WorksMathematics
Explicit Measurable Counterexamples to Finite-Difference Operators
Abstract
Let be a finite-difference operator, where the coefficients may be complex and the nodes are real. We study the asymptotic implication for Lebesgue measurable functions. In 2017, Ash, Catoiu, and Csörnyei \cite{ACC} showed that this implication holds for every measurable function precisely for operators belonging to a very restricted class of finite differences. To establish that every operator outside the exceptional class admits a counterexample, Ash, Catoiu, and Csörnyei analyze an infinite system of linear equations. Their argument is therefore existential in nature and does not provide an explicit description of the counterexamples. By contrast, we give a substantially simpler and elementary construction based on zeros of finite Laurent polynomials and multiplicative characters on finitely generated multiplicative subgroups. This provides a direct recipe for constructing measurable functions such that while, in fact,
Let , let , and let be distinct. For a function defined on , we consider the finite-difference operator
The numbers are called the nodes, and the non-zero numbers are the corresponding coefficients.
This paper concerns the implication
for measurable functions . In the real-valued setting, Ash, Catoiu, and Csörnyei [1] completely characterized the finite differences for which this ^2020 Mathematics Subject Classification. Primary 39B05; Secondary 26A24, 28A05.
Keywords. Finite differences, measurable functions, functional equations, Laurent polynomials, multiplicative groups. implication holds for every measurable function . Surprisingly, apart from multiplication by a nonzero constant and a rescaling of the variable, the only exceptional difference is
The difficult part of their result is the converse. Namely, if a finite difference is not of this exceptional form, then there exists a measurable function such that
while
The proof in [1] is existential in nature: the counterexample is obtained from the solvability of an infinite system of linear equations. In contrast, our proof gives an explicit construction. Crucially, our framework dramatically simplifies this problem: rather than managing vast infinite systems of equations, the counterexamples can be written down almost immediately by inspection. They are supported on countable multiplicative subgroups and are described by multiplicative characters whose parameters are obtained from zeros of finite Laurent polynomials.
To demonstrate the remarkable ease and algorithmic nature of this approach, we show how a concrete counterexample can be effortlessly constructed from scratch using a simple algebraic recipe. Consider the operator
The corresponding Laurent polynomial is
In order to find a nonzero root of , we first choose an integer point that does not belong to any of the hyperplanes determined by the exponent vectors. In this case, the relevant hyperplanes are
The integer point avoids all three of them. Consequently,
Setting , we see that is a nonzero root of .
Now choose the countable multiplicative subgroup
generated by , together with the auxiliary prime . Since , , and are multiplicatively independent, every element of has a unique representation of this form. The auxiliary generator is used only to provide a sequence ; the definition of does not depend on the exponent . Define by setting if , and . A direct calculation yields
for every . Hence
for every . Since is countable, is trivially Lebesgue measurable. Moreover, because
it follows that as , completing the counterexample.
Lemma 1. Let
be a Laurent polynomial with non-zero coefficients and at least two distinct exponent vectors,
Then there exists such that .
Proof. By combining like terms, we may assume that all of the exponent vectors of are distinct. For each pair of exponent vectors,
consider the hyperplane in defined by
Since a finite union of hyperplanes cannot cover , nor can it cover the lattice , there exists an integer lattice point that does not lie on any of the hyperplanes. Define
Then is a nonconstant Laurent polynomial in one variable. By choice of all of the exponents of are distinct. Multiplying by a suitable power of , we obtain a nonconstant ordinary polynomial
Since has at least two terms, it is not a monomial. Hence, there exists such that . Therefore , and consequently
Since , all coordinates are nonzero.
:::`
Theorem 2. Let
be a finite difference with nonzero coefficients and distinct nodes . Suppose that there are at least two nonzero nodes and that cannot be scaled to be of the form
Then there exists a Lebesgue measurable function such that as and
If the coefficients of are real, then either the real part or the imaginary part of provides a real-valued function with the same properties.
Proof. If one of the nodes is zero, say , define
Then as , and
for every . We may therefore assume that none of the nodes is zero.
Let be the multiplicative subgroup of generated by
Since is finitely generated, there exist positive real numbers such that every element of can be written uniquely in the form
For each node , write
Define the Laurent polynomials
and
Suppose first that, after combining like terms and eliminating terms with zero coefficients, is not a monomial. By Lemma 1, there exists
such that
Choose such that
are linearly independent over , and let
For each element of all the integers are unique.
Define by
and
For every and every node , we have
Indeed, this follows directly from the definition when ; when , both sides vanish because and hence . Consequently,
for every .
Suppose next that, after combining like terms and eliminating terms with zero coefficients, is not a monomial. Choose
such that
Using the same group , define for , and define
Then
for every . It follows that
for every .
In either case, along the sequence
we have and
Thus as . Moreover, is countable, so , being zero outside the countable set , is Lebesgue measurable on .
It remains to consider the case in which both and reduce to monomials or to the zero polynomial.
Suppose first that is a monomial. Then there is a node such that, if is also a node, the sum of the corresponding coefficients is nonzero. Moreover, for every node , the node is also present, and the coefficients of and are negatives of one another.
If is the zero polynomial, then every node occurs together with its opposite, and the coefficients of and are negatives of one another.
Similarly, if is a monomial, then there is a node such that, if is also a node, the difference of the corresponding coefficients is nonzero. Moreover, for every node , the node is also present, and the coefficients of and are the same.
The two polynomials cannot both be zero, since this would force every coefficient to be zero. If is the zero polynomial, then
Define
Then for every , and as .
If is the zero polynomial, then
In this case the constant function satisfies for every .
Finally, suppose that both and are monomials. Then, after relabeling the two pairs if necessary,
After multiplying the difference by a nonzero constant and rescaling the variable , this reduces precisely to the exceptional form
contrary to the hypothesis. □
The characterization of measurable finite differences obtained by Ash, Catoiu, and Csörnyei becomes substantially richer under the additional assumption that as . For example, the finite difference
has the property that
imply
This and related applications to Riemann derivatives are established in [2].
References
References
- [1]J. M. Ash, S. Catoiu, and M. Csörnyei, Generalized vs. ordinary differentiation, Proc. Amer. Math. Soc. 145 (2017), 1553–1565.DOI
- [2]H. Fejzić, The Marcinkiewicz–Zygmund Property for Riemann Differences with Geometric Nodes, Real Anal. Exchange 50(2) (2025), 445–468.arxiv.org/abs/2507.11463