OctoInformBeta

OpenAI preprint, September 2026 · Section 2. Reductions and Newton potentials

Lemma 2.2: Newton potentials represent the scaled components when the opposite power is positive

Positive solutions in dimension n≥2n\ge2 satisfy distributional equations.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

For a nonnegative integer nn, write ℝn\mathbb R^n for real Euclidean coordinate space.✦ Given real parameters p,q,A,Bp,q,A,B, a positive solution (u,v)(u,v) is a pair of continuous, strictly positive real-valued functions on ℝn\mathbb R^n, of class C2C^2 away from the origin, satisfying the Hénon-Lane-Emden system (1) at every nonzero point: {−Δu(x)=|x|Av(x)p,−Δv(x)=|x|Bu(x)q,x≠0.\begin{cases} -\Delta u(x)=|x|^A v(x)^p,\\ -\Delta v(x)=|x|^B u(x)^q, \end{cases} \qquad x\ne0.(1)✦

The weighted power sources f,g:ℝn→ℝf,g:\mathbb R^n\to\mathbb R are the functions in (2), with real powers and value zero at the origin: f(x)={|x|Bu(x)q,x≠0,0,x=0,g(x)={|x|Av(x)p,x≠0,0,x=0.f(x)=\begin{cases}|x|^B u(x)^q,&x\ne0,\\0,&x=0,\end{cases} \qquad g(x)=\begin{cases}|x|^A v(x)^p,&x\ne0,\\0,&x=0.\end{cases}(2)✦

The Potential and Its Directional Kernel

Let σ\sigma be the unnormalized surface measure on the unit sphere Sn−1S^{n-1} in ℝn\mathbb R^n induced by Lebesgue volume, and let ωn−1\omega_{n-1} be its total area.✦ The power kernel knk_n on ℝn\mathbb R^n is the real-valued function in (3), and the Newton normalization constant dnd_n is the real number given there; for n≥3n\ge3, the normalized kernel KnK_n on ℝn\mathbb R^n is the real-valued quotient of knk_n by dnd_n: ωn−1=σ(Sn−1),kn(z)=|z|2−n,dn=(n−2)ωn−1,Kn(z)=kn(z)dn(n≥3).\omega_{n-1}=\sigma(S^{n-1}),\qquad k_n(z)=|z|^{2-n},\qquad d_n=(n-2)\omega_{n-1},\qquad K_n(z)=\frac{k_n(z)}{d_n}\quad(n\ge3).(3)

For a real-valued source hh on ℝn\mathbb R^n, its unnormalized Newton potential 𝒫h:ℝn→ℝ\mathcal P_h:\mathbb R^n\to\mathbb R is the real-valued Lebesgue integral in (4): 𝒫h(x)=∫ℝnkn(x−y)h(y)dy.\mathcal P_h(x)=\int_{\mathbb R^n}k_n(x-y)h(y)\,dy.(4)✦ A superscript ++ on an integral denotes the nonnegative extended Lebesgue integral, whose value may be infinite.

For a real-valued function ww on a real normed vector space EE, a point x∈Ex\in E and a direction e∈Ee\in E, the directional derivative Dew(x)D_e w(x) is the Fréchet derivative of ww at xx applied to ee.✦ The directional Newton kernel Hn,e:ℝn→ℝH_{n,e}:\mathbb R^n\to\mathbb R is given by (5): Dew(x)=Dw(x)[e],Hn,e(z)=(2−n)|z|−n⟨z,e⟩.D_e w(x)=Dw(x)[e],\qquad H_{n,e}(z)=(2-n)|z|^{-n}\langle z,e\rangle.(5)

