OctoInformBeta

OpenAI preprint, September 2026 · §4. Gaussian heat dissipation

Lemma 4.4: Integrability with center-sum and positive-time derivatives, plus strict dissipation for cyclic permutations of (1,11/10,11/10)(1, 11/10, 11/10)

Gaussian plane heat integrals admit derivatives and dissipation bounds.

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

Consider a finite index set ι\iota, three complex ι×ι\iota \times \iota matrices (Av)v∈ℤ/3ℤ(A_v)_{v \in \mathbb{Z}/3\mathbb{Z}} assigned to the edges of a triangle, real node coordinates u:ι→ℝu : \iota \to \mathbb{R}, and a step size h>0h > 0. Weighting the nodes by Gaussians centered at p=(p0,p1,p2)∈ℝℤ/3ℤp = (p_0, p_1, p_2) \in \mathbb{R}^{\mathbb{Z}/3\mathbb{Z}} with positive variances σ=(σ0,σ1,σ2)\sigma = (\sigma_0, \sigma_1, \sigma_2) produces a weighted matrix array whose vertex star energies depend on pp and σ\sigma. Integrating these energies over the constant-sum center plane Πd={p∈ℝℤ/3ℤ:p0+p1+p2=d}\Pi_d = \{p \in \mathbb{R}^{\mathbb{Z}/3\mathbb{Z}} : p_0 + p_1 + p_2 = d\} raises two questions: given a positive rate vector α=(α0,α1,α2)\alpha = (\alpha_0, \alpha_1, \alpha_2), how the resulting plane integrals vary under shifts of dd and under linear variance scaling σv=αvs\sigma_v = \alpha_v s, and whether the total plane energy dissipates at a strictly positive rate as s>0s > 0 increases.

Matrix Energy and Hessian Variations

Let κ\kappa be a finite index set. The real trace Re⁡tr⁡(M)\operatorname{Re}\operatorname{tr}(M) of a complex square matrix M∈ℂκ×κM \in \mathbb{C}^{\kappa \times \kappa} is the real number given by the real part of the trace of MM.✦ The real Hilbert-Schmidt inner product ⟨X,Y⟩HS\langle X, Y \rangle_{\mathrm{HS}} of X,Y∈ℂκ×κX, Y \in \mathbb{C}^{\kappa \times \kappa} is the real number Re⁡tr⁡(X*Y)\operatorname{Re}\operatorname{tr}(X^* Y), where X*X^* is the conjugate transpose of XX.✦ The matrix energy Φ(P)\Phi(P) of P∈ℂκ×κP \in \mathbb{C}^{\kappa \times \kappa} with square root P1/2P^{1/2} is the real number Re⁡tr⁡(PP1/2)\operatorname{Re}\operatorname{tr}(P P^{1/2}) (which equals tr⁡(P3/2)\operatorname{tr}(P^{3/2}) when PP is positive semidefinite)✦, and the matrix energy gradient ∇Φ(P)\nabla\Phi(P) of P∈ℂκ×κP \in \mathbb{C}^{\kappa \times \kappa} is the continuous real-linear functional on ℂκ×κ\mathbb{C}^{\kappa \times \kappa} mapping KK to 32Re⁡tr⁡(P1/2K)\frac{3}{2}\operatorname{Re}\operatorname{tr}(P^{1/2} K).

