Approximation of functions of finite variation by superpositions of a Sigmoidal function

The aim of this note is to generalize a result of Barron [1] concerning the approximation of functions, which can be expressed in terms of the Fourier transform, by superpositions of a fixed sigmoidal function. In particular, we consider functions of the type h(x) = ∫ℝ^d ƒ (〈t, x〉)dμ(t), where μ is a finite Radon measure on ℝ^d and ƒ : ℝ → ℂ is a continuous function with bounded variation in ℝ We show (Theorem 2.6) that these functions can be approximated in L₂-norm by elements of the set G_n = {Σ_i=0^staggeredn c_ig(〈a_i, x〉 + b_i) : a_iℝ^d, b_i, c_iℝ}, where g is a fixed sigmoidal function, with the error estimated by C/n^1/2, where C is a positive constant depending only on f. The same result holds true (Theorem 2.9) for f : ℝ → ℂ satisfying the Lipschitz condition under an additional assumption that ∫ℝ^d6t6e^d|u(t)| > ∞     

Analytic Extension of Smooth Functions

Let F be a closed proper subset of ?ⁿ and let ?* be a class of ultradifferentiable functions. We give a new proof for the following result of Schmets and Valdivia on analytic modification of smooth functions: for every function ? ∈ ?* (?ⁿ) there is ${\widetilde f} \in {\cal E}_{*}(\rm R ^{n})$ which is real analytic on ?ⁿF and such that ?^a ? ¦ _F = ?^a ? ¦ _F for any a ∈ ?₀ ⁿ. For bounded ultradifferentiable functions ? we can obtain ${\widetilde f}$ by means of a continuous linear operator.     

Order of the length of Boolean functions in the class of exclusive-OR sums of pseudoproducts

An exclusive-OR sum of pseudoproducts (ESPP) is a modufo-2 sum of products of affine (linear) Boolean functions. The length of an ESPP is defined as the number of summands in this sum; the length of a Boolean function in the class of ESPPs is the minimum length of an ESPP representing this function. The Shannon length function L ^ESPP(n) on the set of Boolean functions in the class of ESPPs is considered; it is defined as the maximum length of a Boolean function of n variables in the class of ESPPs. It is proved that L ^ESPP(n) = ? (2 ⁿ /n ²). The quantity L ^ESPP(n) also equals the least number l such that any Boolean function of n variables can be represented as a modulo-2 sum of at most l multiaffine functions.     

Improved algorithms for the continuous tree edge-partition problems and a note on ratio and sorted matrices searches

Let p≥2 be an integer and T be an edge-weighted tree. A cut on an edge of T is a splitting of the edge at some point on it. A p-edge-partition of T is a set of p subtrees induced by p−1 cuts. Given p and T, the max-min continuous tree edge-partition problem is to find a p-edge-partition that maximizes the length of the smallest subtree; and the min-max continuous tree edge-partition problem is to find a p-edge-partition that minimizes the length of the largest subtree. In this paper, O(n²)-time algorithms are proposed for these two problems, improving the previous upper bounds by a factor of log (min{p,n}). Along the way, we solve a problem, named the ratio search problem. Given a positive integer m, a (non-ordered) set B of n non-negative real numbers, a real valued non-increasing function F, and a real number t, the problem is to find the largest number z in {b/a|a∈[1,m],b∈B} such that F(z)≥t. We give an O(n+t_F×(logn+logm))-time algorithm for this problem, where t_F is the time required to evaluate the function value F(z) for any real number z.     

Relating three nonlinearity parameters of vectorial functions and building APN functions from bent functions

We survey the properties of two parameters introduced by C. Ding and the author for quantifying the balancedness of vectorial functions and of their derivatives. We give new results on the distribution of the values of the first parameter when applied to F + L, where F is a fixed function and L ranges over the set of linear functions: we show an upper bound on the nonlinearity of F by means of these values, we determine then the mean of these values and we show that their maximum is a nonlinearity parameter as well, we prove that the variance of these values is directly related to the second parameter. We briefly recall the known constructions of bent vectorial functions and introduce two new classes obtained with Gregor Leander. We show that bent functions can be used to build APN functions by concatenating the outputs of a bent (n, n/2)-function and of some other (n, n/2)-function. We obtain this way a general infinite class of quadratic APN functions. We show that this class contains the APN trinomials and hexanomials introduced in 2008 by L. Budaghyan and the author, and a class of APN functions introduced, in 2008 also, by Bracken et al.; this gives an explanation of the APNness of these functions and allows generalizing them. We also obtain this way the recently found Edel?CPott cubic function. We exhibit a large number of other sub-classes of APN functions. We eventually design with this same method classes of quadratic and non-quadratic differentially 4-uniform functions.     

