Searcharxiv⌕ Search

arXiv subjects

Zsolt Páles

Publications and source records attributed to Zsolt Páles.

At least 19 recordsLinked to original sources

Expected loss of quasi-arithmetic means of exchangeable random variables

The purpose of this paper is to establish certain relationships between the theories of means (aggregation functions) and loss functions. Namely, we derive sufficient conditions under which the sequence of the expected losses of a quasi-arithmetic mean of the first $n$ members from a sequence of exchangeable random variables is (strictly) decreasing with respect to $n$. Some new properties of functions which are strongly convex with respect to a nonnegative even error function, such as a Jensen-type inequality, are described. Additionally, non-trivial new examples for such functions are presented as well.

math.PR↗

Monotone representation and measurability of generalized $ψ$-estimators

We investigate the monotone representation and measurability of generalized $ψ$-estimators introduced by the authors in 2022. Our first main result, applying the unique existence of a generalized $ψ$-estimator, allows us to construct this estimator in terms of a function $ψ$, which is decreasing in its second variable. We then interpret this result as a bridge from a nonconvex optimization problem to a convex one. Further, supposing that the underlying measurable space (sample space) has a measurable diagonal and some additional assumptions on $ψ$, we show that the measurability of a generalized $ψ$-estimator is equivalent to the measurability of the corresponding function $ψ$ in its first variable.

math.ST↗

On subadditive quasi-arithmetic means

Let $f\colon \mathbb{R}_+\to\mathbb{R}$ be a continuous and strictly monotone function. In the main result of this paper, we show that, for a fixed $n\geq 2$, the $n$-variable mean $\mathscr{A}_f \colon \mathbb{R}_+^n \to \mathbb{R}_+$ defined by $$ \mathscr{A}_f(x_1,\dots,x_n):=f^{-1} \bigg( \frac{f(x_1)+\cdots+f(x_n)}n \bigg) $$ is subadditive if and only if $f$ is differentiable with a continuously semi-differentiable and nonvanishing first derivative, and there exists an $α\in[0,\infty]$ such that $f''_+:=(f')'_+$ is positive on $(0,α)$ and $f''_+=0$ on $[α,\infty)$, furthermore, $\frac{f'}{f''_+}$ is increasing and superadditive on $(0,α)$.

math.CA↗

Axiomatic characterisation of generalized $ψ$-estimators

We give axiomatic characterisations of generalized $ψ$-estimators and (usual) $ψ$-estimators (also called $Z$-estimators), respectively. The key properties of estimators that come into play in the characterisation theorems are the symmetry, the (strong) internality and the asymptotic idempotency. In the proofs, a separation theorem for Abelian subsemigroups plays a crucial role.

math.ST↗

Hölder- and Minkowski-type inequalities for generalized quasi-arithmetic means

The purpose of this paper is to establish several necessary and sufficient conditions to ensure the validity of a general functional inequality in terms of generalized quasi-arithmetic means. In particular cases, we consider Hölder-, Minkowski-, and Jensen-type inequalities. Generalized quasi-arithmetic means are defined by taking strictly monotone generating functions instead of strictly monotone and continuous ones.

math.CA↗

Jensen convex functions and doubly stochastic matrices

Given an nxn doubly stochastic matrix P satisfying an appropriate condition of linear algebraic-type, and a function f defined on a nonempty interval, we show that the validity of a convexity-type functional inequality for f in terms P implies that f is Jensen convex. We also prove that if f is convex, then the functional inequality in question holds for all doubly stochastic matrices of any order. The particular case when the doubly stochastic matrix is a circulant one is also considered.

math.CA↗

A convexity-type functional inequality with infinite convex combinations

Given a function $f$ defined on a nonempty and convex subset of the $d$-dimensional Euclidean space, we prove that if $f$ is bounded from below and it satisfies a convexity-type functional inequality with infinite convex combinations, then $f$ has to be convex. We also give alternative proofs of a generalization of some known results on convexity with infinite convex combinations due to Daróczy and Páles (1987) and Pavić (2019) using a probabilistic version of Jensen inequality.