The Sylvester operator SMS_M of M∈ℂκ×κM \in \mathbb{C}^{\kappa \times \kappa} is the linear map on ℂκ×κ\mathbb{C}^{\kappa \times \kappa} given by SM(X)=MX+XMS_M(X) = M X + X M✦, and the Sylvester penalty of MM and X∈ℂκ×κX \in \mathbb{C}^{\kappa \times \kappa} is the real number ⟨X,SM(X)⟩HS\langle X, S_M(X) \rangle_{\mathrm{HS}}.✦ For M,K∈ℂκ×κM, K \in \mathbb{C}^{\kappa \times \kappa}, the Sylvester variational functional is the real-valued function on ℂκ×κ\mathbb{C}^{\kappa \times \kappa} mapping XX to 2⟨K,X⟩HS−⟨X,SM(X)⟩HS2\langle K, X\rangle_{\mathrm{HS}} - \langle X, S_M(X)\rangle_{\mathrm{HS}}✦, and the root variational cost ρM(K)\rho_M(K) is the real number equal to 32\frac{3}{2} times the supremum of 2⟨K,X⟩HS−⟨X,SM(X)⟩HS2\langle K, X\rangle_{\mathrm{HS}} - \langle X, S_M(X)\rangle_{\mathrm{HS}} over X∈ℂκ×κX \in \mathbb{C}^{\kappa \times \kappa}.✦ For P,K,L∈ℂκ×κP, K, L \in \mathbb{C}^{\kappa \times \kappa}, the Hessian quadratic cost bP(K)b_P(K) and polarized Hessian pairing ℬP(K,L)\mathcal{B}_P(K, L) are the real-valued quadratic and bilinear forms defined by (1): bP(K)=ρP1/2(K),ℬP(K,L)=14(bP(K+L)−bP(K−L)).b_P(K) = \rho_{P^{1/2}}(K), \qquad \mathcal{B}_P(K, L) = \frac{1}{4}\bigl(b_P(K + L) - b_P(K - L)\bigr).(1)

The triangle block space (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota over a finite index set ι\iota is the Cartesian product of the triangle vertex set ℤ/3ℤ={0,1,2}\mathbb{Z}/3\mathbb{Z} = \{0, 1, 2\} under cyclic addition (v+1v + 1 and v−1=v+2v - 1 = v + 2 modulo 33) with ι\iota.✦ Given a family W=(Wv)v∈ℤ/3ℤW = (W_v)_{v \in \mathbb{Z}/3\mathbb{Z}} in ℂι×ι\mathbb{C}^{\iota \times \iota}, the triangle matrix blocks form the (ℤ/3ℤ)×(ℤ/3ℤ)(\mathbb{Z}/3\mathbb{Z}) \times (\mathbb{Z}/3\mathbb{Z}) matrix of ι×ι\iota \times \iota blocks whose (i,j)(i, j) block is 00 when i=ji = j, equals WiW_i when j=i+1j = i + 1, and equals Wj*W_j^* when i=j+1i = j + 1.✦ Block matrix assembly flattens any (ℤ/3ℤ)×(ℤ/3ℤ)(\mathbb{Z}/3\mathbb{Z}) \times (\mathbb{Z}/3\mathbb{Z}) matrix of ι×ι\iota \times \iota blocks into a matrix on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota✦, and the triangle matrix T(W)T(W) on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota is the matrix obtained by assembling the triangle matrix blocks of WW.✦

For each v∈ℤ/3ℤv \in \mathbb{Z}/3\mathbb{Z}, the vertex block projection QvQ_v is the diagonal matrix on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota with entry 11 at (a,k)(a, k) when a=va = v and 00 otherwise.✦ The vertex star row Rv(W)R_v(W) of WW at vv is the matrix QvT(W)Q_v T(W) on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota that isolates the block row at vv✦, and the vertex star Gram matrix of WW at vv is the matrix Rv(W)*Rv(W)R_v(W)^* R_v(W) on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota.✦

The weighted matrix array A[g]A^{[g]} of a family A=(Av)v∈ℤ/3ℤA = (A_v)_{v \in \mathbb{Z}/3\mathbb{Z}} in ℂι×ι\mathbb{C}^{\iota \times \iota} and real node weights g=(gv)v∈ℤ/3ℤg = (g_v)_{v \in \mathbb{Z}/3\mathbb{Z}} with gv:ι→ℝg_v : \iota \to \mathbb{R} is the family in ℂι×ι\mathbb{C}^{\iota \times \iota} indexed by ℤ/3ℤ\mathbb{Z}/3\mathbb{Z} with entries Av[g](i,j)=gv(i)Av(i,j)gv+1(j)A^{[g]}_v(i, j) = \sqrt{g_v(i)}\, A_v(i, j)\, \sqrt{g_{v+1}(j)}.✦ The vertex root matrix DgD_g of real node weights g=(gv)v∈ℤ/3ℤg = (g_v)_{v \in \mathbb{Z}/3\mathbb{Z}} with gv:ι→ℝg_v : \iota \to \mathbb{R} is the diagonal matrix on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota with entry ga(k)\sqrt{g_a(k)} at (a,k)(a, k).✦ For v,w∈ℤ/3ℤv, w \in \mathbb{Z}/3\mathbb{Z}, the vertex star base row of AA and gg at vv is the matrix Rv(A)DgR_v(A) D_g✦, the vertex star base column of AA and gg at vv is the matrix Rv(A)*DgR_v(A)^* D_g✦, and the fixed base matrix Bv,w(A,g)B_{v,w}(A, g) of AA and gg at v,wv, w is the matrix on (ℤ/3ℤ)×ι(\mathbb{Z}/3\mathbb{Z}) \times \iota equal to Rv(A)DgR_v(A) D_g when w=vw = v and Rv(A)*DgR_v(A)^* D_g when w≠vw \neq v.✦ The weighted Gram matrix of a matrix B∈ℂκ×κB \in \mathbb{C}^{\kappa \times \kappa} and a real weight function ω:κ→ℝ\omega : \kappa \to \mathbb{R} is the matrix B*diag⁡(ω)B∈ℂκ×κB^* \operatorname{diag}(\omega) B \in \mathbb{C}^{\kappa \times \kappa}.✦

