Abstract
Let \(p\) and \(q\) be distinct primes, and consider the expression \(S_{p,q}\left(\mathfrak{z}\right)\) defined by the formal series \(\sum_{n=0}^{\infty}q^{\#_{1}\left(\left[\mathfrak{z}\right]_{2^{n}}\right)}/p^{n}\), where \(\mathfrak{z}\) is a \(2\)-adic integer variable, \(\left[\mathfrak{z}\right]_{2^{n}}\) is the integer in \(\left\{ 0,\ldots,2^{n}-1\right\} \) congruent to \(\mathfrak{z}\) mod \(2^{n}\), and where, for any integer \(m\geq0\), \(\#_{1}\left(m\right)\) is the number of \(1\)s in the binary expansion of \(m\). When \(\mathfrak{z}\in\left\{ 0,1,2,\ldots\right\} \), \(S_{p,q}\left(\mathfrak{z}\right)\) reduces to a geometric series with \(1/p\) as its common ratio. This series converges in the topology of \(\mathbb{R}\), and its sum is a rational number which is also a \(q\)-adic integer. For \(\mathfrak{z}\in\mathbb{Z}_{2}\backslash\left\{ 0,1,2,\ldots\right\} \), \(\#_{1}\left(\left[\mathfrak{z}\right]_{2^{n}}\right)\rightarrow\infty\) as \(n\rightarrow\infty\), and so \(S_{p,q}\left(\mathfrak{z}\right)\) converges in the \(q\)-adic topology to a \(q\)-adic integer. In this way, we can define a “\(\left(2,q\right)\)-adic function” \(X_{p,q}:\mathbb{Z}_{2}\rightarrow\mathbb{Z}_{q}\) by the sum of \(S_{p,q}\left(\mathfrak{z}\right)\) in \(\mathbb{R}\) for \(\mathfrak{z}\in\mathbb{Z}_{2}\backslash\left\{ 0,1,2,\ldots\right\} \) and as the sum of \(S_{p,q}\left(\mathfrak{z}\right)\) in \(\mathbb{Z}_{q}\) for all other \(\mathfrak{z}\). Thus, while \(X_{p,q}\left(\mathfrak{z}\right)\) is well-defined as a \(q\)-adic integer for all \(\mathfrak{z}\in\mathbb{Z}_{2}\), the topology of convergence used to sum the series representation of \(X_{p,q}\left(\mathfrak{z}\right)\) to compute its value at any given \(\mathfrak{z}\) depends on the value of \(\mathfrak{z}\). This represents an entirely new type of point-wise convergence, one where the topology in which the limit of a sequence of functions \(\left\{ f_{n}\right\} _{n\geq1}\) is evaluated depends on the point at which the sequence is evaluated. In a manner comparable to the adèle ring of a number field, functions defined by \(\mathcal{F}\)-series require considering different metric completions of an underlying field in order to be properly understood. This paper catalogues a variety of examples of this new, unstudied convergence phenomenon, and presents the concept of a “frame”, a rigorous formalism for defining and studying \(\mathcal{F}\)-series and their peculiar modes of convergence.
Notes
It is not difficult to show that this convergence is strictly point-wise. We can make the \(3\)-adic convergence of \(S_{2,3}\left(\mathfrak{z}\right)\) occur arbitrarily slowly by choosing a \(\mathfrak{z}\in\mathbb{Z}_{2}^{\prime}\) whose \(1\)s digits are separated by sufficiently large number of \(0\)s.
Number theorists may be somewhat uncomfortable with this notation. Traditionally, given a prime \(p\), the symbol \(q\) is used to represent a number of the form \(p^{\nu}\) where \(\nu\in\mathbb{N}_{1}\). When working with a given prime \(p\), the symbol \(\ell\) is preferred to denote a prime different from \(p\). Unfortunately, we will need to use all three symbols. In general, \(p\) and \(q\) will be reserved for the primes associated to the domain and target space of our functions; these two will be fixed. Meanwhile, \(\ell\) will be allowed to vary, as it is the prime we will use to investigate series convergence. Heuristically, one could say that our functions will go from \(p\)-adics to \(q\)-adics by way of the \(\ell\)-adic topology, for appropriate choices of \(\ell\).
If this terminology is unfamiliar to the reader, a brief review of absolute values on fields is given at the start of Section 2.
Recall that a number field is a finite-degree field extension of \(\mathbb{Q}\).
Recall that a number ring is a ring which is the ring of integers of some number field.