The Claim

Theorem 1 (Lemma 2.2). Let nn be a nonnegative integer, let p,q,A,Bp,q,A,B be real, and let (u,v)(u,v) be a positive solution. Component exchange gives (6): (v,u) is a positive solution for the parameters (n,q,p,B,A).(v,u)\text{ is a positive solution for the parameters }(n,q,p,B,A).(6)✦ Assume n≥2n\ge2. The sources satisfy (7) and (8): g∈Lloc1(ℝn).g\in L^1_{\mathrm{loc}}(\mathbb R^n).(7)✦ f∈Lloc1(ℝn).f\in L^1_{\mathrm{loc}}(\mathbb R^n).(8)✦ For every real-valued test function φ∈Cc2(ℝn)\varphi\in C_c^2(\mathbb R^n), (9) and (10) hold: ∫ℝnu(x)Δφ(x)dx=−∫ℝng(x)φ(x)dx.\int_{\mathbb R^n}u(x)\Delta\varphi(x)\,dx =-\int_{\mathbb R^n}g(x)\varphi(x)\,dx.(9)✦ ∫ℝnv(x)Δφ(x)dx=−∫ℝnf(x)φ(x)dx.\int_{\mathbb R^n}v(x)\Delta\varphi(x)\,dx =-\int_{\mathbb R^n}f(x)\varphi(x)\,dx.(10)✦ For every x∈ℝnx\in\mathbb R^n, (11) and (12) hold, together with the decomposition (13): ∫ℝn+kn(x−y)g(y)dy≤dnu(x).\int_{\mathbb R^n}^{+}k_n(x-y)g(y)\,dy\le d_nu(x).(11)✦ ∫ℝn|kn(x−y)g(y)|dy<∞.\int_{\mathbb R^n}|k_n(x-y)g(y)|\,dy<\infty.(12)✦ There exists c≥0 such that dnu(x)=𝒫g(x)+cfor almost every x∈ℝn.\text{There exists }c\ge0\text{ such that } d_nu(x)=\mathcal P_g(x)+c\quad\text{for almost every }x\in\mathbb R^n.(13)✦ Assume further that q>0q>0. Then (14) holds, and (15) and (16) hold for every x≠0x\ne0, with arbitrary direction e∈ℝne\in\mathbb R^n: dnu(x)=𝒫g(x)for almost every x∈ℝn.d_nu(x)=\mathcal P_g(x)\quad\text{for almost every }x\in\mathbb R^n.(14)✦ dnu(x)=𝒫g(x),x≠0.d_nu(x)=\mathcal P_g(x),\qquad x\ne0.(15)✦ ∫ℝn|Hn,e(x−y)g(y)|dy<∞,dnDeu(x)=∫ℝnHn,e(x−y)g(y)dy.\begin{gathered} \int_{\mathbb R^n}|H_{n,e}(x-y)g(y)|\,dy<\infty,\\ d_nD_eu(x)=\int_{\mathbb R^n}H_{n,e}(x-y)g(y)\,dy. \end{gathered}(16)✦

In the paper, the result reads: for n≥3n\ge3, p,q>0p,q>0 and A,B>−2A,B>-2, every positive solution has f,g∈Lloc1(ℝn)f,g\in L^1_{\mathrm{loc}}(\mathbb R^n) and satisfies −Δu=g-\Delta u=g, −Δv=f-\Delta v=f distributionally on ℝn\mathbb R^n; the Newton potentials Kn*gK_n*g and Kn*fK_n*f are finite everywhere, satisfy Kn*g≤uK_n*g\le u and Kn*f≤vK_n*f\le v everywhere, and give u=cu+Kn*gu=c_u+K_n*g, v=cv+Kn*fv=c_v+K_n*f for constants cu,cv≥0c_u,c_v\ge0, almost everywhere globally and pointwise off the origin; off the origin both potentials are continuously differentiable, with ∇u(x)=∫ℝn∇Kn(x−y)g(y)dy\nabla u(x)=\int_{\mathbb R^n}\nabla K_n(x-y)g(y)\,dy and ∇v(x)=∫ℝn∇Kn(x−y)f(y)dy\nabla v(x)=\int_{\mathbb R^n}\nabla K_n(x-y)f(y)\,dy, where ∇Kn(z)=−(n−2)zKn(z)/|z|2\nabla K_n(z)=-(n-2)zK_n(z)/|z|^2 for z≠0z\ne0, and both gradient integrals converge absolutely.

What Is Checked

The assertions are stated over the positive solutions of (1), with the source convention (2) and the Euclidean potential and directional derivative constructions in (4) and (5).