Gaussian Heat Weights and Plane Integrals

The variance-σ\sigma Gaussian density gσ(t)g_\sigma(t) at argument t∈ℝt \in \mathbb{R} for σ>0\sigma > 0 is the real number (2πσ)−1/2exp⁡(−t2/(2σ))(2\pi\sigma)^{-1/2} \exp(-t^2/(2\sigma)).✦ Given a point x∈ℝx \in \mathbb{R}, center c∈ℝc \in \mathbb{R}, variance σ∈ℝ\sigma \in \mathbb{R}, and step size h∈ℝh \in \mathbb{R}, the Gaussian score is the real number (x−c)/σ(x - c)/\sigma✦ and the Gaussian weight is the real number hgσ(x−c)h\, g_\sigma(x - c).✦ Fixing A=(Av)v∈ℤ/3ℤA = (A_v)_{v \in \mathbb{Z}/3\mathbb{Z}} in ℂι×ι\mathbb{C}^{\iota \times \iota}, a step size h∈ℝh \in \mathbb{R}, and node coordinates u:ι→ℝu : \iota \to \mathbb{R}, for any centers p:ℤ/3ℤ→ℝp : \mathbb{Z}/3\mathbb{Z} \to \mathbb{R} and variances σ:ℤ/3ℤ→ℝ\sigma : \mathbb{Z}/3\mathbb{Z} \to \mathbb{R} the vertex heat weights 𝒘=(wv)v∈ℤ/3ℤ\mathbf{w} = (w_v)_{v \in \mathbb{Z}/3\mathbb{Z}} form the family of functions wv:ι→ℝw_v : \iota \to \mathbb{R} defined by wv(i)=hgσv(ui−pv)w_v(i) = h\, g_{\sigma_v}(u_i - p_v).✦ The heat matrix array A[𝒘]A^{[\mathbf{w}]} is the weighted matrix array of AA with vertex heat weights 𝒘\mathbf{w}✦, and the heat matrix energy ev(p,σ)e_v(p, \sigma) at vertex v∈ℤ/3ℤv \in \mathbb{Z}/3\mathbb{Z} is the real number Φ(Rv(A[𝒘])*Rv(A[𝒘]))\Phi(R_v(A^{[\mathbf{w}]})^* R_v(A^{[\mathbf{w}]})).✦