If the reader is troubled by this, observe that \(\overline{\mathbb{Q}}\) and any subfield thereof is discrete; in particular, for such a field, the trivial absolute value is the only absolute value with respect to which such a field is metrically complete.
Since this function is uniformly continuous, it is then \(L^{1}\)-integrable with respect to the real-valued Haar probability measure on \(\mathbb{Z}_{2}\). So, we can use the theory of Fourier analysis on locally compact abelian groups (Pontryagin duality, etc.) to study it. This also includes distributional notions of differentiation (see [25], for example).
That is \(\mathfrak{z}\in\mathbb{Z}_{2}\backslash\mathbb{Q}\). The irrational \(2\)-adic integers are precisely those \(2\)-adic integers whose digit sequences are not eventually periodic.
In fact, the assumption that such a universality property holds for general power series is arguably the oldest mistake in \(p\)-adic analysis, dating back to Kurt Hensel himself and his erroneous \(p\)-adic proof that the constant \(e\) is transcendental [23].
Specifically, \(m\) is the non-negative integer whose \(2\)-adic digits are the initial string of \(\mathfrak{z}\)’s \(2\)-adic digits before \(\mathfrak{z}\)’s digits become periodic. Meanwhile, \(n\) is the non-negative integer whose \(2\)-adic digits generate the periodic part of \(\mathfrak{z}\)’s digit sequence. For example, if the periodic part of \(\mathfrak{z}\) is \(01010101\ldots\), then the generating sequence is \(01\), which represents \(n=2\). On the other hand, for something like \(110110110110\ldots\), note that the generating sequence is \(1101\) (which represents \(n=11\)), rather than \(110\). As a \(2\)-adic integer, we have \(110=1+1\cdot2+0\cdot2^{2}=3=11\). This is because, when writing a non-zero \(2\)-adic integers’ digits left-to-right in order of increasing powers of \(2\), the right-most digit has to be \(1\).
Thus, \(U_{1}\) is the set of all \(p\)-adic integers with infinitely many \(1\)s digits; \(U_{2}\) is the set of all \(p\)-adic integers with infinitely many \(2\)s digits, but only finitely many \(1\)s digits; \(U_{3}\) is the set of all \(p\)-adic integers with infinitely many \(3\)s digits, but only finitely many \(1\)s or \(2\)s digits; etc.
In practice, \(K\) will be a field.
Or, perhaps, as the “Radon-Nikodym derivative” of the measure \(\left|\mathfrak{z}\right|_{p}^{\alpha}d\mathfrak{z}\), for an appropriate sense of “Radon-Nikodym derivative”; the issue is not entirely straightforward in the \(\left(p,q\right)\)-adic context, when \(p\neq q\); see for example, [22].
The functional equations are those generated by the affine linear maps contained in \(H\).
That is, there is an integer \(n\geq1\) so that \(H^{\circ n}\left(x\right)=x\), where \(H^{\circ n}\left(x\right)\overset{\text{def}}{=}\underbrace{H\circ\cdots\circ H}_{n\text{ times}}\).
\(x\in\mathbb{Z}\) is a divergent point when the ordinary absolute value \(\left|H^{\circ n}\left(x\right)\right|_{\infty}\) tends to \(\infty\) as \(n\rightarrow\infty\). The author conjectures in [23] that the converse of (2) is also true: if \(x\in\mathbb{Z}\) is a divergent point of \(H\), then there is a \(\mathfrak{z}\in\mathbb{Z}_{p}\backslash\mathbb{Q}\) so that \(x=\chi_{H}\left(\mathfrak{z}\right)\).
A more rigorous, comprehensive exposition of the adèles is given in [13]; that resource also details \(p\)-adic and adèlic distributions and the Schwartz-Bruhat functions which are used as test functions. It should be mentioned, however, that, like virtually all of the extant literature, to the extent analysis is done with the adèles, it is with functions that accept adèlic inputs and produce real or complex numbers as outputs. This is the opposite of what we do here, which is to consider functions that produce adèles as outputs. That being said, there may be a way to combine these two approaches and have functions whose inputs and outputs are both adèlic.
It would be interesting to see if there was an algebraic way to define this formal summation process, perhaps by playing around with quotients of rings of formal power series?
References
J. Aguayo and M. Moraga, “A Radon Nikodym theorem in the non-archimedean setting,” Proyecciones (Antofagasta) 20 (3), 263–279 (2001).
