No label defined (Q3159757)

\textit{I. Lakatos} has shown in [Philosophical Papers, Vol. 2, 45--60 (1978)] that, from a non-standard analysis point of view, Cauchy's theorem from 1821 and 1823 on the continuity of the limit function is correct. This evoked a polemic: it was pointed out that Cauchy himself in 1853 clearly stated that his theorem needed the uniform convergence hypothesis. The most recent account of this polemic was credited to \textit{T. Koetsier} [see his: Lakatos' Philosphy of Mathematics, Amsterdam, North Holland (1991; Zbl 0743.00017)]. The present paper seeks to insert a brief set-theoretic introduction to non-standard analysis into the discussion of the so-called Cauchy's `mistakes'. The main conclusion is that, in 1853, Cauchy actually added point-wise, not uniform convergence to his 1831 theorem, because from a non-standard analysis perspective, in 1821 he already had sufficient conditions for uniform convergence. A non-standard interpretation is developed. It is declared that Cauchy worked practically with a non-standard continuum, and his 1821 and 1853 proofs of the theorem should be interpreted accordingly. For this a more precise presentation of infinitesimals is needed. This is given here from the point of view of non-standard analysis. An extensive bibliography is added and discussion with several authors is developed, complemented with citations. In the final part the historical evolution is discussed: Cauchy-Weierstrass-Robinson. The authors hope to have shown that Cauchy may be considered as predecessor of Weierstrass but not in the sense of a continuous development as the logic of understanding puts it. The last sentence is: In any case, what modern non-standard analysis provides us is the standardization of a \textit{new concept of number}: In the future, will Robinson be to Dedekind as Dedekind is to Eudoxus?