Given a step size h∈ℝh \in \mathbb{R}, a finite index set κ\kappa, coordinates ũ:κ→ℝ\tilde{u} : \kappa \to \mathbb{R}, a group assignment ℓ:κ→ν\ell : \kappa \to \nu into an index set ν\nu, and p,σ:ν→ℝp, \sigma : \nu \to \mathbb{R}, the grouped heat weight W(0):κ→ℝW^{(0)} : \kappa \to \mathbb{R} is the function given by W(0)(a)=hgσℓ(a)(ũa−pℓ(a))W^{(0)}(a) = h\, g_{\sigma_{\ell(a)}}(\tilde{u}_a - p_{\ell(a)}).✦ For a group w∈νw \in \nu, the grouped first variation weight Ww(1):κ→ℝW^{(1)}_w : \kappa \to \mathbb{R} and grouped second variation weight Ww(2):κ→ℝW^{(2)}_w : \kappa \to \mathbb{R} are the functions that vanish when ℓ(a)≠w\ell(a) \neq w and when ℓ(a)=w\ell(a) = w are defined by (2): Ww(1)(a)=ũa−pwσwhgσw(ũa−pw),Ww(2)(a)=((ũa−pwσw)2−σw−1)hgσw(ũa−pw).W^{(1)}_w(a) = \frac{\tilde{u}_a - p_w}{\sigma_w} h\, g_{\sigma_w}(\tilde{u}_a - p_w), \qquad W^{(2)}_w(a) = \biggl(\Bigl(\frac{\tilde{u}_a - p_w}{\sigma_w}\Bigr)^2 - \sigma_w^{-1}\biggr) h\, g_{\sigma_w}(\tilde{u}_a - p_w).(2) Specializing to κ=(ℤ/3ℤ)×ι\kappa = (\mathbb{Z}/3\mathbb{Z}) \times \iota and ν=ℤ/3ℤ\nu = \mathbb{Z}/3\mathbb{Z} with ũ(a,i)=ui\tilde{u}(a, i) = u_i and ℓ(a,i)=a\ell(a, i) = a, write Bv,w=Bv,w(A,𝒘)B_{v,w} = B_{v,w}(A, \mathbf{w}) for v,w∈ℤ/3ℤv, w \in \mathbb{Z}/3\mathbb{Z}. The local Gram matrix Pv,wP_{v,w}, local Gram first variation Kv,wK_{v,w}, and local Gram second variation Mv,wM_{v,w} are the weighted Gram matrices of Bv,wB_{v,w} with W(0)W^{(0)}, Ww(1)W^{(1)}_w, and Ww(2)W^{(2)}_w, respectively. From these matrices evaluated at (p,σ)(p, \sigma), the heat energy gradient ev,w(1)(p,σ)e^{(1)}_{v,w}(p, \sigma), heat Hessian cost Vv,w(p,σ)V_{v,w}(p, \sigma), pure heat second variation ev,wpure(p,σ)e^{\mathrm{pure}}_{v,w}(p, \sigma), heat variance term ev,wvar(p,σ)e^{\mathrm{var}}_{v,w}(p, \sigma), and mixed heat Hessian pairing mv,w,z(p,σ)m_{v,w,z}(p, \sigma) for v,w,z∈ℤ/3ℤv, w, z \in \mathbb{Z}/3\mathbb{Z} are the real numbers given by (3): ev,w(1)(p,σ)=∇Φ(Pv,w)(Kv,w),Vv,w(p,σ)=bPv,w(Kv,w),ev,wpure(p,σ)=bPv,w(Kv,w)+∇Φ(Pv,w)(Mv,w),ev,wvar(p,σ)=12∇Φ(Pv,w)(Mv,w),mv,w,z(p,σ)=ℬPv,w(Kv,w,Kv,z).\begin{aligned} e^{(1)}_{v,w}(p, \sigma) &= \nabla\Phi(P_{v,w})(K_{v,w}), & V_{v,w}(p, \sigma) &= b_{P_{v,w}}(K_{v,w}), \\ e^{\mathrm{pure}}_{v,w}(p, \sigma) &= b_{P_{v,w}}(K_{v,w}) + \nabla\Phi(P_{v,w})(M_{v,w}), & e^{\mathrm{var}}_{v,w}(p, \sigma) &= \frac{1}{2}\nabla\Phi(P_{v,w})(M_{v,w}), \\ m_{v,w,z}(p, \sigma) &= \mathcal{B}_{P_{v,w}}(K_{v,w}, K_{v,z}). \end{aligned}(3)

