Independent audit of Ferreira–Simon's diagonal monotonicity conjecture

What Lean checks

Complete proposed proof of strict increase throughout the full unit parameter interval. The independent audit checks the Pollard integral, classical sine quantile, derivative signs, reciprocal pairing and strictness. No numerical or finite-case argument is used.

Date: 2026-11-21. Related issue: #171.

Scope and source

The primary source is Rui A. C. Ferreira and Thomas Simon, Mittag-Leffler functions and convex ordering, arXiv:2512.04940v1. Remark 5(b), printed page 14, proposes that the function α↦Eα,α(−1)\alpha\mapsto E_{\alpha,\alpha}(-1) increases throughout (0,1)(0,1). The proof under review establishes strict increase for every pair 0<α<β<10<\alpha<\beta<1.

This audit concerns mathematical validity and the exact statement. A separate novelty audit is required before counting a new resolution.

Integral identity and normalization

Equation (8) in Section 3 of the source gives

Eα(−xα)=∫0∞e−xrgα(r)dr,gα(r)=sin(πα)rα−1π(r2α+2cos(πα)rα+1).E_\alpha(-x^\alpha)=\int_0^\infty e^{-xr}g_\alpha(r)\,dr,\qquad g_\alpha(r)=\frac{\sin(\pi\alpha)r^{\alpha-1}}{\pi(r^{2\alpha}+2\cos(\pi\alpha)r^\alpha+1)}.

The denominator is positive for r>0r>0 and 0<α<10<\alpha<1. The identity at x=0x=0 normalizes gαg_\alpha to mass one. The defining Mittag–Leffler series gives

ddxEα(−xα)=−xα−1Eα,α(−xα).\frac{d}{dx}E_\alpha(-x^\alpha)=-x^{\alpha-1}E_{\alpha,\alpha}(-x^\alpha).

Differentiation of the integral is justified near x=1x=1: for x≥1/2x\ge1/2, the absolute derivative re−xrr e^{-xr} is bounded by 2/e2/e, an integrable constant with respect to the probability density gαg_\alpha. Therefore

Eα,α(−1)=∫0∞re−rgα(r)dr.E_{\alpha,\alpha}(-1)=\int_0^\infty r e^{-r}g_\alpha(r)\,dr.

There is no extra factor of α\alpha in this identity.

Quantile substitution

For u∈(0,1)u\in(0,1), put

qα(u)=(sin(παu)sin(πα(1−u)))1/α.q_\alpha(u)=\left(\frac{\sin(\pi\alpha u)}{\sin(\pi\alpha(1-u))}\right)^{1/\alpha}.

Let a=παa=\pi\alpha, θ=au\theta=au, and y=qα(u)α=sinθ/sin(a−θ)y=q_\alpha(u)^\alpha=\sin\theta/\sin(a-\theta). Direct differentiation and expansion give

dydθ=sinasin2(a−θ),y2+2cos(a)y+1=sin2asin2(a−θ).\frac{dy}{d\theta}=\frac{\sin a}{\sin^2(a-\theta)},\qquad y^2+2\cos(a)y+1=\frac{\sin^2a}{\sin^2(a-\theta)}.

Consequently

dqαdu=πqα1−αsina(qα2α+2cos(a)qαα+1)=1gα(qα(u)).\frac{dq_\alpha}{du} =\frac{\pi q_\alpha^{1-\alpha}}{\sin a} \left(q_\alpha^{2\alpha}+2\cos(a)q_\alpha^\alpha+1\right) =\frac{1}{g_\alpha(q_\alpha(u))}.

The function qαq_\alpha is strictly increasing, with endpoint limits zero and infinity. It is exactly the inverse distribution function for gαg_\alpha. This direct calculation avoids any dependency on stable-variable conventions.

Pointwise contraction and strictness

The reciprocal symmetry qα(1−u)=qα(u)−1q_\alpha(1-u)=q_\alpha(u)^{-1} is exact. For u>1/2u>1/2,

sin(au)−sin(a(1−u))=2cos(a/2)sin(a(u−1/2))>0,\sin(au)-\sin(a(1-u)) =2\cos(a/2)\sin(a(u-1/2))>0,

so sα(u)=logqα(u)>0s_\alpha(u)=\log q_\alpha(u)>0. At fixed uu, differentiation gives

∂sα(u)∂α=h(παu)−h(πα(1−u))α2,h(z)=zcotz−logsinz.\frac{\partial s_\alpha(u)}{\partial\alpha} =\frac{h(\pi\alpha u)-h(\pi\alpha(1-u))}{\alpha^2}, \qquad h(z)=z\cot z-\log\sin z.

Since h′(z)=−zcsc2z<0h'(z)=-z\csc^2z<0 on (0,π)(0,\pi), this derivative is strictly negative for u>1/2u>1/2.

Pairing uu with 1−u1-u converts the desired expectation to

Eα,α(−1)=2∫1/21f(sα(u))du,f(s)=ese−es+e−se−e−s2.E_{\alpha,\alpha}(-1)=2\int_{1/2}^1 f(s_\alpha(u))\,du,\qquad f(s)=\frac{e^s e^{-e^s}+e^{-s}e^{-e^{-s}}}{2}.

For s>0s>0,

2f′(s)=ese−es(1−es)−e−se−e−s(1−e−s)<0.2f'(s) =e^s e^{-e^s}(1-e^s) -e^{-s}e^{-e^{-s}}(1-e^{-s})<0.

Both displayed terms are strictly negative. Hence for every 0<α<β<10<\alpha<\beta<1 and every u∈(1/2,1)u\in(1/2,1),

f(sα(u))<f(sβ(u)).f(s_\alpha(u))<f(s_\beta(u)).

The integrands are bounded and measurable. Integrating proves strict increase, without differentiating an improper integral in the parameter α\alpha.

Verdict

The integral normalization, quantile Jacobian, parameter derivative, symmetry, strict comparison, and full source scope all check. No mathematical gap was found in this proof. It is an analytic proof for the full open interval, with no finite verification or numerical inference.