math.CA↗

Determining classes for generalized $ψ$-estimators

We prove that the values of a generalized $ψ$-estimator (introduced by Barczy and Páles in 2025) on samples of arbitrary length but having only two different observations uniquely determine the values of the estimator on any sample of arbitrary length without any restriction on the number of different observations. In other words, samples of arbitrary length but having only two different observations form a determining class for generalized $ψ$-estimators. We also obtain a similar statement for the comparison of generalized $ψ$-estimators using comparative functions, and, as a corollary of this result, we derive the Schweitzer's inequality (also called Kantorovich's inequality).

math.ST↗

Orderings on measures induced by higher-order monotone functions

The main aim of this paper is to study the functional inequality \begin{equation*} \int_{[0,1]}f\bigl((1-t)x+ty\bigr)dμ(t)\geq 0, \qquad x,y\in I \mbox{ with } x<y, \end{equation*} for a continuous unknown function $f:I\to{\mathbb R}$, where $I$ is a nonempty open real interval and $μ$ is a signed and bounded Borel measure on $[0,1]$. We derive necessary as well as sufficient conditions for its validity in terms of higher-order monotonicity properties of $f$. Using the results so obtained we can derive sufficient conditions under which the inequality $${\mathbb E} f(X)\leq {\mathbb E} f(Y)$$ is satisfied by all functions which are simultaneously: $k_1$-increasing (or decreasing), $k_2$-increasing (or decreasing), \dots , $k_l$-increasing (or decreasing) for given nonnegative integers $k_1,\dots,k_l.$ This extends several well-known results on stochastic ordering. A necessary condition for the $(n,n+1,\dots,m)$-increasing ordering is also presented.

math.CA↗

Existence and uniqueness of weighted generalized $ψ$-estimators

We introduce the notions of generalized and weighted generalized $ψ$-estimators as unique points of sign change of some appropriate functions, and we give necessary as well as sufficient conditions for their existence. We also derive a set of sufficient conditions under which the so-called $ψ$-expectation function has a unique point of sign change. We present several examples from statistical estimation theory, where our results are well-applicable. For example, we consider the cases of empirical quantiles, empirical expectiles, some $ψ$-estimators that are important in robust statistics, and some examples from maximum likelihood theory as well. Further, we introduce Bajraktarević-type (in particular, quasi-arithmetic-type) $ψ$-estimators. Our results specialized to $ψ$-estimators with a function $ψ$ being continuous in its second variable provide new results for (usual) $ψ$-estimators (also called Z-estimators).

math.ST↗

Equality and comparison of generalized quasiarithmetic means

The purpose of this paper is to extend the definition of quasiarithmetic means by taking a strictly monotone generating function instead of a strictly monotone and continuous one. We establish the properties of such means and compare them to the analogous properties of standard quasiarithmetic means. The comparability and equality problems of generalized quasiarithmetic are also solved. We also provide an example of a mean which, depending on the underlying interval or on the number of variables, could be or could not be represented as a generalized quasiarithmetic mean.

math.CA↗

Hermite-type interpolation in terms of exponential polynomials

Motivated by classical results of approximation theory, we define an Hermite-type interpolation in terms of $n$-dimensional subspaces of the space of $n$ times continuously differentiable functions. In the main result of this paper, we establish an error term in integral form for this interpolation in the case when the $n$-dimensional subspace is the kernel of an $n$th order linear differential operator with constant coefficients. Several corollaries are deduced illustrating the applicability of this result.

math.CA↗

Estimates for approximately Jensen convex functions

In this paper functions $f:D\to\mathbb{R}$ satisfying the inequality \[ f\Big(\frac{x+y}{2}\Big)\leq\frac12f(x)+\frac12f(y) +φ\Big(\frac{x-y}{2}\Big) \qquad(x,y\in D) \] are studied, where $D$ is a nonempty convex subset of a real linear space $X$ and $φ:\{\frac12(x-y) : x,y \in D\}\to\mathbb{R}$ is a so-called error function. In this situation $f$ is said to be $φ$-Jensen convex. The main results show that for all $φ$-Jensen convex function $f:D\to\mathbb{R}$, for all rational $λ\in[0,1]$ and $x,y\in D$, the following inequality holds \[ f(λx+(1-λ)y) \leq λf(x)+(1-λ)f(y)+\sum_{k=0}^\infty \frac{1}{2^k}φ\big(\mbox{dist}(2^kλ,\mathbb{Z})\cdot(x-y)\big). \] The infinite series on the right hand side is always convergent, moreover, for all rational $λ\in[0,1]$, it can be evaluated as a finite sum.