The plane coordinate chart φi,d:ℝ2→Πd\varphi_{i,d} : \mathbb{R}^2 \to \Pi_d associated with a vertex i∈ℤ/3ℤi \in \mathbb{Z}/3\mathbb{Z} and sum level d∈ℝd \in \mathbb{R} is the map sending z=(z1,z2)∈ℝ2z = (z_1, z_2) \in \mathbb{R}^2 to the triple in Πd\Pi_d equal to d−z1−z2d - z_1 - z_2 at ii and to z1,z2z_1, z_2 at the remaining two vertices in increasing order.✦ The plane integral ∫Πdf(p)dπd(p)\int_{\Pi_d} f(p)\, d\pi_d(p) at level d∈ℝd \in \mathbb{R} of a real-valued function ff on ℝℤ/3ℤ\mathbb{R}^{\mathbb{Z}/3\mathbb{Z}} is the Lebesgue integral ∫ℝ2f(φ2,d(z))dz\int_{\mathbb{R}^2} f(\varphi_{2,d}(z))\, dz. A tuple (A,h,u,d,σ)(A, h, u, d, \sigma) consisting of a matrix family A=(Av)v∈ℤ/3ℤA = (A_v)_{v \in \mathbb{Z}/3\mathbb{Z}} in ℂι×ι\mathbb{C}^{\iota \times \iota}, a step size h∈ℝh \in \mathbb{R}, node coordinates u:ι→ℝu : \iota \to \mathbb{R}, a sum level d∈ℝd \in \mathbb{R}, and variances σ:ℤ/3ℤ→ℝ\sigma : \mathbb{Z}/3\mathbb{Z} \to \mathbb{R} is heat-plane integrable if for all i,v,w∈ℤ/3ℤi, v, w \in \mathbb{Z}/3\mathbb{Z}, the compositions with φi,d\varphi_{i,d} of ev(⋅,σ)e_v(\cdot, \sigma), ev,w(1)(⋅,σ)e^{(1)}_{v,w}(\cdot, \sigma), ev,wpure(⋅,σ)e^{\mathrm{pure}}_{v,w}(\cdot, \sigma), ev,wvar(⋅,σ)e^{\mathrm{var}}_{v,w}(\cdot, \sigma), and Vv,w(⋅,σ)V_{v,w}(\cdot, \sigma) are Lebesgue integrable on ℝ2\mathbb{R}^2, and for all pairwise distinct v,w,z∈ℤ/3ℤv, w, z \in \mathbb{Z}/3\mathbb{Z}, the composition of mv,w,z(⋅,σ)m_{v,w,z}(\cdot, \sigma) with φi,d\varphi_{i,d} is Lebesgue integrable on ℝ2\mathbb{R}^2.✦

Given a sum level d∈ℝd \in \mathbb{R}, variances σ:ℤ/3ℤ→ℝ\sigma : \mathbb{Z}/3\mathbb{Z} \to \mathbb{R}, rate vector α:ℤ/3ℤ→ℝ\alpha : \mathbb{Z}/3\mathbb{Z} \to \mathbb{R}, and time s∈ℝs \in \mathbb{R}, the total plane heat energy J(d,σ)J(d, \sigma) and total plane time derivative Θ(d,α,s)\Theta(d, \alpha, s) are the real numbers defined by (4): J(d,σ)=∑v∈ℤ/3ℤ∫Πdev(p,σ)dπd(p),Θ(d,α,s)=∑v,w∈ℤ/3ℤαw∫Πdev,wvar(p,sα)dπd(p).J(d, \sigma) = \sum_{v \in \mathbb{Z}/3\mathbb{Z}} \int_{\Pi_d} e_v(p, \sigma)\, d\pi_d(p), \qquad \Theta(d, \alpha, s) = \sum_{v, w \in \mathbb{Z}/3\mathbb{Z}} \alpha_w \int_{\Pi_d} e^{\mathrm{var}}_{v,w}(p, s\alpha)\, d\pi_d(p).(4) For each i∈ℤ/3ℤi \in \mathbb{Z}/3\mathbb{Z}, the strict dissipation rate α(i):ℤ/3ℤ→ℝ\alpha^{(i)} : \mathbb{Z}/3\mathbb{Z} \to \mathbb{R} is the vector with αw(i)=1\alpha^{(i)}_w = 1 for w=iw = i and αw(i)=11/10\alpha^{(i)}_w = 11/10 for w≠iw \neq i, and the strict positivity constant c*c_* is the real number given by (5): c*=12min⁡{1−352,1110−122}.c_* = \frac{1}{2} \min\Bigl\{1 - \frac{3}{5}\sqrt{2},\, \frac{11}{10} - \frac{1}{2}\sqrt{2}\Bigr\}.(5)✦

