Qu'est ce qu'une différentielle d'ordre n? (What is a differential of order n?) (Q1263879)

The stochastic integral, or ``Itô's integral'', which arises naturally in probability theory, does not obey the usual rules of calculus. The most famous example occurs in Itô's change of variables formula, which has an ``extra'' term, which is intuitively understood as keeping the second term in a Taylor expansion on an infinitesimal level; this second term disappears (i.e., is zero) in the normal calculus. The development of stochastic integration on manifolds, led by L. Schwartz, the author, and M. Emery, has clarified this ``pathology'', and has led to a reconsideration of a symmetric calculus on manifolds, rather than the customary antisymmetric one. Indeed, the identity \(d^ 2f=0\) of the exterior calculus is exactly what geometric probabilists do not want. In this provocative article, written in an expository, eminently readable way, the author makes a convincing case for the reexamination of some nineteenth century ideas revolving around differentials of order n. Only the order two has applications to date (to stochastic integration), but the theory developed here is elegant, simple, and very well presented.