A recurring theme is my use of hyperreal or hypercomplex polynomials of infinite degree. These functions behave (within certain limits) like ordinary polynomials but their number of terms is uncountably infinite.
These 'infinitely long' polynomials (they are infinitely much longer than any power series) provide a tool whose flexibility is unmatched in 'ordinary' analysis.
The main results regarding continuity of nonstandard polynomials can be found in
The last paper shows that uniform continuity is a poor substitute for nonstandard (hyperreal or hypercomplex) properties. In the hyperreal framework, e.g., Weierstrass' Approximation Theorem is upgraded from a sufficient condition on compact sets to a necessary and sufficient condition on G-delta sets.
For some time now I have been trying to reduce Schwartzian distributions to nonstandard polynomials. A first approach can be found in
This reduction leads to some new problems concerning Bernstein operators. In treating these I was happy to collaborate with Heiner Gonska (Duisburg, Germany), Ioan Gavrea (Cluj-Napoca, Romania), Daniela Kacso (Cluj-Napoca, Romania) and my former Ph.D.-student, now my colleague Hans Vernaeve. See the latter's Ph.D. thesis Nonstandard Contributions to the Theory of Generalized Functions (May 30, 2002) for a comprehensive and successful treatment of the original problem along entirely different lines.
As the Ph.D. research of my assistant Sam Sanders went along, emphasis has shifted from the ultrapower approach to Elementary Recursive Nonstandard Analysis (ERNA), an axiomatic system developed in the last decade by Chuaqui, Sommer and Suppes. ERNA is a partial implementation of Hilbert's Finitistic Program. We were most happy with the expertise in logic and model theory generously offered by Ulrich Kohlenbach (Darmstadt) and Andreas Weiermann, now our colleague at Gent University.
Preliminary version ERNA at Work appeared in The Strength of Nonstandard Analysis, Imme van den Berg and Vitor Neves, eds., SpringerWienNewYork 2007, pp. 64-75.
(More papers submitted.)
Preliminary version Stirling's series for n! made easy appeared in Real Analysis Exchange, 26th Summer Symposium Conference Reports, June 2002, pp.67-71.
For some time now, I've been working on a major project with working title Mathematics, a fragment constructed from zero. It originates in my anger facing the decline of secondary school mathematics. Originally, I intended to systematically develop secondary school mathematics, as it used to be in the good old days in our finest schools. (Actually, those days turned out to be less good than I remembered.) I first called it Deductive intermediate mathematics but changed the title after switching to a constructive approach. It was never intended for secondary school students, though, and in its present form even teachers may find it hard going. It's highly "no nonsense". There are no axioms. Real numbers are defined as decimal expansions; functions like the exponential and sine are defined by their series. Everything dispensable is dispensed with. (For instance: so far, I have no need for derivatives beyond the first.) Proofs are as short as I could provide them. But everything comes with its proof. There are also no loops or loose ends; everything (so far) is deduced by pure logic. Geometry will be analytic. If you look carefully, you'll find hat most calculus books at some point rely on geometry and vice versa. This I want to avoid. The style will be austere, without external elements begging for the attention of uninterested people. No applications will be included. (You read it correctly: no applications.)