The Setting
Consider a finite index set , three complex matrices assigned to the edges of a triangle, real node coordinates , and a step size . Weighting the nodes by Gaussians centered at with positive variances produces a weighted matrix array whose vertex star energies depend on and . Integrating these energies over the constant-sum center plane raises two questions: given a positive rate vector , how the resulting plane integrals vary under shifts of and under linear variance scaling , and whether the total plane energy dissipates at a strictly positive rate as increases.
Matrix Energy and Hessian Variations
Let be a finite index set. The real trace of a complex square matrix is the real number given by the real part of the trace of .✦ The real Hilbert-Schmidt inner product of is the real number , where is the conjugate transpose of .✦ The matrix energy of with square root is the real number (which equals when is positive semidefinite)✦, and the matrix energy gradient of is the continuous real-linear functional on mapping to .
The Sylvester operator of is the linear map on given by ✦, and the Sylvester penalty of and is the real number .✦ For , the Sylvester variational functional is the real-valued function on mapping to ✦, and the root variational cost is the real number equal to times the supremum of over .✦ For , the Hessian quadratic cost and polarized Hessian pairing are the real-valued quadratic and bilinear forms defined by (1): (1)
The triangle block space over a finite index set is the Cartesian product of the triangle vertex set under cyclic addition ( and modulo ) with .✦ Given a family in , the triangle matrix blocks form the matrix of blocks whose block is when , equals when , and equals when .✦ Block matrix assembly flattens any matrix of blocks into a matrix on ✦, and the triangle matrix on is the matrix obtained by assembling the triangle matrix blocks of .✦
For each , the vertex block projection is the diagonal matrix on with entry at when and otherwise.✦ The vertex star row of at is the matrix on that isolates the block row at ✦, and the vertex star Gram matrix of at is the matrix on .✦
The weighted matrix array of a family in and real node weights with is the family in indexed by with entries .✦ The vertex root matrix of real node weights with is the diagonal matrix on with entry at .✦ For , the vertex star base row of and at is the matrix ✦, the vertex star base column of and at is the matrix ✦, and the fixed base matrix of and at is the matrix on equal to when and when .✦ The weighted Gram matrix of a matrix and a real weight function is the matrix .✦
Gaussian Heat Weights and Plane Integrals
The variance- Gaussian density at argument for is the real number .✦ Given a point , center , variance , and step size , the Gaussian score is the real number ✦ and the Gaussian weight is the real number .✦ Fixing in , a step size , and node coordinates , for any centers and variances the vertex heat weights form the family of functions defined by .✦ The heat matrix array is the weighted matrix array of with vertex heat weights ✦, and the heat matrix energy at vertex is the real number .✦
Given a step size , a finite index set , coordinates , a group assignment into an index set , and , the grouped heat weight is the function given by .✦ For a group , the grouped first variation weight and grouped second variation weight are the functions that vanish when and when are defined by (2): (2) Specializing to and with and , write for . The local Gram matrix , local Gram first variation , and local Gram second variation are the weighted Gram matrices of with , , and , respectively. From these matrices evaluated at , the heat energy gradient , heat Hessian cost , pure heat second variation , heat variance term , and mixed heat Hessian pairing for are the real numbers given by (3): (3)
The plane coordinate chart associated with a vertex and sum level is the map sending to the triple in equal to at and to at the remaining two vertices in increasing order.✦ The plane integral at level of a real-valued function on is the Lebesgue integral . A tuple consisting of a matrix family in , a step size , node coordinates , a sum level , and variances is heat-plane integrable if for all , the compositions with of , , , , and are Lebesgue integrable on , and for all pairwise distinct , the composition of with is Lebesgue integrable on .✦
Given a sum level , variances , rate vector , and time , the total plane heat energy and total plane time derivative are the real numbers defined by (4): (4) For each , the strict dissipation rate is the vector with for and for , and the strict positivity constant is the real number given by (5): (5)✦
The Claim
Theorem 1. Fix a finite index set , a family of complex matrices, a step size , node coordinates , and a center sum . For every positive variance vector , the plane integrability condition (6) holds: (6)✦ Under the same positivity assumption on , for all the plane energy and first variation integrals are differentiable at with derivatives (7) and (8): (7)✦ (8)✦ When are pairwise distinct, the first variation integral is also differentiable at with mixed derivative (9), and for any the pure and mixed plane integrals coincide as in (10): (9)✦ (10)✦
Let be a positive rate vector. For every , the scaled total plane energy is differentiable at with derivative (11): (11)✦ and for every the total plane time derivative satisfies the integrability condition (12): (12)✦ When , the general plane dissipation bound (13) holds: (13)✦
Independently of these parameters, the strict positivity constant satisfies (14): (14)✦ Given , the strict instantaneous dissipation bound (15) holds for all , and the integrated dissipation bound (16) holds for all : (15)✦ (16)✦
In the paper, the result reads: fixing finite edge arrays and a positive vector , evaluating the energies at , and setting , , and (with , , and standing for , , and ), these integrals are finite, continuously differentiable in , and twice continuously differentiable in , satisfy , and, for each permutation of , satisfy with . The theorem above answers this formulation for fixed and by expressing the derivatives in and via the variation functionals and , specializing to through , and integrating the strict inequality over .
What Is Checked
The claims are stated for a finite index set , a family of complex matrices indexed by , a step size , node coordinates , a center sum , positive variances or positive rate vectors on , and the strict dissipation rates and constant , built from the plane coordinate chart, Gaussian heat weights, local Gram matrices, matrix energy variations, and plane integrals. For any finite index set , the real trace on is additive and homogeneous under real scalar multiplication.