The Claim

Theorem 1. Fix a finite index set ι\iota, a family A=(Av)v∈ℤ/3ℤA = (A_v)_{v \in \mathbb{Z}/3\mathbb{Z}} of complex ι×ι\iota \times \iota matrices, a step size h>0h > 0, node coordinates u:ι→ℝu : \iota \to \mathbb{R}, and a center sum d∈ℝd \in \mathbb{R}. For every positive variance vector σ:ℤ/3ℤ→(0,∞)\sigma : \mathbb{Z}/3\mathbb{Z} \to (0, \infty), the plane integrability condition (6) holds: (A,h,u,d,σ) is heat-plane integrable.(A, h, u, d, \sigma) \text{ is heat-plane integrable.}(6)✦ Under the same positivity assumption on σ\sigma, for all v,w∈ℤ/3ℤv, w \in \mathbb{Z}/3\mathbb{Z} the plane energy and first variation integrals are differentiable at dd with derivatives (7) and (8): ddx|x=d∫Πxev(p,σ)dπx(p)=∫Πdev,w(1)(p,σ)dπd(p),\frac{d}{dx}\Big|_{x=d} \int_{\Pi_x} e_v(p, \sigma)\, d\pi_x(p) = \int_{\Pi_d} e^{(1)}_{v,w}(p, \sigma)\, d\pi_d(p),(7)✦ ddx|x=d∫Πxev,w(1)(p,σ)dπx(p)=∫Πdev,wpure(p,σ)dπd(p).\frac{d}{dx}\Big|_{x=d} \int_{\Pi_x} e^{(1)}_{v,w}(p, \sigma)\, d\pi_x(p) = \int_{\Pi_d} e^{\mathrm{pure}}_{v,w}(p, \sigma)\, d\pi_d(p).(8)✦ When v,w,z∈ℤ/3ℤv, w, z \in \mathbb{Z}/3\mathbb{Z} are pairwise distinct, the first variation integral is also differentiable at dd with mixed derivative (9), and for any i∈ℤ/3ℤi \in \mathbb{Z}/3\mathbb{Z} the pure and mixed plane integrals coincide as in (10): ddx|x=d∫Πxev,w(1)(p,σ)dπx(p)=∫Πdmv,w,z(p,σ)dπd(p),\frac{d}{dx}\Big|_{x=d} \int_{\Pi_x} e^{(1)}_{v,w}(p, \sigma)\, d\pi_x(p) = \int_{\Pi_d} m_{v,w,z}(p, \sigma)\, d\pi_d(p),(9)✦ ∫Πdev,ipure(p,σ)dπd(p)=∫Πdmv,w,z(p,σ)dπd(p).\int_{\Pi_d} e^{\mathrm{pure}}_{v,i}(p, \sigma)\, d\pi_d(p) = \int_{\Pi_d} m_{v,w,z}(p, \sigma)\, d\pi_d(p).(10)✦

Let α:ℤ/3ℤ→(0,∞)\alpha : \mathbb{Z}/3\mathbb{Z} \to (0, \infty) be a positive rate vector. For every s>0s > 0, the scaled total plane energy is differentiable at ss with derivative (11): ddt|t=sJ(d,tα)=Θ(d,α,s),\frac{d}{dt}\Big|_{t=s} J(d, t\alpha) = \Theta(d, \alpha, s),(11)✦ and for every 0<a≤b0 < a \le b the total plane time derivative satisfies the integrability condition (12): Θ(d,α,⋅)∈L1([a,b]).\Theta(d, \alpha, \cdot) \in L^1([a, b]).(12)✦ When s>0s > 0, the general plane dissipation bound (13) holds: 2Θ(d,α,s)≤∑w∈ℤ/3ℤ(−αw+2max⁡{0,12∑v∈ℤ/3ℤαv−αw})∫ΠdVw,w(p,sα)dπd(p).2\Theta(d, \alpha, s) \le \sum_{w \in \mathbb{Z}/3\mathbb{Z}} \Bigl(-\alpha_w + \sqrt{2}\max\Bigl\{0,\, \frac{1}{2}\sum_{v \in \mathbb{Z}/3\mathbb{Z}} \alpha_v - \alpha_w\Bigr\}\Bigr) \int_{\Pi_d} V_{w,w}(p, s\alpha)\, d\pi_d(p).(13)✦

