Mathemat- ical Monographs, vol. 191. American Mathematical Society, Providence, RI, and Oxford University Press, Oxford, 2000. 35

2000

Show all 23 references

[9]

Shima.The Geometry of Hessian Structures

H. Shima.The Geometry of Hessian Structures. World Scientific, Hackensack, NJ, 2007

2007
[10]

R. T. Rockafellar.Convex Analysis. Princeton Mathematical Series, vol. 28. Princeton University Press, Princeton, NJ, 1970

1970
[11]

Boyd and L

S. Boyd and L. Vandenberghe.Convex Optimization. Cambridge University Press, Cam- bridge, 2004

2004
[12]

G. H. Hardy, J. E. Littlewood, and G. Pólya.Inequalities. Cambridge University Press, Cambridge, 2nd edition, 1952

1952
[13]

E. Noether. Invariante Variationsprobleme.Nachrichten von der Königlichen Gesellschaft der Wissenschaften zu Göttingen, Mathematisch-physikalische Klasse, 235–257, 1918

1918
[14]

Bessel-Hagen

E. Bessel-Hagen. Über die Erhaltungssätze der Elektrodynamik.Mathematische Annalen, 84(3–4):258–276, 1921

1921
[15]

P. J. Olver.Applications of Lie Groups to Differential Equations. Graduate Texts in Mathematics, vol. 107. Springer-Verlag, New York, 2nd edition, 1993

1993
[16]

I. M. Gelfand and S. V. Fomin.Calculus of Variations. Translated and edited by R. A. Sil- verman. Prentice-Hall, Englewood Cliffs, NJ, 1963; Dover reprint, 2000

1963
[17]

Arsigny, P

V. Arsigny, P. Fillard, X. Pennec, and N. Ayache. Geometric means in a novel vector space structure on symmetric positive-definite matrices.SIAM Journal on Matrix Analysis and Applications, 29(1):328–347, 2007

2007
[18]

Bhatia, T

R. Bhatia, T. Jain, and Y. Lim. On the Bures–Wasserstein distance between positive definite matrices.Expositiones Mathematicae, 37(2):165–191, 2019

2019
[19]

J. G. Skellam. The frequency distribution of the difference between two Poisson variates belonging to different populations.Journal of the Royal Statistical Society, 109(3):296, 1946

1946
[20]

C. Léonard. A survey of the Schrödinger problem and some of its connections with optimal transport.Discrete and Continuous Dynamical Systems, 34(4):1533–1574, 2014

2014
[21]

F. Otto. The geometry of dissipative evolution equations: the porous medium equation. Communications in Partial Differential Equations, 26(1–2):101–174, 2001

2001
[22]

Ambrosio, N

L. Ambrosio, N. Gigli, and G. Savaré.Gradient Flows in Metric Spaces and in the Space of Probability Measures. Lectures in Mathematics ETH Zürich. Birkhäuser, Basel, 2nd edition, 2008

2008
[23]

Villani.Optimal Transport: Old and New

C. Villani.Optimal Transport: Old and New. Grundlehren der mathematischen Wis- senschaften, vol. 338. Springer-Verlag, Berlin, 2009. 36

2009

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.

[1] [1]

V. I. Arnold.Mathematical Methods of Classical Mechanics. Graduate Texts in Mathemat- ics, vol. 60. Translated by K. Vogtmann and A. Weinstein. Springer-Verlag, New York, 2nd edition, 1989

1989

[2] [2]

Coherent Comparison as Informa- tion Cost: Axiomatic Foundations for Discrete Ledger Dynamics

Pardo-Guerra, S.; Thapa, A.; Simons, M.; Washburn, J. Coherent Comparison as Informa- tion Cost: Axiomatic Foundations for Discrete Ledger Dynamics. Foundations 2026, 6, 17. https://doi.org/10.3390/foundations6020017

doi:10.3390/foundations6020017 2026

[3] [3]

Uniqueness of the Canonical Reciprocal Cost

Washburn, J.; Zlatanović, M. Uniqueness of the Canonical Reciprocal Cost. Mathematics 2026, 14, 935. https://doi.org/10.3390/math14060935

doi:10.3390/math14060935 2026

[4] [4]

Multidimensional Cost Geometry

Washburn, J.; Zlatanović, M.; Beltracchi, P. Multidimensional Cost Geometry. Axioms 2026, 15, 378. https://doi.org/10.3390/axioms15050378

doi:10.3390/axioms15050378 2026

[5] [5]

Aczél.Lectures on Functional Equations and Their Applications

J. Aczél.Lectures on Functional Equations and Their Applications. Academic Press, New York, 1966

1966

[6] [6]

Stetkær.Functional Equations on Groups

H. Stetkær.Functional Equations on Groups. World Scientific, Singapore, 2013

2013

[7] [7]

L. M. Bregman. The relaxation method of finding the common point of convex sets and its application to the solution of problems in convex programming.USSR Computational Mathematics and Mathematical Physics, 7(3):200–217, 1967

1967

[8] [8]

Amari and H

S. Amari and H. Nagaoka.Methods of Information Geometry. Translations of Mathemat- ical Monographs, vol. 191. American Mathematical Society, Providence, RI, and Oxford University Press, Oxford, 2000. 35

2000