A. Baker, Transcendental Number Theory, Cambridge Mathematical Library (2nd ed.) (Cambridge Univ. Press, 1990).
J. Bell, “Harmonic analysis on the \(p\)-adic numbers,” <https://individual.utoronto.ca/jordanbell/notes/padicharmonic.pdf> (2016).
Y. Amice, Les nombres \(p\)-adiques (Presses universitaires de France, 1975).
N. Childress, Class Field Theory (Springer Science & Business Media, 2008).
H. Cohen, Number Theory: Vol. II: Analytic and Modern Tools (Springer Science & Business Media, 2008).
K. Conrad, “Infinite series in \(p\)-adic fields.” <https://kconrad.math.uconn.edu/blurbs/gradnumthy/infseriespadic.pdf>.
B. Dragovich, A. Khrennikov, S. Kozirev, I. Volovich and E. Zelenov, “\(p\)-Adic mathematical physics: the first 30 years,” \(p\)-Adic Numb. Ultrametr. Anal. Appl. 9 (2), 87–121 (2017).
G. B. Folland, Real Analysis: Modern Techniques and their Applications (John Wiley & Sons, 1999).
G. B. Folland, A Course in Abstract Harmonic Analysis (CRC Press, 2016).
D. Goldfeld and J. Hundley, Automorphic Representations and L-Functions for the General Linear Group: Vol. 1 (Cambridge Univ. Press, 2011).
N. Koblitz, \(p\)-Adic Analysis: A Short Course on Recent Work (Cambridge Univ. Press, 1980).
A. Y. Khrennikov, A. V. Kosyak and V. M. Shelkovich, “Wavelet analysis on adeles and pseudo-differential operators,” J. Fourier Anal. Appl. 18 (6), 1215–1264 (2012).
A. Y. Khrennikov, V. M. Shelkovich, J. H. van der Walt, “Adelic multiresolution analysis, construction of wavelet bases and pseudo-differential operators,” J. Fourier Anal. Appl. 19 (6), 1323–1358 (2013).
S. Lang, Algebraic Number Theory (Springer Science & Business Media, 2013).
D. A. Marcus and S. Emanuele Number Fields (New York, Springer, 1977).
A. R. Robert, A Course in \(p\)-Adic Analysis (Springer Science & Business Media, Chicago, 2013).
A. C. M. van Rooij, Non-Archimedean Functional Analysis, Pure and Applied Math. 51 (Marcel Dekker, New York, 1978),
A. C. M van Rooij and W. H. Schikhof, “Non-Archimedean integration theory,” Indag. Math. (Proceedings) 72 (2) (North-Holland, 1969).
W. Schikhof, Ultrametric Calculus: An Introduction to \(p\)-Adic Analysis, Cambridge Studies in Advanced Mathematics (Cambridge, Cambridge Univ. Press, 1985).
W. H. Schikhof, Non-Archimedean Harmonic Analysis, Ph.D. Thesis (Catholic University of Nijmegen, The Netherlands, 1967).
W. H. Schikhof, “A Radon-Nikodym theorem for non-Archimedean integrals and absolutely continuous measures on groups,” Indag. Math. 33, 78–85 (1971).
M. C. Siegel, \(\left(p,q\right)\)-Adic analysis and the Collatz conjecture, Ph.D. Thesis, (University of Southern California, 2022). <https://siegelmaxwellc.files.wordpress.com/2022/08/phd_dissertation_final.pdf>.
M. C. Siegel, “The Collatz conjecture & non-Archimedean spectral theory: Part I -- Arithmetic dynamical systems and non-Archimedean value distribution theory”. <https://arxiv.org/abs/2007.15936> (2007).
V. S. Vladimirov, “Generalized functions over the field of \(p\)-adic numbers,” Russian Math. Surveys 43 (5), 19 (1988).
Acknowledgments
As mentioned, this paper presents and builds upon the author’s PhD (Mathematics) dissertation [23] done at the University of Southern California under the supervision of Professors Sheldon Kamienny and Nicolai Haydn. Thanks must also be given to the author’s friends and family, as well as Jeffery Lagarias, Steven J. Miller, Alex Kontorovich, Andrei Khrennikov, K.R. Matthews, Susan Montgomery, Amy Young and all the helpful staff of the USC Mathematics Department; all the kindly strangers became and acquainted with along the way, and probably lots of other people, too.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Siegel, M.C. Infinite Series Whose Topology of Convergence Varies From Point to Point. P-Adic Num Ultrametr Anal Appl 15, 133–167 (2023). https://doi.org/10.1134/S2070046623020061
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S2070046623020061