Independently of these parameters, the strict positivity constant satisfies (14): c*>0.c_* > 0.(14)✦ Given i∈ℤ/3ℤi \in \mathbb{Z}/3\mathbb{Z}, the strict instantaneous dissipation bound (15) holds for all s>0s > 0, and the integrated dissipation bound (16) holds for all 0<a≤b0 < a \le b: c*∑w∈ℤ/3ℤ∫ΠdVw,w(p,sα(i))dπd(p)≤−Θ(d,α(i),s),c_* \sum_{w \in \mathbb{Z}/3\mathbb{Z}} \int_{\Pi_d} V_{w,w}(p, s\alpha^{(i)})\, d\pi_d(p) \le -\Theta(d, \alpha^{(i)}, s),(15)✦ c*∫ab∑w∈ℤ/3ℤ∫ΠdVw,w(p,tα(i))dπd(p)dt≤J(d,aα(i))−J(d,bα(i)).c_* \int_a^b \sum_{w \in \mathbb{Z}/3\mathbb{Z}} \int_{\Pi_d} V_{w,w}(p, t\alpha^{(i)})\, d\pi_d(p)\, dt \le J(d, a\alpha^{(i)}) - J(d, b\alpha^{(i)}).(16)✦

In the paper, the result reads: fixing finite edge arrays and a positive vector α\alpha, evaluating the energies at σi=αis\sigma_i = \alpha_i s, and setting Jv(d,s)=∫ΠdevdπdJ_v(d, s) = \int_{\Pi_d} e_v\, d\pi_d, J(d,s)=∑vJv(d,s)J(d, s) = \sum_v J_v(d, s), and Σα=∑iαi\Sigma_\alpha = \sum_i \alpha_i (with eve_v, VwV_w, and J(d,s)J(d, s) standing for ev(⋅,sα)e_v(\cdot, s\alpha), Vw,w(⋅,sα)V_{w,w}(\cdot, s\alpha), and J(d,sα)J(d, s\alpha)), these integrals are finite, continuously differentiable in s>0s > 0, and twice continuously differentiable in dd, satisfy 2∂sJ(d,s)≤∫Πd∑w[−αw+2max⁡{0,Σα/2−αw}]Vwdπd2\partial_s J(d, s) \le \int_{\Pi_d} \sum_w [-\alpha_w + \sqrt{2}\max\{0, \Sigma_\alpha/2 - \alpha_w\}] V_w\, d\pi_d, and, for each permutation of (1,λ,λ)(1, \lambda, \lambda), satisfy −∂sJ(d,s)≥c*∫Πd∑wVwdπd-\partial_s J(d, s) \ge c_* \int_{\Pi_d} \sum_w V_w\, d\pi_d with c*=12min⁡{1−352,1110−122}>0c_* = \frac{1}{2}\min\{1 - \frac{3}{5}\sqrt{2}, \frac{11}{10} - \frac{1}{2}\sqrt{2}\} > 0. The theorem above answers this formulation for fixed h>0h > 0 and uu by expressing the derivatives in dd and ss via the variation functionals and Θ\Theta, specializing λ\lambda to 11/1011/10 through α(i)\alpha^{(i)}, and integrating the strict inequality over [a,b][a, b].

What Is Checked

The claims are stated for a finite index set ι\iota, a family AA of complex ι×ι\iota \times \iota matrices indexed by ℤ/3ℤ\mathbb{Z}/3\mathbb{Z}, a step size h>0h > 0, node coordinates u:ι→ℝu : \iota \to \mathbb{R}, a center sum d∈ℝd \in \mathbb{R}, positive variances σ\sigma or positive rate vectors α\alpha on ℤ/3ℤ\mathbb{Z}/3\mathbb{Z}, and the strict dissipation rates α(i)\alpha^{(i)} and constant c*c_*, built from the plane coordinate chart, Gaussian heat weights, local Gram matrices, matrix energy variations, and plane integrals. For any finite index set κ\kappa, the real trace on ℂκ×κ\mathbb{C}^{\kappa \times \kappa} is additive and homogeneous under real scalar multiplication.