Tranchage et cône tangent des courants localement normaux

Let T be a locally normal current on an open set Ω of ℝ″ = ℝ x ℝ″⁻¹ and let π: ℝⁿ → ℝ denote the projection π(x₁, x″) = x₁. We define the current 〈T, π, 0〉 (called slice of T at 0 by π) as the limit, as ɛ → 0, of the family ɛ⁻¹TΛπ ^* (ψ(x¹ /ɛ)dx¹), where ψ is a C^∞ function on ℝ with compact support such that ∝ψ(x¹)dx¹ = 1, provided the limit exists and doesn't depend on the choice of ψ. We first prove that the limit lim_R→+∞(h_R)_#T exists, where h_R(x₁,x″) = (Rx¹,x″). We apply this result to the study of the existence of the tangent cone at 0 associated to a locally normal current, and especially associated to a subanalytic chain. We finally give a necessary and sufficient condition relative to T for the existence of the slice 〈T, π, 0〉.     

Characterization of Linear Structures

We study the notionof linear structure of a function defined from F ^mto F ⁿ, and in particular of a Boolean function.We characterize the existence of linear structures by means ofthe Fourier transform of the function. For Boolean functions,this characterization can be stated in a simpler way. Finally,we give some constructions of resilient Boolean functions whichhave no linear structure.     

Periodic behaviour of generalized threshold functions

It is shown that, for a function Δ from {0, 1}ⁿ to {0, 1}ⁿ whose components from a symmetric set of threshold functions the repeated application of Δ, leads either to a fixed point or to a cycle of length two.     

Edge-bipancyclicity of a hypercube with faulty vertices and edges

A bipartite graph G=(V,E) is said to be bipancyclic if it contains a cycle of every even length from 4 to |V|. Furthermore, a bipancyclic G is said to be edge-bipancyclic if every edge of G lies on a cycle of every even length. Let F_v (respectively, F_e) be the set of faulty vertices (respectively, faulty edges) in an n-dimensional hypercube Q_n. In this paper, we show that every edge of Q_n-F_v-F_e lies on a cycle of every even length from 4 to 2ⁿ-2|F_v| even if |F_v|+|F_e|?n-2, where n?3. Since Q_n is bipartite of equal-size partite sets and is regular of vertex-degree n, both the number of faults tolerated and the length of a longest fault-free cycle obtained are worst-case optimal.     

VERTEX-FAULT-TOLERANT CYCLES EMBEDDING ON ENHANCED HYPERCUBE NETWORKS

In this paper, we study the enhanced hypercube, an attractive variant of the hypercube and obtained by adding some complementary edges from a hypercube, and focus on cycles embedding on the enhanced hypercube with faulty vertices. Let Fu be the set of faulty vertices in the n-dimensional enhanced hypercube Qn,k （n ≥ 3, 1 ≤ k 〈≤n - 1）. When IFvl = 2, we showed that Qn,k - Fv contains a fault-free cycle of every even length from 4 to 2n - 4 where n （n ≥ 3） and k have the same parity; and contains a fault-free cycle of every even length from 4 to 2n - 4, simultaneously, contains a cycle of every odd length from n-k ＋ 2 to 2^n-3 where n （≥ 3） and k have the different parity. Furthermore, when ｜Fv｜ = fv ≤ n - 2, we prove that there exists the longest fault-free cycle, which is of even length 2^n - 2fv whether n （n ≥ 3） and k have the same parity or not; and there exists the longest fault-free cycle, which is of odd length 2^n - 2fv ＋ 1 in Qn,k - Fv where n （≥ 3） and k have the different parity.     

Finite Morse Index Implies Finite Ends

We prove that finite Morse index solutions to the Allen-Cahn equation in ℝ² have finitely many ends and linear energy growth. The main tool is a curvature decay estimate on level sets of these finite Morse index solutions, which in turn is reduced to a problem on the uniform second-order regularity of clustering interfaces for the singularly perturbed Allen-Cahn equation. Using an indirect blowup technique, in the spirit of the classical Colding-Minicozzi theory in minimal surfaces, we show that the obstruction to the uniform second-order regularity of clustering interfaces in ℝⁿ is associated to the existence of nontrivial entire solutions to a (finite or infinite) Toda system in ℝ^n–1. For finite Morse index solutions in ℝ², we show that this obstruction does not exist by using information on stable solutions of the Toda system. © 2019 Wiley Periodicals, Inc.     

A note on a boundary value problem for linear differential difference equations of mixed type

The statistical theory of linear chain molecules often has rational functions of a few analytic functions as the form of the generating function for the partition function Z⁽ⁿ⁾ of the linear chain molecule, n units long. For applications, it is necessary to know the limiting form of Z⁽ⁿ⁾ as n → ∞. Renewal theory from probability theory is applied to determine this limiting form in important cases.     

Interpolating and orthogonal polynomials on fractals

A special class of closed subsetsF ofR ⁿ , referred to as sets preserving Markov's inequality, are considered. Typically,F may be a fractal such as the Cantor set or von Koch's curve, butF may also be a closed Lipschitz domain. We investigate interpolation to smooth functions onF where the points of interpolation belong toF. We also consider orthogonal polynomials onF∩B, whereB is a ball with center inF, and their relation to spaces of smooth functions onF.     