math.CA↗

Local and global Hölder- and Minkowski-type inequalities for nonsymmetric generalized Bajraktarević means

The aim of this paper is to investigate inequalities that are analogous to the Minkowski and Hölder inequalities by replacing the addition and the multiplication by a more general operation, and instead of using power means, generalized Bajraktarević means are considered, in particular, Gini means. A further aim is to introduce the concept of local and global validity of such inequalities and to characterize them in both senses.

math.CA↗

Estimates of linear expressions through factorization

The aim of this paper is to establish various factorization results and then to derive estimates for linear functionals through the use of a generalized Taylor theorem. Additionally, several error bounds are established including applications to the trapezoidal rule as well as to a Simpson formula-type rule.

math.CA↗

On the equality of generalized Bajraktarević means under first-order differentiability assumptions

In this paper we consider the equality problem of generalized Bajraktarević means, i.e., we are going to solve the functional equation \begin{equation}\label{E0}\tag{*} f^{(-1)}\bigg(\frac{p_1(x_1)f(x_1)+\dots+p_n(x_n)f(x_n)}{p_1(x_1)+\dots+p_n(x_n)}\bigg)=g^{(-1)}\bigg(\frac{q_1(x_1)g(x_1)+\dots+q_n(x_n)g(x_n)}{q_1(x_1)+\dots+q_n(x_n)}\bigg), \end{equation} which holds for all $x=(x_1,\dots,x_n)\in I^n$, where $n\geq 2$, $I$ is a nonempty open real interval, the unknown functions $f,g:I\to\mathbb{R}$ are strictly monotone, $f^{(-1)}$ and $g^{(-1)}$ denote their generalized left inverses, respectively, and the vector-valued weight functions $p=(p_1,\dots,p_n):I\to\mathbb{R}_{+}^n$ and $q=(q_1,\dots,q_n):I\to\mathbb{R}_{+}^n$ are also unknown. This equality problem in the symmetric two-variable case (i.e., when $n=2$ and $p_1=p_2$, $q_1=q_2$) was solved under sixth-order regularity assumptions by Losonczi in 1999. The authors of this paper improved this result in 2023 by reaching the same conclusion assuming only first-order differentiability. In the nonsymmetric case, assuming third-order differentiability of $f$, $g$ and the first-order differentiability of at least three of the functions $p_1,\dots,p_n$, Grünwald and Páles proved that \eq{0} holds if and only if there exist four constants $a,b,c,d\in\mathbb{R}$ with $ad\neq bc$ such that $$ cf+d>0,\qquad g=\frac{af+b}{cf+d},\qquad\mbox{and}\qquad q_\ell=(cf+d)p_\ell\qquad (\ell\in\{1,\dots,n\}). $$ The main goal of this paper is to establish the same conclusion under first-order differentiability.

math.CA↗

Comparison and equality of generalized $ψ$-estimators

We solve the comparison problem for generalized $ψ$-estimators introduced in Barczy and Páles (2022). Namely, we derive several necessary and sufficient conditions under which a generalized $ψ$-estimator less than or equal to another $ψ$-estimator for any sample. We also solve the corresponding equality problem for generalized $ψ$-estimators. For applications, we solve the two problems in question for Bajraktarević-type- and quasi-arithmetic-type estimators. We also apply our results for some known statistical estimators such as for empirical expectiles and Mathieu-type estimators and for solutions of likelihood equations in case of normal, a Beta-type, Gamma, Lomax (Pareto type II), lognormal and Laplace distributions.

math.ST↗

Basic properties of generalized $ψ$-estimators

We establish several properties of (weighted) generalized $ψ$-estimators introduced by Barczy and Páles in 2022: mean-type, monotonicity and sensitivity properties, bisymmetry-type inequality and some asymptotic and continuity properties as well. We also illustrate these properties by providing several examples including statistical ones as well.

math.ST↗