The Setting
Fix , an even natural number , vectors of and real numbers , the widths. With the linear functional of a vector , for in , and , the soft slab potential and its slab summands are (1) and the soft slab law is the measure (2)✦
The question is whether is the moment measure of a smooth function on the whole space: whether some function on has a gradient map that carries the source measure , with density the weight against Lebesgue measure, onto , written , and that is a change of coordinates of . The variable of is called the source coordinate, where lives, and the target coordinate, where lives.
Derivatives and Measures
The directional derivative of a function on a real normed space along a vector is the function , where is the Fréchet derivative of at .✦ The partial derivative of a function on is ✦, for the standard basis vector with coordinate at and at every other index✦, and its gradient is the vector .✦ The Hessian of at is the matrix with entry in row and column .✦
A Euclidean space is a finite-dimensional real inner product space, with norm . Euclidean -space is with the inner product , and its standard orthonormal basis consists of the unit coordinate vectors. The coordinate isomorphism is the continuous linear isomorphism that keeps every coordinate.✦
The gradient of a differentiable function on at is the vector with for every , and the second derivative of at is the derivative of at , a linear map of into itself. For an orthonormal basis of , the Hessian matrix is the matrix of in , with entry in row and column .✦
For real matrices, means that is symmetric and positive definite, and that is symmetric and positive semidefinite; is the identity matrix. The symmetric second difference of at with step is (3)✦
The normalized measure of a function on a Euclidean space is the measure , whose density is the weight divided by its integral .✦
The Claim
Theorem 1. Let and even, and let , and be as in (1) and (2). On each of the spaces and there is a function whose gradient is a homeomorphism of the space onto itself with inverse and which satisfies, on , at every point , (4)✦ and, on , at every point , (5)✦ Moreover, let be a Euclidean space and an orthonormal basis of .
If , , and is a continuous even function on with integrable and for all and , then there are a convex even function on whose gradient maps onto , a constant and a real number such that, for all , and , (6)✦
Let be a convex function on whose gradient is Lipschitz and maps onto , and a function on with integrable and . If is convex and , then (7)✦ If is and is convex and , then (8)✦ If is and is continuous, then at every point (9)✦
If is a function on with at every , and is a real number with for all and , then at every point (10)✦
If is a map of onto with for every and every nonzero , then (11)✦
In the paper, the result reads: for , and an even with , and , there is an even such that is a probability measure, is a diffeomorphism of onto with , and , and moreover for some and some real .
What Is Checked
The claim is stated over the soft slab potential and law (1), (2), the normalized measures , the source measures , the gradient and Hessian built from partial derivatives on , the Hessian matrix in an orthonormal basis, the second difference (3) and the coordinate isomorphism .