The eigenvalue problem for linear and affine iterated function systems

The eigenvalue problem for a linear function L centers on solving the eigen-equation . This paper generalizes the eigenvalue problem from a single linear function to an iterated function system F consisting of possibly an infinite number of linear or affine functions. The eigen-equation becomes F(X)=λX, where λ>0 is real, X is a compact set, and F(X)=?_f∈Ff(X). The main result is that an irreducible, linear iterated function system F has a unique eigenvalue λ equal to the joint spectral radius of the functions in F and a corresponding eigenset S that is centrally symmetric, star-shaped, and full dimensional. Results of Barabanov and of Dranisnikov-Konyagin-Protasov on the joint spectral radius follow as corollaries.     

On the length of Boolean functions in the class of exclusive-OR sums of pseudoproducts

Representations of Boolean functions by exclusive-OR sums (modulo 2) of pseudoproducts is studied. An ExOR-sum of pseudoproducts (ESPP) is the sum modulo 2 of products of affine (linear) Boolean functions. The length of an ESPP is defined as the number of summands in this form, and the length of a Boolean function in the class of ESPPs is defined as the minimum length of an ESPP representing this function. The Shannon function L ^ESPP(n) of the length of Boolean functions in the class of ESPPs is considered, which equals the maximum length of a Boolean function of n variables in this class. Lower and upper bounds for the Shannon function L ^ESPP(n) are found. The upper bound is proved by using an algorithm which can be applied to construct representations by ExOR-sums of pseudoproducts for particular Boolean functions.     

Orthogonal exponentials on the ball

We show that there does not exist an infinite sequence of vectors λ_n in ℝ^d, d > 1, such that the corresponding exponentials eⁱ〈λ_n,x〉, x ∈ ℝ^d, when considered on the unit ball B in ℝ^d, are pairwise orthogonal in L²(B) (B being endowed with Lebesgue measure). The weaker result that L²(B) does not have an infinite orthogonal base of exponentials has recently been established by A. Iosevich, N. Katz, and S. Pedersen in [2]. For d = 2 the present result was announced in the author's 1974 paper [1].     

On the solvability of the jump problem in clifford analysis

Let Ω be a bounded open and oriented connected subset of ? ⁿ which has a compact topological boundary Γ, let C be the Dirac operator in ? ⁿ , and let ?_0,n be the Clifford algebra constructed over the quadratic space ? ⁿ . An ?_0,n -valued smooth function f : Ω → ?_0,n in Ω is called monogenic in Ω if Df = 0 in Ω. The aim of this paper is to present the most general condition on Γ obtained so far for which a Hölder continuous function f can be decomposed as F ⁺ ? F ^? = f on Γ, where the components F ^± are extendable to monogenic functions in Ω^± with Ω⁺ := Ω, and Ω^? := ? ⁿ \ (Ω ? Γ), respectively.     

Obstacle Problems with Linear Growth: Hölder Regularity for the Dual Solution

For a strictly convex integrand f : ℝⁿ → ℝ with linear growth we discuss the variational problem among mappings u : ℝⁿ ⊃ Ω → ℝ of Sobolev class W¹₁ with zero trace satisfying in addition u ≥ ψ for a given function ψ such that ψ|_∂Ω < 0. We introduce a natural dual problem which admits a unique maximizer σ. In further sections the smoothness of σ is investigated using a special J-minimizing sequence with limit u* ∈ C^1,α (Ω) for which the duality relation holds.     

The crooked property

Crooked permutations were introduced twenty years ago to construct interesting objects in graph theory. These functions, over ${\mathbb{F}}_{2^n}$ with odd n, are such that their derivatives have as image set a complement of a hyperplane. The field of applications was extended later, in particular to cryptography. However binary crooked functions are rare. It is still unknown if non quadratic crooked functions do exist. We extend the concept and propose to study the crooked property for any characteristic. A function F, from ${\mathbb{F}}_{p^n}$ to itself, satisfies this property if all its derivatives have as image set an affine subspace. We show that the partially-bent vectorial functions and the functions satisfying the crooked property are strongly related. We later focus on the components of these functions, establishing that the existence of linear structures is here decisive. We then propose a symbolic approach to identify the linear structures. We claim that this problem consists in solving a system of linear equations, and can often be seen as a combinatorial problem.     

Sufficient Conditions and Duality for a Class of Multiobjective Fractional Programming Problems with Higher-Order (F,α,ρ,d)-Convexity

In this paper, a class of multiobjective fractional programming problems (denoted by (MFP)) is considered. First, the concept of higher-order (F,α,ρ,d)-convexity of a function f:C→R with respect to the differentiable function φ:R ⁿ ×R ⁿ →R is introduced, where C is an open convex set in R ⁿ and α:C×C→R ₊?{0} is a positive value function. And an important property, which the ratio of higher-order (F,α,ρ,d)-convex functions is also higher-order (F,α ^′,ρ ^′,d ^′)-convex, is proved. Under the higher-order (F,α,ρ,d)-convexity assumptions, an alternative theorem is also given. Then, some sufficient conditions characterizing properly (or weakly) efficient solutions of (MFP) are obtained from the above property and alternative theorem. Finally, a class of dual problems is formulated and appropriate duality theorems are proved.